Episode 114: Campbell’s Evolutionary Epistemology

• 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: Hello out there. This week on The Theory of Anything podcast, we look at Popperian Donald Campbell's Evolutionary Epistemology, which expands Karl Popper's ideas about scientific knowledge and learning to the natural world. Could it be that nature creates knowledge through analogous process called natural selection? How far reaching is Popper's theory? Could it be that this is how cultures create knowledge? Perhaps this even has cosmological implications. Is Karl Popper just describing how the universe works? I enjoyed listening to Bruce greatly like I always do, and I hope someone out there gets something out of this too.

[00:00:54]  Bruce: Welcome to The Theory of Anything Podcast. Hey, Peter.

[00:00:56]  Peter: Hey, Bruce. How are you today?

[00:00:58]  Unidentified speaker: Good.

[00:01:00]  Bruce: So today is gonna be our most behemoth episode yet.

[00:01:04]  Unidentified speaker: Oh boy.

[00:01:05]  Bruce: Because I am thoroughly incapable of cutting down my presentations to something reasonable.

[00:01:11]  Unidentified speaker: Okay.

[00:01:13]  Unidentified speaker: I'll buckle up then.

[00:01:15]  Unidentified speaker: Right.

[00:01:16]  Peter: Most behemoth, that's...

[00:01:17]  Unidentified speaker: Yeah.

[00:01:19]  Unidentified speaker: All right.

[00:01:21]  Bruce: So I've mentioned the Theory of Evolutionary Epistemology as developed by Donald Campbell and also Karl Popper in past podcasts, especially in episode 26. But like I mentioned it all the time.

[00:01:36]  Unidentified speaker: Now, did Popper ever use the word evolutionary epistemology?

[00:01:40]  Bruce: He did, but he found it pretentious.

[00:01:42]  Unidentified speaker: Extrapolated from him.

[00:01:44]  Unidentified speaker: Yeah.

[00:01:44]  Bruce: He found the term pretentious.

[00:01:46]  Unidentified speaker: Okay.

[00:01:47]  Bruce: Though he tried to avoid the term, he would say towards an evolutionary epistemology or something along those lines.

[00:01:55]  Peter: So he considered it a fair interpretation of his work then at least.

[00:01:59]  Bruce: Oh, yeah. Absolutely.

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

[00:02:01]  Bruce: So I have an opinion as to why he found it pretentious. I think the simple truth is that the evolutionary epistemology that Donald Campbell developed, which he loved, we will discuss that, is just not the correct, fully correct, generalized version of his epistemology. I think it needed a lot of work that it just wasn't. Campbell put a ton of work into it.

[00:02:26]  Unidentified speaker: He did a great job.

[00:02:26]  Bruce: I'm not trying to downplay Campbell's achievement, right? But I think Popper was right to not see it as the final general evolutionary epistemology that we are seeking, right? So at least that's my interpretation as to why he found it pretentious. I haven't researched it enough. Maybe I'll change my mind when I read more about what Popper wrote about it, wrote about his own version of it, I mean. But here's the thing, though. Donald Campbell called it evolutionary epistemology. That's the name of it, right?

[00:03:02]  Unidentified speaker: That he named it.

[00:03:03]  Bruce: He coined the term evolutionary epistemology to describe his theory, right? So it's the proper thing to call it, regardless of whether ultimately this is the correct version of a theory or not, if that makes any sense.

[00:03:20]  Peter: And this is, would you describe this as a subset of Popperian epistemology?

[00:03:26]  Unidentified speaker: [Superset?]

[00:03:26]  Peter: Or is this more of a, okay, that's what I was getting at. It's more of a all-encompassing.

[00:03:31]  Unidentified speaker: Yeah.

[00:03:31]  Unidentified speaker: Okay.

[00:03:32]  Unidentified speaker: Okay.

[00:03:33]  Unidentified speaker: Yeah.

[00:03:34]  Bruce: An attempt at being more encompassing, right? It's a generalization. So a generalization means that you're trying to, that Popper's epistemology is part of evolutionary epistemology, but evolutionary epistemology includes many other things.

[00:03:51]  Peter: So it's basically saying, this isn't just how humans create knowledge.

[00:03:55]  Unidentified speaker: This is how nature.

[00:03:58]  Unidentified speaker: That is correct.

[00:03:59]  Unidentified speaker: [Knowledge?]

[00:04:01]  Unidentified speaker: Yeah.

[00:04:01]  Unidentified speaker: Yes.

[00:04:02]  Bruce: So he was swinging for the fences, right? That's just no doubt. He absolutely was trying to come up with an all-encompassing theory of epistemology.

[00:04:13]  Unidentified speaker: He meaning Popper or?

[00:04:15]  Unidentified speaker: Donald Campbell.

[00:04:17]  Unidentified speaker: Campbell.

[00:04:17]  Unidentified speaker: Okay.

[00:04:18]  Unidentified speaker: So you know what?

[00:04:19]  Unidentified speaker: You ask a question.

[00:04:20]  Bruce: Actually, Popper was trying to do that too, right? But Popper's approach was very different. So I'm going to actually give quotes that make this more clear.

[00:04:29]  Unidentified speaker: Okay.

[00:04:29]  Bruce: What the relationship is. Because Campbell spends quite a bit of time on this in his papers.

[00:04:35]  Unidentified speaker: Okay.

[00:04:36]  Bruce: So I think I can clarify this all using Campbell's and Popper's own words, if that makes any sense. So now I've talked about the relevance of Donald Campbell's theory to animal learning, and I'm going to try to drive that point home again in this episode today. I've also discussed how it is, how it is, how Campbell's Evolutionary Epistemology is at odds with many of Deutsch's theories of knowledge, particularly Deutsch's Two Sources Hypothesis, the idea that there are only two sources of knowledge, biological evolution and human ideas. I've not, I've been very open on this podcast that I disagree with that theory, but it's because of Donald Campbell's theories that I disagree with that theory. I think Campbell refuted it, like prior to him ever creating it, of course, because Campbell isn't contemporary with, with Deutsch, obviously. But I, I, and, and, and in saying that, I'm not trying to endorse Campbell's theory. I actually think it's a wrong theory. But like, it's okay that it's wrong. There's nothing wrong with a wrong theory. We're trying to work our way through and solve problems, and having a theory that's partially right and has verisimilitude, but it's still wrong, that's like a genuinely interesting thing, right? And so I'd like, that's why I get excited about it, even though I will always say it's a wrong theory.

[00:06:05]  Unidentified speaker: Okay.

[00:06:06]  Bruce: Trying to get information on Campbell's actual theories is very tough. I only know of one source today, a book that is called Evolutionary Epistemology, Rationality and the Sociology of Knowledge. It's available on Amazon, I'll put a link to it in the show notes, etc. But like, it's not easy to come by. I'm not advocating for breaking copyright laws, but like even if you were to go out to like libgen.is and try to download a pirated copy, I don't think you'd find it, right? Like it's just not available. It's difficult to come by, let me just say. It's hard to find online.

[00:06:52]  Unidentified speaker: It's hard to just find anywhere that you can get your hands on it and you can actually read his theory using his own words, right?

[00:07:02]  Bruce: So today's podcast is going to be a very in-depth dive into Campbell's theories, largely quoting him, so that I'm getting him in his own words, before I've always summarized him, which of course means there's a layer of interpretation. Like I do my best, like I've tried really hard, like in episode 26, to explain why I think my interpretation is the correct interpretation of Campbell. Therefore, in my mind, my interpretation of Campbell is what Campbell was saying.

[00:07:28]  Unidentified speaker: But you know what?

[00:07:29]  Unidentified speaker: I don't know that for sure.

[00:07:30]  Bruce: Like how could I ever know that for sure, right? I'm simply using conjecture refutation. I'm trying to work out against alternative interpretations, why I think those are refuted by what he says. It's the same as anything, right? I'm going to do this episode so I can make some of his writings and ideas available to the critical rationalist community for those that are interested. And that's why I finally decided I'm going to just do one really, really long podcast, just going over his theories in his own words as much as possible, just so it's available to people. I'm not going to read it word for word. I'm still curating it down to what I think are the most interesting points that he makes. I'm trying to find the points that the CritRat community would find interesting. He spends a ton of time, I'm going to discuss this as we get further, but he spends a ton of time trying to prove that there's no such thing as direct knowledge. The entire CritRat community buys that. I don't really need to dwell on that aspect of his work, where that's a huge aspect of his work. I'm trying to curate it down to what today a modern critical rationalist would find interesting. Here's what I'm going to cover in this one epic podcast. First an introduction to Donald Campbell and the importance of his work in Evolutionary Epistemology. Second, the motivations for a theory of Evolutionary Epistemology, that came from Popper himself. And in fact, arguably, Campbell's theory is just a slight improvement on Popper's ideas. In fact, almost all of Campbell's ideas came from Popper in all truth.

[00:09:17]  Bruce: So when I call it Campbell's theory, it's almost unfair. I should really be calling it the Campbell-Popper Theory of Evolutionary Epistemology. Finally, I'm going to take a detailed look at his own examples of the theory. And as a bonus, I'm going to read for you Popper's reaction to Campbell's theory. In a future podcast, I will go into more detail of Popper's reaction. Popper did have some minor nits with it, which I'm not going to cover in today's episode. And I think what his nits are are very interesting. It's surprising how few nits he had with it, though. Along the way, I will call out how this theory differs from Deutsch's current thinking on the concept of knowledge and explain why Campbell's theory is the better theory and why it uncovers fatal flaws in Deutsch's Two Sources Hypothesis. Though that doesn't mean Deutsch's Constructor Theory of Knowledge is wrong. I've argued that the Two Sources Hypothesis is not even an implication of Deutsch's Constructor Theory of Knowledge. So I honestly, I think it can totally be the case that Deutsch's Constructor Theory of Knowledge is completely right and also Campbell's Theory is completely right. They do not have to be seen as at odds with each other. But the Two Sources Hypothesis is for sure at odds with Campbell's Theory. Indeed, once you drop the Two Sources Hypothesis, I honestly think the Two Theories complement each other very well. So what I'm really seeking then is a synthesis of the Two Theories.

[00:10:49]  Unidentified speaker: If that makes any sense.

[00:10:52]  Bruce: So here's the introduction. Donald Campbell was an American social scientist that worked on a generalization of Popper's Theory starting in the 1950s, and he called it Evolutionary Epistemology. That term will be very familiar to you even if you follow this podcast, it will be very familiar to you because I talk about it all the time. But even if you didn't follow this podcast, that's probably a term you've heard many times. If you're in any way adjacent to the Critical Rationalist Community, because they throw that term around a lot. They don't realize Campbell coined the term, and that it actually means his theory. They almost assuredly accept a diluted version of the theory today. It's just a more general term that means something like, biological evolution and human ideas, they use similar processes, something along those lines, if that makes any sense. Campbell also famously coined the term downward causation, by the way. So Wikipedia, in their entry on Evolutionary Epistemology, they say the following, Campbell founded the domain of Evolutionary Epistemology. This can be seen as a generalization of Karl Popper's philosophy of science, which conceives the development of new theories as a process of proposing conjectures, blind variation, followed by the refutation, selective elimination, of those conjectures that are empirically falsified. Campbell added that the logic of blind variation and selective elimination slash retention underlies all knowledge processes, not only scientific ones.

[00:12:26]  Bruce: Thus, the blind variation and selective retention mechanism explains not only creativity, but also the evolution of instinctive knowledge and of our cognitive abilities in general. Now, my sources for this are going to come from the book I just mentioned, Evolutionary Epistemology, Rationality and the Sociology of Knowledge, primarily from the two chapters that Campbell wrote himself. One of the chapters, so when I say chapters, it's not like this book is just a collection of papers that were previously published. So these are two papers that he wrote that then got collected into this collection. So the first one is called, and the main one that I'm going to be using in this podcast is called Evolutionary Epistemology, that's a 1974 paper. But I'm also going to be taking from a paper that has a long name called Blind Variation and Selective Retention in Creative Thought as in Other Knowledge Processes, which is for AGI nerds, it's a 1960s attempt to work out human creativity using Popper's epistemology as a starting point. Which if you're an AGI nerd, I mean like, especially a CritRat AGI nerd, oh my gosh, right? Like this is a paper you absolutely should have read. And if you haven't, like it's a scandal. But note that he started work on his theory back in the 1950s. So we are talking about, assuming Evolutionary Epistemology was his final work, which I don't know if it was or wasn't, we are talking about two decades at least of work that went into this theory. So Campbell's work inspired entire new fields.

[00:14:01]  Bruce: There is a field today known as Universal Darwinism, which is still active to this day and that was directly inspired by Campbell's work on Evolutionary Epistemology. Indeed, Universal Darwinism is a synonym today for Evolutionary Epistemology, and the two terms are often used interchangeably. So here is what Wikipedia says about Universal Darwinism. Universal Darwinism, also known as General Darwinism, is a variety of approaches that extend the Theory of Darwinism beyond its original domain of biological evolution on Earth. Universal Darwinism aims to formulate a generalized version of the mechanisms of variation, selection, and heredity proposed by Charles Darwin so that they can apply to explain evolution in a wide variety of other domains, including psychology, linguistics, economics, culture, medicine, computer science, and physics. Now, on that same entry, Wikipedia specifically places Campbell as inspiring the field. Starting in the 1950s, Donald T. Campbell was one of the first and most influential authors to revive the tradition and to formulate a generalized Darwinian algorithm, directly applicable to phenomena outside of biology. Note the use of the term generalized Darwinian algorithm. I suggested back in episode 26 that we should instead think of this as a meta-algorithm rather than a specific algorithm, as it is, as we'll see, supposed to capture a whole class of algorithms that are different in many ways.

[00:15:40]  Bruce: This is also going to become a point of contention between me and members of the CritRat community when we do podcasts about other possible interpretations or criticisms of this theory.

[00:15:53]  Peter: While you were reading that on Wikipedia, I was looking it up on ChatGPT, too, and it lists other examples of applications of Universal Darwinism. I just thought this was so interesting. Other than culture, science and knowledge, artificial intelligence, there's also cosmological Universal Darwinism, as advocated by Lee Smolin. And then, of course, Quantum Darwinism, which is Zurek or something, I've never heard of him, but the state says that this could really be a far-reaching idea. This could be the theory of everything.

[00:16:34]  Unidentified speaker: Then, really?

[00:16:34]  Unidentified speaker: I mean, I'm not saying it is, but...

[00:16:38]  Bruce: Let me actually take the two examples you just used. So Lee Smolin has this idea that there are multiple universes. So this is a multiverse, but not the quantum multiverse that we're talking about.

[00:16:47]  Peter: I think you've talked about it before a little bit.

[00:16:48]  Unidentified speaker: Yeah.

[00:16:49]  Bruce: So, and the idea is, is that these universes evolve and that the reason why our universe is in such a, like, low entropy state, is this uncannily low entropy state, is because there is in fact a... The reason why our universe has an appearance of design, let's put it that way, okay, is because he's arguing that perhaps we exist inside of a set of universes that evolved. So that cosmology itself is based on evolution. And that might be an explanation for the fine-tuning problem. Now, keep in mind, this is just a really early conjecture. There is no evidence for this whatsoever at this point, right?

[00:17:32]  Unidentified speaker: Okay.

[00:17:33]  Bruce: So Lee Smolin is conjecturing, maybe this will turn out to be true, but the theory must be turned into a testable theory before it can be considered a scientific theory. Right now, it's more like an armchair philosophical theory, if that makes any sense.

[00:17:47]  Unidentified speaker: Okay.

[00:17:48]  Bruce: But I find it a really interesting theory, right? Like, who knows, maybe he's right, right? Maybe that does explain the fine-tuning problem. And in which case, that would mean that evolution happens at the level of universes. I mean, like, that's an amazing, amazing idea.

[00:18:09]  Unidentified speaker: Yeah.

[00:18:09]  Peter: So it could explain everything from the quantum level to the universe level, supposedly.

[00:18:16]  Bruce: Quantum Darwinism, I don't know much about at all. So I've mentioned that I read David Deutsch's books. He advocated for many worlds. I thought, okay, that's crap. And because I'm someone who goes out and does my best to criticize theories, I immediately went and read alternative views, starting with Roger Penrose, who's one of the smartest men in the world. And Roger Penrose made me start to realize that maybe Deutsch was right, that many worlds was the best explanation of quantum physics. I went out and I checked out Bohmian Bohm’s Theory and found it to be a bad explanation. I checked out the transactional theory and found it to be a bad explanation. In fact, both of those theories actually are kind of quasi crypto many worlds theories. Deutsch mentions this, but it's the truth. Okay, I had to figure that out. I didn't take Deutsch's word for it.

[00:19:19]  Unidentified speaker: I went and actually worked it out for myself.

[00:19:23]  Bruce: The one I have not checked out that I know exists is quantum Darwinism. And I think what happened was is I lost momentum. It took, you have to understand the amazing amount of effort and hours I must put in to solve problems like this as a layman, right? That I'm not getting paid for, that I'm doing purely for fun as a labor of love. So I never made it. I always meant to go on to check out quantum Darwinism and I got sidetracked into doing other things. But quantum Darwinism is a competitor to many worlds and it tries to explain the resolution of, as I understand it today, which I don't, it tries to explain the resolution of observations as a sort of struggle between many different possibilities that then resolves to, through Darwinism, to a best possible outcome. And so it would get rid of the multiverse, the quantum multiverse if it were true. So it is a competitor to MWI.

[00:20:21]  Peter: It's not a very popular theory, though, right?

[00:20:23]  Unidentified speaker: It is not, no.

[00:20:24]  Unidentified speaker: I haven't heard it mentioned.

[00:20:25]  Peter: I've only heard it mentioned very rarely.

[00:20:28]  Bruce: Do you know where you probably heard it mentioned the first time?

[00:20:30]  Unidentified speaker: [Bobby Azarian’s book?]

[00:20:31]  Peter: The Azarian, yeah, yeah.

[00:20:33]  Bruce: So he totally endorses that theory. I should probably mention that, as much as I love Bobby Azarian's book and love Bobby Azarian, he is not an expert in what he's writing about in the same way that Deutsch is. I mean, he's an expert in his own field. I'm not trying to downplay his expertise, but he's intentionally writing outside his field. Whereas, Deutsch was far more into his own field because-

[00:20:57]  Peter: It's just a different kind of a book.

[00:20:59]  Unidentified speaker: Yeah.

[00:21:00]  Bruce: So I don't know. Deutsch just makes better arguments because he just thoroughly understands the material in a way that I would not expect Bobby Azarian to be. That doesn't mean anything though. Like it still could be that Bobby Azarian is actually correct, right? So I mean, I'd have to go look into it myself. I have not ruled out quantum Darwinism as a possibility. Let me just throw that out there.

[00:21:28]  Unidentified speaker: Yeah.

[00:21:28]  Unidentified speaker: Okay.

[00:21:29]  Bruce: So probably the most prolific researcher in the domain of universal Darwinism is John O. Campbell. No relation to Donald as far as I'm aware, but that's a name I've seen many times and he is one who's really advocated for universal Darwinism as a matter of fact.

[00:21:48]  Bruce: So, now in the area of evolutionary computation, Campbell is also often cited as the father of evolutionary computation. So for example, in David Fogel's Magnum Opus called Evolutionary Computation, Towards a New Philosophy of Machine Intelligence, he goes over the history and the founding of the field. And here is what he says of Campbell in a section entitled Early Speculations and Specific Attempts. So here he's trying to lay down what was the foundation of the field, right? And he says, Campbell, following his earlier work in 1956, offered a conjecture that in all processes leading to expansions of knowledge, a blind variation and selective survival process is involved.

[00:22:37]  Unidentified speaker: That's page 59 of the book I was just quoting.

[00:22:40]  Bruce: So Campbell's work was one of the inspiring starting points for literally the field of evolutionary computation, which is a huge field, right? So I'm trying to help you catch the vision here that we are not talking about an uninfluential set of papers that he wrote. Like, it absolutely inspired huge fields that are still around today and very active.

[00:23:08]  Unidentified speaker: Okay.

[00:23:09]  Bruce: So for better or worse, Campbell has had a huge influence on both critical rationalism, particularly in his attempt to generalize critical rationalism to evolutionary epistemology and in several other related fields. Consider the fact that we may not, you know, CritRats may not be aware of Donald Campbell's work, but every time they use the term evolutionary epistemology, which they do all the time, they're showing their unconscious dependency upon his work, okay? As we'll see, Popper was a huge fan of Campbell's work and endorsed it with relish. So Campbell's work has had a profound influence on critical rationalism. Yet most lay CritRats today have never heard of him, even as they toss around the term he coined, evolutionary epistemology, as if it's just a given. Even though they aren't aware what it originally meant. However, if you follow this podcast, you know I've struggled with Campbell's ideas. For one thing, I believe I've found potential counterexamples to it, which I discuss in episode 26.

[00:24:15]  Unidentified speaker: Or maybe I haven't.

[00:24:17]  Bruce: Maybe I'm misunderstanding the theory. I guess that's always possible, right? I do my best, but who knows? The problem is that his paper is maybe not as clear as I wish it was. And in future podcasts, I will critique Campbell's papers in detail about why I don't always find it the most clear, what I did to try to take the possible interpretations and narrow it down to one interpretation, which I think is the right one, and then judge for yourself. Judge for yourself if this is Bruce's version of evolutionary epistemology inspired by Campbell, or if, actually, I really am, as I think I am, simply explaining Campbell's evolutionary epistemology. And I guess it doesn't really matter. Like, it sort of doesn't matter which it is, because ultimately the ideas have to stand on their own.

[00:25:10]  Unidentified speaker: Okay?

[00:25:11]  Bruce: To be clear, though, if I say it's my version of his epistemology, that would be the same as saying I've steel-manned his epistemology and I still find it to be wrong. So I'm not advancing it as my own theory in that sense.

[00:25:25]  Unidentified speaker: Okay?

[00:25:25]  Bruce: But I do think the interpretation of him that I have arrived at is by far the best interpretation of him that has the fewest problems, if that makes any sense.

[00:25:35]  Unidentified speaker: Okay?

[00:25:35]  Bruce: I understand it to be the steel-man version of his epistemology. So I've done my best to seek out people to help me make sense of his views, as well as Popper's very solid endorsement of his views. But I found it difficult to find good, thoughtful commentary on his theory. Indeed, I would dare say that the modern CritRat response to Campbell's ideas has been extremely poor or maybe even outright sucky. In future podcasts, I'll elaborate on what I mean by that. Let's talk about the motivation though for why he wanted to create an evolutionary epistemology. So what is Campbell's Evolutionary Epistemology? So first of all, as I mentioned, the term is now diluted, so we want to get specific with what Campbell understood it to be. Here's the thing you have to understand. I read Campbell at least as putting forward a bold testable theory.

[00:26:33]  Unidentified speaker: Now that would make sense.

[00:26:36]  Bruce: I say testable, maybe I mean checkable, if you understand the difference between those two terms. I'm going to use testable to just mean checkable. Obviously, it's not an empirically testable theory per se, but it's a theory that I understand what a counter example would look like to it. I can go look for counter examples and I can go look and check to see if nature matches his theory or not.

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

[00:27:01]  Bruce: And so to me, he was not offering some armchair philosophy or mere intuition pump. He was boldly trying out an incredibly interesting idea that took risks and might be falsified. And indeed, as in Episode 26, I claimed that I falsified it. So there we go. So here is his bold idea in his own words. So first, Campbell admits that his ideas were entirely inspired by Popper's Scientific Epistemology, which Popper himself says his scientific epistemology was really just an attempt to study knowledge growth in general, not specifically scientific knowledge growth. So Campbell, quoting Popper, so this is actually Popper I'm about to read on page 51 of the book that I'm taking this from. It's, Popper says, the central problem of epistemology has always been and still is the problem of the growth of knowledge and the growth of knowledge can be studied best by studying the growth of scientific knowledge. Pages 52 to 53. My interest is not merely in the theory of scientific knowledge, but rather in the theory of knowledge in general, says Popper. Yet the study of the growth of scientific knowledge is, I believe, the most fruitful way of studying the growth of knowledge in general. For the growth of scientific knowledge may be said to be the growth of ordinary human knowledge writ large. Page 51. Moreover, it seems clear that the growth of scientific knowledge is the most important and interesting case of the growth of knowledge.

[00:28:31]  Bruce: But Popper found it easier to study knowledge growth in general, and then tried to generalize from there, because he thought studying common sense knowledge had proved too hard for most philosophers and hadn't gone anywhere. So on page 52, Popper says, scientific knowledge can be more easily studied than common sense knowledge. So Campbell was inspired by Popper's realization that if you study scientific knowledge growth, you were actually studying knowledge growth in general. But how do we generalize what we know about scientific knowledge growth to knowledge growth in general? So Campbell, again, quoting Popper, so this is Popper, says, how and why do we accept one theory in preference to others? We choose the theory which best holds its own in competition with other theories, the one which by natural selection proves itself the fittest to survive.

[00:29:25]  Unidentified speaker: That's page 49.

[00:29:27]  Bruce: This insight from Popper is Campbell's jumping-off point. Campbell then goes on to argue that his epistemology is not limited to scientific theories but is the basis for all knowledge increases. Campbell explains his goal in trying to find a generalized evolutionary epistemology. So this is from starting with page 47. Campbell says, An evolutionary epistemology would at a minimum be an epistemology taking cognizance of and compatible with man's status as a product of biological and social evolution. In the present essay, it is also argued that evolution, even in its biological aspect, is a knowledge process and that the natural selection paradigm for such knowledge increments can be generalized to other epistemic activities, such as learning, thought, and science. So far, this sounds to be entirely compatible with the current CritRat community's ideas around evolutionary epistemology and the growth of knowledge, since only biological evolution and human ideas have been mentioned so far. But here is Campbell's main insight.

[00:30:31]  Unidentified speaker: Page 50.

[00:30:33]  Bruce: If we have made it our task, then there is no more rational procedure than the method of trial and error, of conjecture and refutation, of boldly proposing theories, of trying our best to show that these are erroneous, and of accepting them tentatively if our critical efforts are unsuccessful.

[00:30:53]  Unidentified speaker: Page 50.

[00:30:55]  Bruce: Thus, trial and error will become the basis for Campbell's attempt to unify all knowledge creation processes. Inspired by this approach, Campbell explains his goals. This is from page 52 now. A focus on the growth of knowledge, on acquisition of knowledge, makes it appropriate to include learning as well as perception as a knowledge process. Such an inclusion makes relevant the learning processes of animals. However primitive these may be, they too must conform to an adequate logical epistemology. Animal learning must not be ruled out as impossible by the logic of knowing.

[00:31:32]  Unidentified speaker: Uh-oh.

[00:31:33]  Bruce: So he's including animal learning as evolutionary epistemology. This put Campbell directly at odds with the current CritRat community's views around Deutsch's two sources hypothesis, that there are only two sources of knowledge, biological evolution and human ideas. But Campbell explains why this is non-negotiable for him and Popper. In the process, Popper, this is quoting from page 50 now, Popper has effectively rejected the model of passive induction, even for animal learning, and advocated here too, the typical process involves broad generalizations from single specific initial experiences. Advocated that here too, the typical process involves broad generalizations from single specific initial experiences, generalizations which subsequent experiences edit. Quoting from page 51, it is the logic of variation and selective elimination which has allowed Popper to go on to describe that sense in which animals and scientific knowledge is yet, animal and scientific knowledge is yet possible. So Campbell takes seriously the idea that we do not induce our knowledge, but instead we learn via trial and error. So animal learning must not be seen as a counter example to this. If it is, it refutes Popper's epistemology. Okay, I cannot overstate this, okay? Now, many CritRats, especially if they don't realize how important Campbell was to developing evolutionary epistemology as a theory, may feel they can simply dismiss him upfront. But note that Campbell got this all from Karl Popper. So pages 52 to 53, Campbell is now quoting Popper. So here's Popper.

[00:33:25]  Bruce: Although I shall confine my discussion to the growth of knowledge and science, my remarks are applicable, without much change, I believe, to the growth of pre-scientific knowledge also. That is to say, to the general way in which men, and even animals, acquire new factual knowledge about the world. The method of learning by trial and error, of learning from our mistakes, seems to be fundamentally the same, whether it is practiced by lower or by higher animals, by chimpanzees, or by men of science. So Popper also is seeking a unifying epistemology that explains all knowledge growth, including that of animal learning. So to Popper, his epistemology is refuted if it can't explain animal learning as being a knowledge growth process consistent with his epistemology. To Popper, it is simply unacceptable to claim all animals' knowledge is in their genes. Why does Popper see it that way? Think of it this way. Animal, and I'm going to discuss this a bit further using Campbell's own words, okay? This is my own attempt to explain it to a critical rationalist audience that is struggling with this aspect of evolutionary epistemology. Animal learning is an ability of an animal to change or overwrite, or if you prefer, disobey its default genetic behavior based on its actual environment, even though that behavior was part of its genetic knowledge. That is to say, learning is a legitimate kind of adaption to the environment and captures what we call knowledge about a local environment.

[00:35:06]  Bruce: As Campbell will later discuss in this podcast, there isn't really any difference between learning and instinct other than where the knowledge originated from. Whether an animal's behavior is learned in its lifetime, that's learning, or came pre-wired genetically, that's instinct. Either way, there is an adaption to the environment. So to call an adaption through instinct, i.e. its genes, an increase of knowledge, but then to insist that an adaption via learning, even though it's the same adaption perhaps, is not an increase of knowledge, simply ruins the theory as an explanation. So Popper rightly understood the burden of explanation was his to include animal learning into his overall theory or his theory was refuted. So Campbell wishes to formalize this idea that he took from Popper of a generalized epistemology that applies to all knowledge processes in nature, be it Darwinian evolution, human thought, animal learning, or even as we'll see, nonliving things. People often paraphrase Popper, saying that humans can allow ideas to die in their place, whereas animals must instead die to learn. And if I recall, Popper actually does say that somewhere, so they’re maybe even paraphrasing him correctly. But this next quote from Popper that Campbell uses, it challenges that interpretation. Here's the actual quote from Popper.

[00:36:36]  Unidentified speaker: This is from page 54.

[00:36:38]  Bruce: Error elimination may proceed either by the complete elimination of unsuccessful forms —the killing off of unsuccessful forms by natural selection— or by the tentative evolution of controls which modify or suppress unsuccessful organs or forms of behavior or hypotheses. Our schema allows for the development of error-eliminating controls, warning organs like the eye, feedback mechanisms. That is, controls which can eliminate error without killing the organism and make it possible, ultimately, for our hypotheses to die in our stead. Notice that he is absolutely talking about animals and humans here, okay? Animals have hypotheses in a loose sense, right? We would probably call them behaviors or expectations. And Popper, I think, if I recall, he says, if I can use the word hypothesis or the word theory very loosely, right? It's really some sort of expectation that they have, right?

[00:37:47]  Unidentified speaker: Or behavior that they have.

[00:37:49]  Bruce: And those can get eliminated instead of the animal itself. This is what we call classical conditioning or operant conditioning or trial and error learning, etc. There's a number of names for it. Campbell refers to this idea as a substitute or vicarious selection. He's going to later formulate a related idea of vicarious indicators based on this idea.

[00:38:12]  Unidentified speaker: Okay.

[00:38:14]  Bruce: Now, a side note, our best theories of animal learning have entirely, entirely, entirely corroborated Campbell and Popper on this. Animal learning, according to our best theories on the subject, which I have read up on, so I do know what I'm talking about, also works via trial and error, and fits well with this more generalized form of Popper's epistemology that we're talking about. So, instrumental conditioning, also called trial and error learning, from page 45 of The Thinking Ape by Richard Byrne, is how animals learn by trying out actions and receiving feedback from the environment. If the feedback is positive, they increase those actions. If it's negative, they reduce those actions. I would note that when I raise this counter-example to Deutsch's Two Sources Hypothesis to the CritRat community, I get a response that's something like this. Oh yeah, show me a dog that can invent an airplane. By the way, that's literally one of the responses I got from a CritRat when I tried to raise this. So, now, to be sure, I just at this point said to the CritRat, hey, Popper isn't claiming animal learning is equivalent to human learning. And the CritRat, he immediately, being a critical rationalist, he immediately said, oh, my bad, I was accidentally straw manning Popper. Okay, start again and I'll try to engage the idea better this time.

[00:39:33]  Unidentified speaker: Yeah, that didn't actually happen, by the way, just for the record.

[00:39:37]  Unidentified speaker: Okay.

[00:39:39]  Peter: That doesn't quite ring true.

[00:39:40]  Bruce: The first part did, but the second part didn't.

[00:39:43]  Unidentified speaker: Okay.

[00:39:45]  Bruce: So, a segment of the CritRat community today strongly rejects Popper's and Campbell's ideas around animal learning due to wide acceptance of David Deutsch's Two Sources Hypothesis. Let me just say that we have quite a number of podcasts on animal learning that situated into Popper's Evolutionary Epistemology. I don't have the numbers handy. I think I maybe list them later in this podcast, but if not, I'll try to link them or something in the show notes. But if you are seriously interested in Richard Byrne’s theories around this and what he said about this, we have four podcasts on the subject where I take you through how they compare to Popper's Epistemology, some of Popper's anticipations of what today we now understand to be a correct understanding of animal learning. And sometimes Popper even exceeded what we've learned today, and it's made some really interesting predictions about it. And also a comparison between Richard Byrne’s theories and David Deutsch's interpretations of Richard Byrne’s theories, which they differ from each other. Byrne understands his own theories differently than Deutsch understands Byrne’s theories. I will discuss more about the CritRat community's concerns here in future podcasts when I maybe talk about some of my criticisms of Campbell's theories and some of the other possible interpretations that members of the CritRat community have put out. So Campbell's idea then is that there is a nested hierarchy of blind variation and selective retention processes that include DNA and biological evolution, but are not limited to those.

[00:41:24]  Bruce: So Campbell is not denying the specialness of those either. And I think that's where a lot of the CritRat community's pushback comes from, is we all kind of, I mean, like, I've argued that the two sources hypothesis has some verisimilitude because it captures the idea that, while those aren't really the only two sources of knowledge, they are the only two open-ended sources of knowledge we know, right? And so, I can understand why people have become very attached to that theory as an intuition pump and why they're really hesitant to give up on it. Because there is something special about those. Now, Campbell’s not denying that, okay? So you have to understand that there's no incompatibility here. So, biological evolution may well supply the learning algorithm animals use. In fact, does supply the learning algorithm animals use. But animals do learn new knowledge not contained within their genes using a variation and selection process that has more than a passing resemblance to Darwinian biological evolution. And that's good, because if that weren't true, then that would be some form of induction and Popper's epistemology would be refuted.

[00:42:32]  Unidentified speaker: Okay?

[00:42:33]  Bruce: So, and this is something that I've long been confused on. When I bring up to the CritRat community, oh, but wait, animal learning. They'll go, well, of course, an animal can change its behaviors. Nobody doubts that, but that's not knowledge. You know, it's just an algorithm that lets them change their behaviors to be adapted to the environment better. It's like, guys, if it's not evolutionary epistemology, it's induction. That's the term for it, right? Like, you have to say what it is. You don't get to declare what it isn't. So anyhow, this is why I actually keep making a big deal about this. There's like really an important point here that keeps getting missed. So also DNA is not the start of the process. Now, this is something I brought up in the podcast before. There's kind of this idea that all knowledge came from biological evolution. But if that were true, then how did biological evolution get started? There must have been knowledge creation processes prior to the existence of DNA, or DNA never would have come into existence.

[00:43:39]  Unidentified speaker: Okay.

[00:43:40]  Unidentified speaker: That's an intriguing point.

[00:43:42]  Unidentified speaker: Oh, yeah, absolutely.

[00:43:43]  Unidentified speaker: And this is like that.

[00:43:44]  Unidentified speaker: Okay.

[00:43:45]  Bruce: That's what Lee Cronin is trying to do with assembly theory, he’s trying to come up with a generalized version of evolution that includes non-living things so that he can explain how evolution got off the ground to begin with, right?

[00:44:00]  Peter: So it could be like a kind of cosmological evolution, right? Is that a chemical evolution?

[00:44:08]  Unidentified speaker: Yes.

[00:44:08]  Unidentified speaker: Yes.

[00:44:09]  Unidentified speaker: Yes.

[00:44:09]  Unidentified speaker: I guess.

[00:44:11]  Bruce: Or something like we're not sure, right? We've got kind of nascent ideas as to what might have happened.

[00:44:19]  Unidentified speaker: Okay.

[00:44:20]  Bruce: So by the way, that's one of the ones I raised as it's a giant hole in our current theory of evolution that as a creationist, I could never get a single evolutionist to admit was a hole.

[00:44:34]  Unidentified speaker: But it is.

[00:44:34]  Bruce: I mean, of course, it's a hole. The fact that it's a hole is not a big deal, right?

[00:44:38]  Unidentified speaker: But it is.

[00:44:40]  Bruce: It's something that is absolutely not currently covered by our theory of evolution, that it must cover at some point, right? So, or evolution is going to be refuted. There's going to turn out to be some other process, induction or something, right? Of course, it's not really induction, but I'm trying to make the point that you're either going to explain this in terms of evolutionary epistemology or there's going to be a second epistemology that you're going to have to explain it through, okay? So, Campbell also points out that Popper's epistemology can explain non-living things like crystal formation. Here's a quote from him, from page 55. Thus, crystal formation is seen as a result of a chaotic permutation of molecular adjacencies, some of which are much more difficult to dislodge than others. At temperatures warm enough to provide general change, but not so warm as to disrupt the few stable adjacencies, the number of stable adjacencies will steadily grow, even if their occurrence is but a random affair, okay? That's an example of applying, of Campbell applying his evolutionary epistemology to outside of biology altogether, okay? And that he did see this as all-encompassing like that, okay? So we've discussed Campbell's and Popper's motivations, but what's the actual theory? So one thing I really like about Campbell's theory, really, really, really like about Campbell's theory, at least as I interpret it, is that it is quite bold in the Popperian sense of taking risks that it might get refuted. It is not some armchair philosophy that can't be tested against the real world.

[00:46:23]  Unidentified speaker: More on this later.

[00:46:24]  Bruce: But I do feel Campbell is less than clear on some unfortunately very important points. And this leads to several problems, which I'm going to cover in upcoming podcasts. I'll mention an example in passing in this podcast, just to give you a flavor for the types of problems that we encounter. So consider this quote from Campbell. He says, in other writings, the present author has advocated a... When he says his other writings, he's talking about his 1950s work on evolutionary epistemology.

[00:46:54]  Unidentified speaker: Okay.

[00:46:55]  Bruce: The present author has advocated a systematic extrapolation of this nested hierarchy of selective retention paradigm to all knowledge processes, in a way which, although basically compatible with Popper's orientation, may go further than he Popper would find reasonable in extremity, dogmatism and claims for generality. It may, on these same grounds, alienate the reader. Truer words have never been spoken. And this is why I've kind of been emphasizing the alienation that exists between Donald Campbell's theories and the CritRat community today. Because he is making some really, really bold claims. That really do kind of go against a lot of people's intuitions on the subject.

[00:47:44]  Unidentified speaker: Okay.

[00:47:47]  Bruce: Now, however, as we're going to see, Popper was not alienated by Campbell's theories. And in fact, strongly endorsed it. And claimed they [matched?] his own views, other than in some very minor ways. Now, Campbell proposes the following, page 56. This is his theory in his own words now. Number one, a blind variation selective retention process is fundamental to all inductive achievements, to all genuine increases in knowledge, to all increases in fit of system to environment. Note that all inductive achievement and all increases in knowledge in the statement, because it's going to become important later, and all increases to fit of system. Number two, in such a process, there are three essentials. A, mechanisms for introducing variations. B, consistent selection processes. And C, mechanisms for preserving and or propagating the selected variations. The and or is important here, I think. I think he's saying preserving and or propagating, meaning that they may or may not include replicators. I'm going to explain why this is significant in a second.

[00:49:02]  Unidentified speaker: Okay.

[00:49:03]  Bruce: Number three, the many processes which shortcut a more full blind variation and selective retention process are in themselves inductive achievements containing wisdom about the environment achieved originally by a blind variation and selective retention process. So Campbell sets up this idea that there can be a more or less full blind variation selective retention process where knowledge from somewhere higher in a hierarchy creates knowledge to speed up a lower down variation selection process, what he's going to later call a heuristic.

[00:49:40]  Unidentified speaker: Okay.

[00:49:41]  Bruce: Then number four, in addition, such shortcut processes contain in their own operation a blind variation selective retention process at some level, substituting for overt locomotor exploration or the life and death winnowing of organic evolution. So Campbell is claiming that at each level of the hierarchy, there will always be a blind variation selective retention process, perhaps substituting for another process. This might include things that are vicariously substitute for, say, the animal dying.

[00:50:18]  Unidentified speaker: Okay.

[00:50:18]  Bruce: Or maybe substitute for the animal moving.

[00:50:21]  Unidentified speaker: Okay.

[00:50:22]  Bruce: I'm trying out random movements to try to take care of a problem. And that allows their ideas then to die in their stead. Ideas here being loosely meant to be the knowledge that's contained within the animal via learning or genetic endowment. So to picture Campbell's theory in your head, imagine a tree with nodes where higher level nodes do a blind variation selective retention process, discover knowledge about the world, and then that creates or feeds knowledge to another node that then does its own blind variation selective retention process, but with insight from the higher node that makes the next node down less blind.

[00:51:05]  Unidentified speaker: Okay?

[00:51:06]  Bruce: So by the time you get to the bottom nodes, such as human thought, human creativity, it may not look very much like a blind process anymore. Yet Campbell is arguing that it is still, even human creativity is still partially a blind process. And insofar as it does not appear blind, that is only because there was some blind process at a higher node that gave a useful adaptive insight that allows less of a blind process.

[00:51:38]  Unidentified speaker: Does that make sense?

[00:51:39]  Bruce: Like this is kind of, I want to make sure people are picturing what Campbell has in mind correctly as we move forward, okay? Now, I mentioned I find Campbell to not always be very clear, and I'm going to give a bunch of examples in future podcasts. Here's one example from the first bullet point. Let me read it again. A blind variation selective retention process is fundamental to all inductive achievements, to all genuine increases in knowledge, to all increases in fit of system to environment. Note here that Campbell says, all inductive achievements, and he also says all genuine increases in knowledge, and he also says all increases to fit of system.

[00:52:19]  Unidentified speaker: Okay.

[00:52:20]  Bruce: Now, it seems to me, like when I read that sentence, I equate those three together. That is, I read him as saying inductive achievement is the same as increases in knowledge, which is the same as increases of fit to system. That is, I think he's offering three phrases to try to explain what he has in mind. Inductive achievement increases in knowledge, increases of fit of system to environment. It seems to me that Campbell is using three synonyms here, trying to try three synonyms to try to make sure he hasn't lost you as to what he's talking about, okay? Now, this makes sense to me because induction, as was understood at the time, was understood at the time as the process by which we move from specific observations to general theories, with the general theory being the supposed increase in knowledge, okay? So Campbell is simply saying, yes, there are inductive achievements or increases of knowledge. However, you want to word that is fine. But they all come through a blind variation selective retention process. But it isn't hard to see, and I think that is the right interpretation of Campbell here, okay? I'm going to give you examples of that as we go. But it isn't hard to see that you could read this as three entirely separate things, that you have inductive achievements, and you have increases of knowledge, and you have increases of fit to environment, and maybe Campbell is treating those all as three separate things. Many CritRats in the CritRat community have attempted to interpret Campbell that way. In any case though, he always equates those three together, right?

[00:53:56]  Bruce: So he always says blind variation selective retention is required for all increases in knowledge, for all inductive achievements, for all fits of system to environment, okay? So even if you disagree with my interpretation, it shouldn't ultimately matter, because even if those are three separate things in your mind, he's still saying blind variation selective retention is the basis for all of them, okay? This is the next part of Campbell's Theory that really bothers some critical rationalists. It defines knowledge in terms of an increase of fit to environment, rather than as information that keeps itself instantiated, okay? Now, however, to be fair, there is no reason why we can't accept two definitions of the word knowledge. Words have no essential single definitions. So this really shouldn't be a stumbling block for CritRats, and it's unfortunate that it is. Recall that to Popper, a word is just a label for a concept. We read definitions back to front, says Popper. There is nothing wrong with starting with the concept that we might call information that keeps itself instantiated and slapping the label knowledge on it. It's just a short hand for the concept, information that keeps itself instantiated. There's also nothing wrong with starting with the concept, adaption that causes fit to environment and slapping the label knowledge on it, okay? And indeed, that is how words grow, right? We reuse words for things that have somewhat similar or analogous meanings, and we continually expand the uses of that term over time.

[00:55:38]  Bruce: This is Hofstadter's theory that we've talked about in the podcast, okay? So it is okay, and indeed, there's no way around it, that words will have multiple definitions, or to put this a different way, we will use one word to point to multiple different concepts, okay? There's just no reason to ever get bent out of shape over this. Okay, so also remember my criticism of Deutsch's Two Sources Hypothesis here in the Walking Robot episode. Any variation selection process is by definition an example of a variant that kept itself instantiated compared to the losing variants. Now, I realize that is not what CritRats had in mind in their head when they heard Deutsch's theory, and apparently it's not what Deutsch had in mind when he came up with this theory. And that's fine, I'm just pointing out that there may be a way to bridge these two definitions if we take them both seriously. They don't even have to be seen as at odds with each other, okay? But at a minimum, we can accept that we're just using the word knowledge to point to two different concepts, and that's fine, and we can move on. We don't have to stop and get an allergy over this. So now, here you might ask, wait, what's an inductive achievement, Bruce? Or maybe you even had a strong, allergic, negative reaction to Campbell's theory that I just read, because he is acting as if inductive achievements are real, and you've heard that induction is just a myth, and there's nothing inductive and it's non-existent.

[00:57:16]  Bruce: Therefore, you can't help but think, oh, Campbell must be wrong, and I can dismiss his whole theory because he used the word induction, and I'm done now with Campbell's theory.

[00:57:28]  Unidentified speaker: Okay.

[00:57:30]  Bruce: Of course, let's try to answer the question, what is an inductive achievement? What does Campbell understand the term inductive achievement to mean? Whatever it is that it's pointing to, that concept that it's pointing to, is that concept at odds with Popper's epistemology. The answer is really pretty simple. It's exactly what we already discussed. An inductive achievement would have at the time been understood as an increase of knowledge due to adapting to fitting one's environment. In other words, you have to read all three of these as equivalent. They're just synonyms for the same thing. It's generalizing one's specific experiences or observations that go beyond what can be deduced from what you already know. So of course, that is what the term inductive mean meant at the time, and it even still means that today. Okay, because it was considered a well-known term, Campbell doesn't take much time to define it or explain it. But here's some quotes from Campbell that might be helpful. This is from the other paper, the one I'm not quoting mostly from, from page 92 of the same book. He says, inductive achievements is stage-by-stage expansions of knowledge beyond what could have been deductively derived from what has been previously known. Notice that here, inductive achievement is explicitly equated to expansions of knowledge. I had one CritRat that I will discuss in a future episode that insisted they were not the same thing, okay, but Campbell says they are.

[00:59:04]  Bruce: And Campbell goes on to say, it has represented, [inductive?] achievement has represented repeated breakouts from the limits of available wisdom. For if such expansions had represented only wise anticipations, they would have been exploiting full or partial knowledge already achieved. And then he says, page 51 now, the central requirement becomes discovering an epistemology capable of handling expansions of knowledge, breakouts from the limits of prior wisdom, scientific discovery. In short, he means by inductive achievement pretty much exactly what people normally would have meant at the time, and in fact, normally is meant today. Here's the thing, though. I anticipate CritRats in my audience are now shouting angrily, but Popper refuted induction, so there are no inductive achievements. However, Campbell explains his meaning, okay, from page 56, in the footnotes. He says, the use of the phrase inductive achievements is for convenience in communicating, and does not in the least imply advocacy of the Bacon–[Hume–Mill?] explanation of those achievements, nor disagreement with Popper's brilliant criticisms of induction.

[01:00:23]  Unidentified speaker: Okay.

[01:00:24]  Bruce: So, let me be clear here. To Popper, induction, sorry, to Campbell, induction is wholly explained by blind variation and selective retention, aka evolutionary epistemology. That is, Campbell is intentionally attempting to situate induction into Popper's epistemology and is absorbing all inductive, inductive achievement into Popper's theory.

[01:00:50]  Unidentified speaker: Okay.

[01:00:51]  Bruce: This is and always has been the correct critical rationalist approach to induction. This is why I've argued against the CritRat war on words, okay, which seems to merely boil down to a negative reaction to certain terms, like induction, right? Rather than realize the opportunity to absorb the other theory into critical rationalism by explaining induction in terms of a generalized version of Popper's scientific epistemology. So Campbell and Popper are attempting to make inductive achievement a real thing. It's just that it's actually in their minds created by a generalized form of Popper's epistemology rather than a mysterious inductive procedure or inductive logic.

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

[01:01:34]  Peter: Epistemology is complicated. So basically, basically it's, it's a true statement or, you know, however you want to put that, that Popper refuted induction and it's compatible with the assertion that induction has some validity in the world.

[01:01:53]  Bruce: So I think I would probably put it like this. We don't get upset over people's choice of words, right? So induction could be referring to a certain theory of how it is that we generalize from specific experiences.

[01:02:10]  Unidentified speaker: Okay.

[01:02:10]  Bruce: That theory is wrong and that's what Popper.

[01:02:13]  Unidentified speaker: That's refuted.

[01:02:14]  Unidentified speaker: Okay.

[01:02:15]  Bruce: But the general idea that we do generalize from specific experiences, nobody really doubts that.

[01:02:22]  Unidentified speaker: Sure.

[01:02:22]  Unidentified speaker: Yeah.

[01:02:23]  Peter: It's hard to argue with that.

[01:02:24]  Unidentified speaker: Right.

[01:02:24]  Bruce: So if you want to call that induction, then sure, of course, induction is real.

[01:02:30]  Unidentified speaker: Okay.

[01:02:30]  Unidentified speaker: Okay.

[01:02:31]  Bruce: Now, here's the key thing, though. What Popper would say is, well, you're just using the word induction to refer to my theory.

[01:02:38]  Unidentified speaker: Okay.

[01:02:39]  Bruce: That is what Popper says, right? If you want to call critical methodology induction, that's fine.

[01:02:46]  Unidentified speaker: Right?

[01:02:48]  Bruce: So if you see his theory as not induction, then there really is nothing left. Yes, we, or at least Popper's claiming there's nothing left. Let's not assume he's right. Of course, I think he is, but Popper is hypothesizing that if you look at all cases of inductive achievement and you subtract the ones that are actually just evolutionary epistemology, that the remaining set's the null set. That's what Popper's really trying to say. That's what Campbell's really trying to say.

[01:03:19]  Unidentified speaker: Okay.

[01:03:20]  Unidentified speaker: Okay.

[01:03:22]  Bruce: So this brings us to the boldest claims of Campbell's theory. Campbell's revised version of evolution does not require replicators and does not require randomness. Now, evolution is generally understood as having three main mechanisms, random variation, selection and replication. But Campbell requires only blind variation, more on that in a moment, and, quote, mechanisms for preserving and or propagating the selected variations. That and or, I'm assuming it's meant to do actual work there. So I understood that, understand that to mean, specifically preservation of variants that may or may not be replicators. So biological evolution as a theory is so tied to the idea of random variation and replicators that it's difficult for people to initially wrap their minds around this generalization that Campbell is proposing. Campbell is literally proposing a generalization to the very concept of evolution, okay? So Campbell decides to replace the term random variation with the term blind variation, and to make clear that randomness is optional. So this is from page 56. Campbell says, the word blind is used rather than the more usual random for a variety of reasons. For example, let's say an algorithm involved, and remember, everything can be thought of as an algorithm, a systemic sweep of all variations, okay?

[01:04:54]  Unidentified speaker: A full sweep.

[01:04:55]  Bruce: Campbell says, for the generalizations essayed here, certain processes involving systemic sweep scanning are recognized as blind, insofar as variations are produced without prior knowledge of which ones, if any, will furnish a select worthy encounter. This does make sense, right? Why would random, why would variation selection simply fail to work if you did a systemic sweep instead of a random checking, okay? Now, to be clear, this doesn't mean that blind means full sweep. That's not the point he's trying to make. A random choice of variations is also blind, okay? This is a generalization. It's just that a full sweep, which is clearly non-random, is also a form of blind variation, okay? In fact, a binary search across a sorted list would still be considered blind variation, even though it is neither random nor a full sweep, okay? So blind is way more encompassing. Blind variation is a way more encompassing concept, a generalization of the idea of random variation. So Campbell's point here is that a blind variation, that's what I just said, is a generalization of random variation. It includes random variations, but it also includes non-random variations that are also blind, such as but not limited to a full sweep. Now, I get a lot of pushback here from the CritRat community. Now, not typically over the blind versus random, though sometimes I see pushback on that too. But moreover, the idea that replicators are not required for evolutionary epistemology. Usually, I hear something like, but Darwinian evolution does include replicators.

[01:06:40]  Bruce: So to me, that's the correct understanding of the concept of variation and selection, or trial and error, or whatever you want to say here. Now, this response misses the whole point of the concept of generalization. So let me explain just so that I can address this argument head on. I'm taking it seriously, I'm addressing it head on. Think of a set A as the set of algorithms with blind variation and selective retention, and think of set B as the set of algorithms with blind variation and selective retention plus replicators. It should be very clear that set B is a subset of set A. Okay, the fact that you care more about the set B, doesn't make set A invalid. In fact, it doesn't make anything, but set A is still a superset of set B. So by making an argument like this, you haven't actually advanced the discussion any, okay? Because you've done nothing to show that set A, you would have to then present why set B is the only set that matters and that why we can eliminate set A, which I've never seen a CritRat who makes this argument try to do. Okay, so Campbell's Theory does not threaten the importance of replicators in biological evolution. It just points out that they are a subset of a more general form of evolution that we're going to call evolutionary epistemology.

[01:08:11]  Bruce: Okay, now, while all this may come across as very bold, and obviously the fact that there's been such a negative reaction from a lot of CritRats I've talked to about this, clearly it does cause people very gut-level negative reactions, it should probably be noted that the idea of evolution without replicators actually predates Campbell, and he didn't invent the idea. So I tried to figure out who invented it. It's a really famous idea. It seems like Dawkins brought this up in his book Selfish Gene. Okay, so it comes from HT. Odum, at least according to the Internet. In the 1950s, he came up with a generalization of evolution based on stability.

[01:08:53]  Unidentified speaker: So HT.

[01:08:53]  Bruce: Odum, 1950, from his book on pages 7 and 10 through 11, according to Wikipedia, it says, A more general statement is that a system which has stability with time will exist longer than a system without stability. Nature seeks steady state entities by natural selection. Of course, natural selection in biological systems is a special case of this principle. Now, recall also the quote I gave from Campbell that suggested that crystal formation was an example of his evolutionary epistemology process. So this is clearly where Campbell is trying to include Odum's ideas of evolution via stability into his evolutionary epistemology. So Campbell was taking the best thinking of his time from different sources, Popper being the main inspiration, but not the only inspiration. In fact, Campbell's papers have a ton of quotes from all sorts of sources showing how far back this idea of evolutionary epistemology actually goes, even though it wasn't under that term, obviously, since he coined the term. But the idea of it dates back much further, prior to Popper, okay? And part of the achievement of Campbell's work here was pulling together all these diverse sources and showing this is an idea that's been around for a while. So, one note, I will often refer to Campbell's blind variation selective retention process simply as variation and selection, dropping the blind and retention. Now, in episode 26, I took the stance that these were redundant terms. There are those that disagree with me on this, and their disagreements deserve serious discussion, which will have to be part of future podcasts.

[01:10:42]  Bruce: For the purpose of this podcast, when I say the term variation selection, think of it as just a shorthand for blind variation selective retention. I'm not trying to necessarily challenge anything at this point. It's just a mouthful to keep saying blind variation and selective retention. So, sometimes it's easier to say trial and error or to say variation and selection. Now, a huge part of Campbell's writing is showing over and over again that there is no such thing as direct knowledge. I kind of mentioned this. Let me just kind of make sure I make this clear, though. Recall that at the time, philosophers of epistemology often believed in empiricism, which was the false idea that knowledge comes direct from sense perceptions. Campbell works hard to show that this is a false theory and that sense perceptions are themselves a form of blind variation and selective retention, part of his theory, in other words, and thus are not infallible sources of knowledge. Now, of course, if you're a critical rationalist listening to this show, you already buy that idea. So for me to keep emphasizing it in Campbell's writings doesn't make sense to me. So that's where I'm concentrating more on the animal learning angle, things that I know really are going to be of interest to people because they're actually something that's controversial, okay? But some of his quotes are going to have this nature. In fact, he makes a huge, huge deal out of trying to situate vision and sight into a blind variation and selective retention process. That's why he's spending so much time on that, okay?

[01:12:17]  Bruce: Now details of the theory. So now we have Campbell's theory in his own words. Let's get into detailed examples of some of the various levels of variation selection that he believes exists. He gives 10. Let me give you all 10 first, then we're going to do a side on each one of those. So non-mnemonic problem solving would be number one. Number two, vicarious locomotor devices. Number three, habits, or in other words, animal behaviors learned in the lifetime of the animal. Number four, instinct, or in other words, animal behaviors wired into the genetics of the animal. Number five, visually supportive thought. Number six, mnemonically supportive thought. Number seven, socially vicarious exploration, observational learning and imitation. Number eight, language. Number nine, cultural culmination. And number 10, science. So let's start with number one, non-mnemonic problem solving.

[01:13:18]  Unidentified speaker: I think-

[01:13:19]  Peter: And these are all examples of creating knowledge?

[01:13:23]  Unidentified speaker: Yes.

[01:13:24]  Unidentified speaker: Okay.

[01:13:25]  Bruce: They're examples of his evolutionary epistemology, where he sees these as an increase of knowledge.

[01:13:29]  Unidentified speaker: Okay.

[01:13:31]  Bruce: So consider the example of a single celled life and its ability to solve problems via exploration.

[01:13:37]  Unidentified speaker: Okay.

[01:13:38]  Bruce: So Campbell uses the paramecium example that I used extensively in episode 26 and other episodes. It's an example of how Campbell's theory can apply to an evolutionary type process that has no replicators. That was actually why I brought it up.

[01:13:55]  Unidentified speaker: Okay.

[01:13:56]  Bruce: So imagine a single celled animal either may be searching for food or trapped in a toxic area that's killing it and needs to get out of the toxins. So it's going to try moving in every single direction until the problem is resolved. He says, quote, this is pages 57 to 58. At the level of Jenning's paramecium, homeostat, there is a blind variation of locomotor activity until a setting that is nourishing or non-noxious is found. Such problem solutions are then retained as a cessation of locomotion, as a cessation of variation. Something I want to call out here. Notice that in this case, the selective retention is simply retaining the direction it's going.

[01:14:44]  Unidentified speaker: No replicators involved.

[01:14:46]  Bruce: Okay, so one of the challenges that comes up is, is he really ruling out replicators? I feel like this example definitely shows he was ruling out replicators as being required. Now, this may surprise people that this process, a process this simple would count as a form of quote-unquote evolution. Or if you prefer, if you don't want to call it evolution, we can call it evolutionary epistemology instead. So Campbell quotes Ashby as saying, The work also, in a sense, develops a theory of the natural selection of behavior patterns. Just as in the species the truism that the dead cannot breed implies there is a fundamental tendency for the successful to replace the unsuccessful, so in the nervous system does the truism that the unstable tends to destroy itself imply that there is a fundamental tendency for the stable to replace the unstable. Just as the gene pattern in its encounter with the environment tends towards ever better adaption of the inherited form and function, so does a system of step and part functions tend towards ever better adaption of the learned behavior.

[01:16:01]  Unidentified speaker: That's from page 58.

[01:16:03]  Bruce: A few points about this level of problem solving from page 58. There is, however, no memory, no using of old solutions over again. That's, I think, why he calls it non-mnemonic. And then he says on page 58, it is these chemoreceptors and comparable organs, which in fact provide the immediate selection of responses. In other words, it has some way of detecting if it's in a nourishing environment or a toxic environment or not. That's what the selection is against, not the life and death of the organism. So thus, he says, only indirectly through selecting the selectors, does life and death relevance select the responses. So this is an example of vicarious indicators. There is evolution going on here, knowledge creation going on here, but it's not via the life and death of the organism. It's instead against a vicarious indicator of life and death. Even at this level, however, there is a need, this is a quote from page 58, there is a need, there is needed an interceptive sense organ, organ which monitors nutritional levels and substitutes for the whole organism's death.

[01:17:17]  Unidentified speaker: That was Campbell, just to be clear.

[01:17:21]  Bruce: Now in a separate 1960s article called Blind Variation Selective Retention in Creative Thought, as in other knowledge processes, what a mouthful. Campbell draws an analogy between biological evolution and animal learning from page 93. In this perspective, the epistemologically most fundamental knowledge processes are embodied in those several inventions making possible organic evolution. That is, learning, it's in scare quotes, on the part of the species by the blind variation selective retention of mutant individuals. Okay, so to be clear, that was a very wordy way of saying that the most fundamental knowledge process is biological evolution.

[01:18:02]  Unidentified speaker: Okay, that's what he's saying.

[01:18:04]  Bruce: And then he compares that to animal learning. So same page. The higher evolutionary developments shift a part of the locus of adaption away from a trial and error of whole organisms or gene pools, over to a process occurring within the single organism. Such processes are numerous, each being not only a device for obtaining knowledge, but also representing general wisdom about environmental contingencies. Note again, Campbell's understanding of knowledge is equivalent to fit to environment, or equivalent to knowledge about the environment, contained within an adaption. And that adaption could be either genetic evolution or animal learning. Interestingly, Campbell thought learning as a trial and error process was beyond needing defending. He says on page 94, the presence of a fundamental trial and error process in individual learning, remember the context here includes animals, needs no elaboration or defense. Suffice it to say that recognition of such a process is found in all learning theories which make any pretense of completeness. So again, note how this is entirely at odds with Deutsch's Two Sources Hypothesis. Campbell further develops this idea of vicarious indicators, where selection happens against some indicator that replaces selection via death. So page 94, whereas the ultimate selection is life or death in encounters with the external environment, by the evolutionary stage at which learning is possible, much of this once external criterion has been internalized, now represented by pleasures or pains or as reinforcers.

[01:19:52]  Bruce: This idea of vicarious indicators such as pleasures and pain is going to become more important at other levels. So level 2, vicarious locomotor devices. The first level is vicarious variation selection by physically moving around. What if you could avoid having to actually move and explore in your environment? So on page 58, he says, substituting for spatial exploration by locomotor trial and error are a variety of distance receptors of which a ship's radar is an example. An automated ship could explore the environment of landfalls, harbors and other ships by a trial and error of full movements and collisions. Instead, it sends out substitute locomotions in the form of a radar beam. These are selectively reflected from nearby objects. Notice the careful wording he's using there. Okay, so page 59. The process removes the trial and error component from the overt locomotion, locating instead in the blindly emitted radar beam. The radar beam is not emitted randomly, but it could be so emitted and still work. The radar beam is, however, emitted in a blind exploration, albeit a systemic sweep. Now comes the hard part. Campbell realizes his theory will fail unless he can show that vision is also a blind variation of selective retention process. If he can't show that, then empiricism wins the battle and Popper's theory is refuted. Here is how Campbell goes about it. He goes on for quite a while, and I'm gonna try to shorten it quite a bit and just try to give you a feel for his argument.

[01:21:35]  Bruce: So he says on page 59, assimilating vision to the blind variation and selective retention model is a more difficult task. Now obviously, part of the reason why is because vision is by definition not blind, right? So how do you situate vision into a blind variation and selective retention model? So he says on page 59, consider a one-photo substitute eye, such as was once distributed for use of the blind. He shows how a blind person could use the one-photo cell, which may unlike a radar [emit?] nothing, just takes light in. I'm not sure if they did or didn't [emit?] something, but still explores an environment just like a radar does via variation selection process. So let's say, so still if we're using, we can imagine a single photo cell that takes light in, and it's not emitting something, and you're trying to use that to determine different colors of light to understand your environment. The fact that nothing's being emitted, does that make it less a variation and selection process? So Campbell's arguing it doesn't. It's still the same variation and selection process that a radar would be, okay? Now let's say that you put many photo cells together, pointing in various directions that improves this process. If enough of these are put together, at some point we could think of the eye as being such a photoreceptor, with a whole bunch of photoreceptors pointing in different directions, okay? So this eventually becomes regular vision at some point, okay?

[01:23:07]  Bruce: Now, no one doubts that the radar sweep is a blind variation and selective retention process that replaces searching via blind locomotion. Where people would tend to doubt is that vision is also such a blind variation process. Yet few would doubt that the one photo cell detector used by a blind person is still a blind variation and selective retention process, despite nothing being [emitted?]. So why does it stop being a blind variation process just because you put a lot of them together into the human eye? Okay, this is the essence of Campbell's argument. Indeed, isn't the human eye actually just a bunch of photoreceptors blindly pointing in many different directions, hoping to notice something? And then at another level, don't the eyes flit around somewhat blindly to take in the scene? Consider the following comment from Campbell as typical of his view. From page 95, vision is a very complex and marvelous mechanism, and the brief presentation here does not do justice even to the random search component involved. Hebb, 1949, has well documented the active search of eye movements, correcting the model of the inactive fixed focus eye, which is implicit in both Gestalt psychology and conditioning theory. Riggs and Ditchburn have documented the essential role of the continuous low-amplitude scanning.

[01:24:35]  Bruce: And then Campbell gives a few more examples, and then he says, These and other considerations convince the present writer that although vision represents the strongest challenge to the generality of a blind variation selective retention aspect to all knowledge processes, it is not in fact an exception. So Campbell argues that vision is not an empiricist process, as people thought, but instead it is a Popperian style example of evolutionary epistemology. Campbell now moves on to, and he handles these two in the same section, number three, habit, which is learned in the lifetime of the animal, and number four, instinct, which is wired into the genetics of the animal. So, on page 60, he says, knowledge about the world through expectation and presumptions, and then he goes on in page 61 to say, instinct development can be seen as involving a trial and error of whole mutant animals, whereas trial and error learning involves the much cheaper wastage of responses within the lifetime of a single animal. The same environment is editing habit and instinct development in most cases. And the editing process is analogous, and the epistemological status of the knowledge innate or learned is no different.

[01:25:57]  Unidentified speaker: That was from page 61.

[01:25:59]  Bruce: The idea Deutsch proposed, that all an animal's knowledge is in its genes, has the curious and unfortunate effect of insisting that if an adaption is instinct, instinct, it counts as knowledge. But if the very same adaption came through animal learning, it no longer counts as knowledge. This is why I consider that idea refuted, by the way. Now, on page 61, he says, the same environment is editing habit and instinct development in most cases. The editing process is analogous and the epistemological status of the knowledge. Innate or learned is no different. Then he says, all comprehensive learning theories, including those of Gestalt and Inspiration, contain a trial and error component. Now, this does pose a problem. How in the world would instinct ever evolve under neo-Darwinian evolution in the first place? Now, I've talked about this extensively in episode 38, Animal Learning and Popper's Epistemology. By the way, the animal learning episodes that I mentioned previously, that would probably be episode 37, 38, 39, and 40, I think. So Campbell raises the very same thing that I do in those episodes. In fact, I'm pretty sure I just got the idea from reading Campbell. Where we can really only explain evolution of instinct in light of an animal's ability to actually learn in its lifetime.

[01:27:23]  Unidentified speaker: In fact, there's a term for this.

[01:27:24]  Bruce: It's called the Baldwin Effect. It looks like Lamarckism, but it's actually compatible with neo-Darwinian evolution. The idea is that it should be impossible for complex instinctual behaviors to evolve one body movement at a time, because you need the whole set of movements to be adaptive. Here, think of like Deutsch's example. He often uses a squirrel that needs to bury a nut. So it uses this leg and then that leg, and it tries to dig, and then it tries to move the nut into the hole it dug, and then it tries to cover it with dirt, and it'll even attempt to do this on cement, right?

[01:28:00]  Unidentified speaker: It has no idea.

[01:28:01]  Peter: That's what the Baldwin Effect is?

[01:28:03]  Unidentified speaker: You're talking about?

[01:28:04]  Unidentified speaker: Not quite.

[01:28:04]  Unidentified speaker: I'm getting there.

[01:28:06]  Bruce: What I'm trying to do is I'm trying to set up the problem the Baldwin Effect solves. So those movements of its arms are not adaptive. The movement of its nose, to bury, are not adaptive. None of the individual movements in this process are adaptive. So how would Neo-Darwinian evolution ever create that automatic movement in the first place as instinct?

[01:28:30]  Unidentified speaker: Like think about that for a second.

[01:28:32]  Bruce: It's literally impossible on its own for instinct to ever evolve one movement at a time like this under Neo-Darwinian evolution. And in fact, you know, creationists might use that as a proof that evolution is impossible and that something's missing. And in fact, they're kind of right, there is something missing, but it's not actually a problem with Neo-Darwinian evolution. So here is Campbell, how he explains the Baldwin Effect to solve this problem. On page 61, he says, The adaptive pattern, being thus piloted by learning, any muta- being initialized by learning is what he means, any mutation in the genes that accelerated the learning made it more certain to occur or predispose the animal to certain component responses would be adaptive and selected no matter which component or in what order affected. The habit thus provided a selective template around which the instinctive component could be assembled. Stating it in other terms, learned habits make a new ecological niche available, which niche then selects instinctive components. It is furthermore typical of such instincts that they involve learned components.

[01:29:47]  Unidentified speaker: That's all from page 61.

[01:29:48]  Bruce: Okay, let me state that in English in case it wasn't obvious what he just said. The idea of the Baldwin Effect is that- let's take our squirrel example. Some proto-squirrel learned to bury nuts using its learning algorithms. No knowledge in the genes at all. And then once that passed through the populace through animal memes, let's say, you then have biological evolution can make it so that, let's say, one of those proto-squirrels had a built-in instinct to try to start digging when it had too many nuts, okay, and that's it. Just some really, really basic sort of movement. It doesn't have the rest of the instinct yet. But that move is now adaptive because it exists within an ecology of a bunch of proto-squirrels that all by learning already know how to bury nuts. So that instinct to start burying makes it easier for that new squirrel, that new mutant squirrel, proto-squirrel, I should say, to be able to learn that whole set of mechanisms. Because now it doesn't have to learn the whole thing. It only has to learn part of it. Okay, that's the Baldwin effect. It's this idea that the learned knowledge must come first and the instinct must come second. Okay, it's actually more complicated than that. I'm making this far too easy. Like it may even just start off as an instinct to hide your nuts. I mean, obviously, it could be something much simpler initially and there could be a complex interaction of learned knowledge and instinctual knowledge. But instincts, the ability of an animal to have instincts only exists if the animal first has an ability to learn.

[01:31:43]  Bruce: And that means that learning algorithms, the ability of an animal to create knowledge in its lifetime that was not in its genes, had to exist for evolution of instincts to ever happen in the first place, okay? Popper took that further. He claimed that you could not even explain physical evolution without understanding something like the Baldwin Effect. The example I used in past podcasts is imagine a Boston Dynamics robot, which originally they were programmed without any sort of learning algorithms, and imagine that you suddenly gave it better hands or a better engine, right? Its software wouldn't know about those better hands or better engines, so even though it has an improved hands or engine, it would actually be detrimental to it because the software would be miscalibrated, okay? If animals were like that, then physical evolution wouldn't work, right? So the fact that physical evolution takes place at all has to be understood in terms of something like the Baldwin Effect. According to Popper, this is a theory, he was a conjecture in any case, where animals must have evolved learning algorithms first, and that allowed so that when they had some sort of physical improvement, physical claws improved or something like that, that they were able to learn how to use them, and that it actually was a positive adaption.

[01:33:08]  Peter: Can I just read what ChatGBT says? I think it clarifies it very simply, at least in my mind. If you can learn something that helps you survive and reproduce, the natural selection might eventually favor genes that make it easier for your descendants to learn or even hardwire that behavior.

[01:33:29]  Bruce: That's the Baldwin Effect, yes.

[01:33:31]  Unidentified speaker: Yes.

[01:33:32]  Unidentified speaker: Okay.

[01:33:32]  Unidentified speaker: Just trying to get my mind around that.

[01:33:35]  Unidentified speaker: Okay.

[01:33:36]  Bruce: To be clear, the Baldwin Effect says that it's completely impossible to reconcile Deutsch's idea that all an animal's knowledge is in his genes with neo-Darwinism. That it's absolutely what the Baldwin Effect is saying.

[01:33:49]  Unidentified speaker: And is the Baldwin Effect controversial or not?

[01:33:51]  Bruce: No, not in the slightest.

[01:33:53]  Unidentified speaker: Not in the slightest.

[01:33:54]  Unidentified speaker: Okay.

[01:33:54]  Unidentified speaker: Yeah.

[01:33:54]  Bruce: Campbell uses this example of the Baldwin Effect. So I'm harping on it because of its relevance to the CritRat community, but Campbell actually is using the Baldwin Effect to explain one of his main ideas. And he says on page 61, this can be conceived as an evolution of increasingly specific selection criteria. This is, he's using it to build this idea of hierarchies of evolutionary epistemology, hierarchies of blind variation, selective retention algorithms, right? So instinct and habit existed in a hierarchy together, and they're both themselves kinds of evolutionary epistemology. But they modify each other, okay? So he's using this to bring up this idea of a hierarchy of blind variation selection retention. So he understands blind variation as not really being totally blind, but merely guided by knowledge created higher in the hierarchy. Now, let's go move number five, visually supported thought. We kind of already covered this. This is a little different than what we just said. That was vision itself. So he says on page 62, with the environment represented vicariously through visual search, the previous one we just discussed, there is a substitute trial and error of potential locomotion in thought. Page 62, the successful locomotion at this substitute level, with its substitute selective criteria, are then put into overt locomotion, where they appear intelligent, purposeful, or insightful.

[01:35:24]  Bruce: To put this in just in plain English, he's saying, sight is itself a blind variation selective retention process, and then use of sight is a separate blind variation selective retention process, where you're in your head maybe using that sight to determine where you're going to move. There's some sort of blind variation selective retention process that's going on in your head based on what you're seeing, that's separate from the sight itself, which is itself a blind variation selective retention process. And when you move, this is an animal, it's moving, okay, not necessarily a human thought, that movement is going to seem very intelligent, purposeful, sighted, it's going to seem like that, but really the whole thing is a hierarchy of blind variation selective retention processes.

[01:36:13]  Unidentified speaker: That's Campbell's point.

[01:36:14]  Unidentified speaker: Okay, does that make sense?

[01:36:15]  Unidentified speaker: I think so, yeah.

[01:36:16]  Bruce: Okay, page 111. Processes such as vision and thought substitute for these overt trial and error are of course acknowledged. But each of these is interpreted as containing in its very workable wisdom about the environment originally obtained originally by the blind variation and mutation and natural selection. In addition, each contains a blind variation and selective retention process at its own level. Number six, mnemonically supported thought. So this would be thought supported by memory, I guess is what he's trying to say with mnemonically. I've never quite understood his use of that term, but that's what I think it means.

[01:36:57]  Unidentified speaker: So page 62.

[01:36:59]  Bruce: At this level, the environment being searched is vicariously represented in memory or by knowledge rather than visually. The blindly emitted vicarious thought trials being selected by a vicarious criteria substituting for an external state of affairs. In short, an animal can now at this level, can now work out what it will do via trial and error in its head, rather than just acting immediately based on what it sees or trying to move around randomly, something along those lines. This is what Byrne called animals with insight. I should probably note that he's using it in a more generalized sense than Byrne does. So probably a dog does have mnemonically supported thought. Otherwise, how would it know how to find its way around a house, say? So we're probably talking about something more primitive than Byrne's animals with insight, but similar to it. On page 62, he says, The net result is the intelligent, these are all in scare quotes, intelligent, creative and foresightful product of thought, our admiration for which makes us extremely reluctant to subsume it under the blind variation selective retention model. So he's trying to explain why it looks like it's sighted, right? And that it knows what it's doing, and it's able to just induce this knowledge, right? But really under the hood, it's a collection of blind variation selective retention processes. That's his argument here, okay? This is also the level where Campbell sees human creativity is starting, though he shows animals can, to a more limited extent, do this as well.

[01:38:39]  Bruce: Humans try out ideas in their head, just like many animals do. But humans can do so using language theories, explanations, etc. Therefore, humans can do this at a whole new level than animals can. Many of Campbell's contemporaries really could not accept that human creativity could possibly be a blind variation selective retention process. Obviously, the CritRat community does accept that, that it's a type of evolutionary epistemology. So Campbell dwells on this a bit to try to show how it fits into his theory. This part's really important, okay? I'm going to kind of emphasize and spend a little time on this, and then I'm going to speed up, go through the rest of his, and we can call it good for today. But I really feel like this part is going to be the part that causes the biggest problems and explosions and hatreds towards Campbell's theory within the CritRat community, okay? So Campbell emphasizes creative thought as substitution. In a different paper, the second paper on page 96 of the same book, he says, creative thought provides a substitute exploration of a substitute representation of the environment, the solution being selected from the multifarious exploratory thought trials according to a criterion, which is in itself substituting for an external state of affairs, very similar to the other quote I read. The problem is that when humans and even animals act, it doesn't appear to be random and blind fumbling process at all.

[01:40:10]  Bruce: Thus, other philosophers saw this as proof that Campbell's theory was incorrect and that blind variation is not at the bottom of something like human creativity. But, Campbell argues, page 96, insofar as the three substitutions are accurate, the solutions when put into overt locomotion are adaptive, leading to intelligent behavior which lacks overt blind floundering and is thus a knowledge process. To put this in plain English, the reason why humans don't seem like they're blindly fumbling around is because they think about alternatives in their head first, use criteria to select, to criticize, select which one's the best one, then they do that one. So, of course, it seems intelligent. But the actual process is still a perspiration algorithm, right? They're actually trying out different ideas, they're criticizing them in their head, and it's still a blind variation selection process. Now, I want to call out the CritRat community a little on this. I don't think there's any doubt that they would accept that argument for humans. Like, that argument, they'd say, yeah, of course, that's exactly how humans think. But they really struggle with accepting that animals do it, too.

[01:41:26]  Unidentified speaker: Okay.

[01:41:29]  Bruce: Now, Campbell quotes other experts and shows that many have already shown that human thought is based on blind variation, and that, in fact, that idea is older. So quoting Souriau, I don't even know who that is, some famous philosopher, on page 64, he says, by a kind of artificial selection, we can, in addition, substantially perfect our thought and make it more and more logical. Continuing the quote, those persons who, upon considering the marvelous results at which knowledge has arrived, cannot imagine that the human mind could achieve this by simple fumbling. Do not bear in mind the great number of scholars working at the same time on the same problem, and how much time even the smallest discovery costs them. Page 97 now, same person being quoted, he says, There is something mechanical, so to speak, in the art of finding solutions. Thus, we see the role of logic diminish and that of chance increase as we approach closer to true invention. Notice that Campbell is here, quoting Souriau, is challenging the Deutschian distinction between perspiration and inspiration. Because human inspiration is a perspiration process.

[01:42:47]  Unidentified speaker: Of course it is.

[01:42:49]  Bruce: How could it be anything else if it's rooted in evolutionary epistemology, which implies trial and error? So continuing Souriau, he says, Even genius has need of patience. It is after hours and years of meditation that the sought-after idea presents itself to the inventor. He does not succeed without going astray many times, and if he thinks himself to have succeeded without effort, it is only because the joy of having succeeded has made him forget all the fatigues, all the false leads, all the agonies, with which he has paid for his success. So, interesting side note, trial and error was originally used, that term, trial and error, was originally used to describe human thinking, not Darwinian evolution. So, from page 97, Campbell says, historically, however, the phrase trial and error was first used to describe thinking by Alexander Bain as early as 1855, two years before Darwin's publication of the Doctrine of Natural Selection. Interestingly, Campbell sees this level of mnemonically supported thought as including not just humans and animals, but computer problem-solving. This is where we're going to have an eruption with the CritRat community. By the time Campbell wrote his 1974 paper, the term artificial intelligence had been coined at a Dartmouth Conference. At that conference, Herbert Simon showed off Logic Theorist, which is today considered the first artificial intelligence program. And interestingly, Simon himself took issue with Campbell's theory and offered Logic Theorist as a refuting counter-example to Campbell's theory.

[01:44:39]  Bruce: But Campbell then explained that Simon had misunderstood his theory and that artificial intelligence programs like Logic Theorist fit neatly into his theory. So, continuing here, page 66, he says, Computer problem-solving is a highly relevant topic and is perhaps best introduced at this point. Like thinking, it requires vicarious explorations of a vicarious representation of the environment and the exploratory trials being selected by criteria which are vicarious representatives of solution requirements or external realities. Page 66, he says, The present writer would insist here, too, that if discovery or expansion of knowledge are achieved, blind variation is requisite. So, there's no doubt that Campbell is including artificial intelligence into his epistemology, evolutionary epistemology, okay? Note that this is a possible way to test Campbell's theory. If there are artificial intelligence programs that show off inductive achievement or increases in knowledge without blind variation, then we have refuted Campbell's theory. So what was Simon's criticism using logic theorists? Here is Campbell's account.

[01:46:00]  Unidentified speaker: Page 66.

[01:46:02]  Bruce: It is only fair to note that Herbert Simon, both a leading computer simulator of thought, and notice that he calls it simulator of thought. Artificial intelligence really believed they were back then in the 1950s or whatever when the Dartmouth Conference took place, they seriously thought they almost had AGI worked out.

[01:46:21]  Unidentified speaker: Okay.

[01:46:21]  Bruce: So they considered themselves simulating thought, right?

[01:46:25]  Unidentified speaker: Yeah.

[01:46:25]  Peter: People have thought that for a long time now, I guess.

[01:46:28]  Bruce: I have a blog post on this on my company website and a video about it. I will actually link to those because I think they're a little bit funny. It's the pre-history of artificial intelligence, the history of artificial intelligence starting with like Aristotle and moving forward in time until the Dartmouth Conference where the term artificial intelligence got coined.

[01:46:47]  Unidentified speaker: Okay.

[01:46:48]  Bruce: They all thought that they were working out general intelligence, right? They really believed they were discovering the laws of human thought. So let me go back and read that now that I've said that. Herbert Simon, both a leading computer simulator of thought, and an epistemologically sophisticated scholar, rejects my point of view, at least in the extreme form advocated here. For example, Simon says, the more difficult and novel the problem, the greater is likely to be the amount of trial and error required to find a solution. At the same time, the trial and error is not completely random or blind. It is, in fact, highly selective. Now, at this point, you probably have a pretty good idea from the way I've formulated this that that's not really a counter example to Campbell's theory because he's actually talking about knowledge coming from a higher node in the tree. Campbell's now going to explain that in context of how AI works. He never uses the term AI, by the way, but we know he's talking about AI because he's talking specifically about the first artificial intelligence program. The selectivity, insofar as it is appropriate, this is from page 66 to 67, represents already achieved wisdom of a more general sort. And as such, selectivity does not in any sense explain an innovative solution. Insofar as selectivity is inappropriate, it limits areas of search in which a solution might be found and rules out classes of possible solutions. Insofar as the selectivity represents a partial general truth, some unusual solutions are ruled out.

[01:48:29]  Bruce: Simon's heuristic, used in his Logic Theorist, are such partial truths. And a computer which would generate its own heuristics would have to do so by a blind trial and error of heuristic principles, selection from which would represent achieved general knowledge. The principle of hierarchy in problem-solving depends upon such discoveries, and once again can, of course, greatly reduce the total search space, but without at all violating the requirement of blindness as here conceived.

[01:49:03]  Unidentified speaker: Okay.

[01:49:04]  Bruce: In plain English, he's saying, yeah, of course the Logic Theorist used was selective in what trials it actually tried. But that's a heuristic, and a heuristic itself must come from a blind variation selective retention process. Therefore, this does not violate my theory. That's Campbell's argument in a nutshell. For example, one of Simon's, one of the heuristics used in Simon's Logic Theorist program, this is a quote from Campbell, is that any substitution or transformation which will increase the similarity between a proposition and the desired outcome should be retained as a stem from which further variations are to be tried. Any transformation decreasing similarity should be discarded. Similarity is crudely scored by counting the number of identical terms, with more points for similarity of location. This rule enables selection to be introduced at each transformational stage, greatly reducing the total search space. It employs an already achieved partial truth. It produces computer search similar to human problem-solving, in failing to discover roundabout solutions, requiring initial decreases in similarity. Beyond thus applying what is already known, albeit only a partial truth, the new discoveries must be produced by a blind generation of alternatives. So here Campbell clarifies several things about his theory. The fact that a blind variation process is highly selective does not represent a counter example to his theory, so long as the heuristic used to reduce the blindness of the process was itself created by a blind variation process.

[01:50:49]  Bruce: Campbell also clearly sees heuristics as being themselves in some sense still blind, since they do not represent infallible truths and may even cause the search process to miss out on an improved solution. Campbell was apparently not familiar with the difference between an admissible and inadmissible heuristic. Admissible heuristics do not cause the search process to ever miss a better solution. Nevertheless, this does give us valuable insight into how Campbell understands blindness. In fact, let me put this somewhat bluntly. Campbell defines blind variation in such such that even an apparent sighted variation process will always count as a blind variation, so long as the sight came from a blind variation process. This will become important later for a future podcast, for one of my disagreements with another member of the CritRat community about Campbell's Theory. Okay, moving on. Number seven, socially vicarious exploration, observational learning and imitation, page 67. In this, the trial and error exploration of one member of the group substitutes for renders unnecessary trial and error exploration on the part of other members. The use of trial and error by scouts on the part of migrating social insects and humans, human bands, illustrates this general knowledge process. Notice he's again including animals. At the simplest level in social animals are procedures whereby one animal can profit from observing the consequences to another of that other's acts, even or especially when the acts are fatal to the model. He then gives examples of apes, ants, termites doing this.

[01:52:37]  Bruce: So here Campbell is talking about humans and animals, and the idea is is that you can learn by observing what others do, instead of having to learn it yourself. Page 67 to 68, he says, Also noted in social animals, perhaps particularly in their young, is a tendency to imitate the actions of other animals, even when the outcomes of those actions cannot be observed. This is much more presumptive but still rational procedure. It involves the assumption that the model animal is capable of learning and is living in a learnable world. If this is so, then the model has probably eliminated punished responses and has increased its tendency to make rewarded responses, resulting in a net output of predominantly rewarded response. The more so, the longer the learning period and the stabler the environment. To put this in plain English, this is exactly what we talked about. That theory that we talked about in our last episode, last record, I don't know if they'll be aired in order or not. But the idea that we have a tendency to want to imitate or learn from other people, particularly if they are prestigious or they show some sorts of types of competence.

[01:53:53]  Peter: Yeah, and that's why I learned so much from you, Bruce.

[01:53:55]  Bruce: Exactly, exactly, right?

[01:54:01]  Bruce: The argument Campbell is using here is that the evolution of this impulse to imitate is based on this idea that an older animal has had a chance to modify its behaviors against the environment, learn from the environment, gain knowledge from the environment, created knowledge that isn't part of the gene pool. And that therefore it makes sense that the younger animals that don't have this knowledge yet, because it's not part of the genes, that they will have learned, they can learn by imitating an older animal, because that older animal has already had a chance to figure out which behaviors are going to result in rewards and which ones are going to result in punishments by the environment. Okay, that's the argument that he's using here. Number eight, language.

[01:54:51]  Unidentified speaker: Page 68.

[01:54:52]  Bruce: Overlapping with levels six and seven, above is language, in which the outcome of explorations can be relayed from scout to follower with neither the illustrative locomotion nor the environment explored being present, not even visually vicariously present. So, but that's not just with humans, by the way. He says, page 69, ants and termites have independently discovered the use of pheromones for this purpose. He's referring to this as language, maybe we should call it communication, okay. An explorer who has encountered food exudes a special external hormone on his walk back to the nest. The other workers backtrack on the special scent. If they too are successful, if the food supply remains plentiful, they keep the hormone track renewed. That's a variation selection process, okay. The knowledge of the environment upon which the worker bases his trip is profoundly indirect. That would be an example of language being a variation selection process. Campbell, to situate language in his theory, develops, and this is a really interesting thing he comes up with that I totally think deserves some thought and some consideration. He develops a Popperian model of language training. So, note the relevance to Hofstadter's theories from Episodes 85 and 86 here. The idea is that there is no way to directly transmit the meaning of a word from person to person. This is the bucket theory of mind, obviously. Thus, each person must discover the meaning themselves via conjecture and refutation, based on how their community uses a term.

[01:56:38]  Bruce: Note that this is identical to Deutsch's theory about how and why creativity evolved to allow humans to transmit memes and the older probabilistic approach that animals used, without explanations, couldn't handle the requirements well enough. So, humans evolved creativity to expand their memetic capacity. Note how Campbell explains this, and I think you'll see strong agreement between Campbell and Deutsch on this part.

[01:57:04]  Unidentified speaker: Page 69.

[01:57:06]  Bruce: For human language too, the representability of things and actions by words is a contingent discovery, a non-entailed relationship, and only approximate. Then he says, we need a Popperian model of language learning, I wasn't making that term up, in the child and of language development in the race. Regarding the child, this would emphasize the word meanings cannot be directly transferred to the child. Rather, the child must discover these by a presumptive trial and error of meanings, which the initial instance only limits but does not determine.

[01:57:48]  Unidentified speaker: Page 69.

[01:57:49]  Bruce: Rather than logically complete ostensive definitions being possible, there are instead extended incomplete sets of ostensive instances, each instance of which equivocally leaves possible multiple interpretations, although the whole series edits out many wrong trial meanings. The logical nature of children's errors in word usage amply testifies to such a process, and testifies against an inductivist version of a child's passively observing the adult usage contingencies. This trial and error of meanings requires more than the communication of mentor and child, requires a third party of objects referred to. Language cannot be taught by telephone, but requires visually or [tactually?], present ostensive reference stimulating and editing the trial meanings. Now here is where things get interesting. If you accept the trial and error approach to language, then this has implications as to why it is that words never have single precise essential definitions. This is something I've emphasized on this show a lot, right? For it is not possible for there to not be mutant usages of terms where no two people understand a term exactly the same. But some terms and meanings will become more effective than others at spreading and having utility presumably due to having close enough meanings in the heads of others in the community to allow communication. All of that was, all of what I'm about to read now comes from pages 69 to 70. Quote, a social trial and error of meanings and namings can be envisioned.

[01:59:38]  Bruce: Trial wordings, words designating reference, which the other speaker in the community rarely correct—guess correctly, either fail to become common coinage, or are vulgarized towards commonly guessed designations. All words have to go through the teaching sieve, have to be used, have to be usefully, if incompletely communicable by finite steps of ostensive instances. Stable, sharp, striking object boundaries useful in manipulating the environment have a greater likelihood of utilization in word meanings than do subtler designations, and when used, achieve a greater universality of meaning within the community of speakers. Such natural boundaries for words exist in much greater number than are actually used, and the alternate boundaries for highly overlapping concepts abounds. This is what I've been saying all along.

[02:00:37]  Unidentified speaker: A word has multiple meanings.

[02:00:38]  Bruce: We use multiple words for the same meaning. That's just the way words evolve. Just as certain knowledge is never achieved in science, so certain equivalence of word meanings is never achieved in the iterative trial and error of meanings in language learning. This equivocality and heterogeneity of meanings is more than trivial logical technicality. It is a practical fringe imperfection, and even where meanings uniform, the word-to-object equivalence is a [corrigible?] contingent relationship, a product of a trial and error of metaphors. This is exactly what Hofstadter's theory says. Of greater and greater appropriateness, but never complete perfection, never a formal nor entailed isomorphism. This is almost exactly what Hofstadter's theory was. Recall from Episodes 58 to—sorry, 85 to 86 that the CritRat community rejected Hofstadter's theory as inductive, and even Hofstadter himself claimed it was an inductive theory. But now we're starting to see the problem with merely having an allergy to the term induction. For a proper understanding of induction for a critical rationalist is to do what Campbell just did, explain what is going on in terms of a generalization of Popper's epistemology. When put in this way, in the way Campbell puts it, it's easy to see that Hofstadter's theory actually fits very nicely with critical rationalism, despite its supposed inductive flavor. Number nine, cultural culmination. What today we’d call memes, but also includes cultural ideas, institutions, social standing, etc. And finally, number ten, science.

[02:02:30]  Bruce: The demarcation, this is quotes from page 71, the demarcation of science from other speculations is that the knowledge claims to be testable and that there be available mechanisms for testing or selecting which are more than social.

[02:02:45]  Unidentified speaker: i.e.

[02:02:45]  Bruce: what I've called hard objective criticisms vs.

[02:02:50]  Unidentified speaker: subjective criticisms.

[02:02:51]  Bruce: Campbell compares social selection of, say, religion or fads against science, which quote, What is characteristic of science is that the selective system which weeds out among the variety of conjectures involves deliberate contact with the environment through experiment and quantified prediction, designed so that outcomes quite independent of the preferences of the investigator are possible. This is the subjective-objective divide that makes science special and makes rapid progress possible. Quote, it is preeminently this feature that gives science its greater objectivity and its claim to a culminating increase in the accuracy with which it describes the world. This has some relevance to our discussion about Deutsch's theory of static societies, namely to the question of which came first, the culture of criticism or the scientific revolution and which is really the start of rapid progress. So now, here is, from page 89, Campbell's own summary of his own work here. This essay has identified Popper as the modern founder and leading advocate of a natural selection epistemology. The characteristic focus is on the growth of knowledge. The problem of knowledge is so defined that the knowledge of other animals than man is included. The variation selective retention process of evolutionary adaption is generalized to cover a nested hierarchy of vicarious knowledge processes, including vision, thought, imitation, linguistic instruction and science. Okay, let me give you one more thing and then we're going to call it quits.

[02:04:30]  Unidentified speaker: All right.

[02:04:31]  Bruce: I'm going to read to you the first two pages of Popper's reaction to Campbell's theories.

[02:04:37]  Unidentified speaker: Oh, this will be interesting.

[02:04:39]  Bruce: It's from the exact same book. This is from page 115. Okay, we're going to cover Campbell's response in-depth in a future podcast. Because I really think it's very important what Popper says about Campbell's theories. But this is a teaser for it. Okay, first two paragraphs. Imagine my voice with a German accent. Professor Campbell's remarkable contribution, which is chapter two, the main article that I've been utilizing, shows the greatest achievement, agreement, with my epistemology and what he cannot know, an astonishing anticipation of some things which I had not yet published when he wrote his paper. In addition, it is a treaty—how do you pronounce the term? Treaty—of prodigious historical learning. There is scarcely anything in the whole of modern epistemology to compare with it. Certainly, not in my own work. His historical references are all highly relevant. They are a real treasure house, and they often surprised me greatly. For me, the most striking thing about Campbell's essay is the almost complete agreement down even to minute details between Campbell's views and my own. I shall try to develop one or two of these points a little further still, and shall then turn to the very rare and comparatively minor points where there may be some difference of opinion. And that is the first two paragraphs of Popper's response to Campbell. So as you can see, Popper really liked Campbell's Evolutionary Epistemology. That is an astonishingly strong endorsement.

[02:06:30]  Peter: Yeah, you don't really hear him endorse too many other people.

[02:06:35]  Bruce: I don't think I've ever heard him say something like that. And when we get to what he actually disagreed with Campbell on, it truly is minor points, like surprisingly minor points. And this is why I honestly think that we could refer to Evolutionary Epistemology as Campbell and Popper's Evolutionary Epistemology. It's obvious that a huge bulk of it came from Popper. Campbell developed it further. Popper greatly agreed with the way in which Campbell developed it.

[02:07:05]  Peter: And did Campbell ever write any books for laymen in the way that Popper did?

[02:07:10]  Bruce: I have no idea. That might be worth looking into. Obviously, he isn't as well remembered as Popper. As I've kind of argued here, he's had a huge influence, right? Even if he himself has been sort of forgotten. His ideas were explosive and really did make a huge difference in the way we perceive Popper's epistemology today. That's really all I had for today. We're going to get way more into this, not necessarily back to back, but this is going to be another thread we're going to pull. And I actually think these threads are all going to end up tying together, right? Like, we're exploring static societies, we're exploring Campbell's Evolutionary Epistemology, Campbell Popper's Evolutionary Epistemology, we're exploring probability theory and Deutsch's treatment of probability theory. We're kind of therefore using that as a jumping off place to talk about Bayesian reasoning and Bayesian epistemology. I think all these ideas end up tying together, and that there's actually a unified whole that we need to understand to make sense of these theories. And so that's why I'm kind of going to intermix them, and we're going to talk them through as we go.

[02:08:26]  Unidentified speaker: Okay.

[02:08:28]  Bruce: And let me just say that I don't agree with Campbell's theory. I have offered what I think are counter examples to it. Therefore I think there is something wrong with the way it's formulated. That doesn't really reduce my admiration for the theory. One could say that I don't agree with Popper on some things. Therefore I have rejected quote unquote Popper's epistemology. But of course the epistemology that I actually accept is just a modified version of that, right? And piecewise improvements on what is Popper's epistemology. I guess I sort of feel the same way. There are huge swaths of Campbell's theory that I do accept. At a minimum, I accept that in cases where there is a trial and error process, that those are knowledge creating processes. I think there's some doubt whether that's the only way to do it. But surely in the cases that match a trial and error process, I do agree with him.

[02:09:22]  Peter: Okay, well, I don't know how you do it week after week, Bruce.

[02:09:26]  Unidentified speaker: You know what?

[02:09:27]  Bruce: This episode probably took me 30 or 40 hours to put together.

[02:09:33]  Peter: Well, I hope you're making time in your life to think about other things other than epistemology.

[02:09:39]  Unidentified speaker: Oh, it's...

[02:09:40]  Bruce: Yeah, I need to probably... That's why we sometimes do easier episodes.

[02:09:46]  Unidentified speaker: Okay.

[02:09:47]  Peter: Well, I've loved listening to you, and I think in your own way, you are forging new ground, at least in the level of pulling together things from other people. And I value you a lot.

[02:10:04]  Unidentified speaker: So thank you, Bruce.

[02:10:04]  Unidentified speaker: Thank you very much.

[02:10:13]  Unidentified speaker: Hello again.

[02:10:14]  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.

[02:10:57]  Unidentified speaker: Thank you.