Episode 144: Memes, Words, and Concepts

• Listen to this episode: 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 talk about meaning memes, words, concepts, and conceptual leakage. At least from what I get, this is Bruce's exploration into where concepts come from. How is a meaning meme turned into a broader concept and then a theory? Maybe this is a pretty good description for what humans do in this universe with our minds. We move memes to concepts to theories, aided along the way by our unique capacity to use words. I hope our listeners enjoy this shortest episode ever.

[00:00:51]  Bruce: Welcome to The Theory of Anything podcast. Hey, Peter.

[00:00:53]  Peter: Hey, Bruce. How are you doing?

[00:00:56]  Bruce: Good. We're going to do a really short one today.

[00:00:58]  Peter: Famous last words.

[00:01:00]  Bruce: So this is actually just me reading some tweets that I wrote a while ago. I called this episode Meaning Memes, Words, Concepts, and Conceptual Leakage. I had a really interesting conversation with somebody out on X, and I thought, you know what? This would make an interesting podcast, so I'm going to just read and comment on what I wrote. It was me trying to get my thoughts together on a certain subject. The background here was that we were talking about Michael Levin, who I'm a big fan of. Michael Levin, we've done some episodes on him. He's the guy who created like the, what are they called? I want to say xenomorph, but that's the aliens from Alien. He created the little xenorobots that are biological machines created out of cells, and he got it so that you could have like the planarian worms so that they have two heads. And then he showed that the two heads will replicate, even though the genes are exactly the same with or without two heads.

[00:02:08]  Peter: I had no idea. Did we talk about that? I had no idea he did that. But I know he's a Third Way evolution guy.

[00:02:16]  Bruce: So he doesn't actually identify with third way, but like he's very similar to third way.

[00:02:23]  Peter: Okay.

[00:02:23]  Bruce: We used him like when we were doing our episodes on third wayers.

[00:02:27]  Peter: Yeah.

[00:02:28]  Bruce: I used him as an example of like everybody argues with the third wayers, but nobody argues with Michael Levin. And I said, this is like trying to argue with Superman. Like he's got the evidence, there's just nothing you can do. So he doesn't strictly align himself with third way, but like he's very similar. Right?

[00:02:47]  Peter: Okay.

[00:02:49]  Bruce: Okay, here's the beginning. I'm not going to read the other guy's tweets; I'm just going to read my own. So you point out that Levin is clearly religious, but unwilling to openly put it on the table. I agree this is true, but perhaps unlike you, I think this is exactly the right approach. There are no bad sources of conjecture. Hard-headed realists often struggle with this, but from a Popperian standpoint, literally anything is an acceptable source of conjecture. By necessity, that includes religion. Taking drugs and talking to quantum elves has reportedly led to scientific breakthroughs. I don't believe in quantum elves, but I fully accept that they are a valid source of conjecture, because everything is.

[00:03:30]  Peter: Quantum elves?

[00:03:32]  Bruce: Yes. If you're not familiar with quantum elves, look it up on Wikipedia.

[00:03:36]  Peter: Is that from DMT? You take DMT and you talk to them?

[00:03:39]  Bruce: Yes, you take DMT and you talk to quantum elves. We need to do a podcast where we take that or something. So, quantum elves is a known real phenomena, where you take DMT and you have this, I would say, hallucination of these quantum elves, and you have conversations with them, right? And they teach you about the universe or something.

[00:03:58]  Peter: Oh, I've talked to people who've done this, who really believe that they've talked to other... I don't know if I've heard quantum elves exactly, but they believe they've talked to aliens or other entities, or something, yeah.

[00:04:18]  Bruce: Yes. I listened to a scientist who didn't say specifically ‘quantum elves,’ but in his talk described going to Burning Man, taking drugs, and talking to an entity from another dimension who explained evolution to him. And then he took what he learned about evolution to the scientist he was working with, and they worked out together a revised version of evolution that did a better job of explaining evolution. And I thought, okay, this is total garbage. And I didn't recognize his name, but the name of the scientist he was working with was this world famous scientist. And I'm like, huh. I was looking up some of what he said, and this wasn't an impossibility: the two of them, probably as part of a team—so more than two scientists—were having discussions. And the discussions were partially sharing information with each other. And some of the conjectures that led to some of the later ideas had come from this talking to another entity that he had talked to while on drugs. And that may have in fact led to an actual scientific breakthrough, right?

[00:05:35]  Bruce: So this seems really bad until you realize that every source of conjecture is valid. There are no invalid sources of conjecture. And as long as you understand that a conjecture is different than the concept of severe testing, which is the critical side of this, then there's really no problem here. So how does this apply to Michael Levin? Okay, so Levin is clearly inspired by his religious background. And we were debating whether it was Buddhist or Eastern Orthodox. For our purposes, let's assume Buddhist, because I think I've heard him talk about Buddhism before, but I don't really know what his religious background is. But he doesn't hide that his theories were inspired by his religious background. What he does do by his own stated policy is restrict his public discussion to ideas that have survived testing. He avoids publicly discussing the spiritual untestable one-third of his worldview. That one-third comes from him, not me. Because those ideas currently lack scientific value.

[00:06:36]  Bruce: This is precisely the correct stance under critical rationalism. You are free to use your spiritual beliefs as sources of conjecture. But you should not introduce them into public scientific contexts, such as papers, unless the theory has been developed beyond ad hoc speculation and yields independently testable predictions that have survived real attempts at falsification. Praxeologists, take note. And I would add that his spiritual worldview, though I personally reject it, has proven extraordinarily fruitful as a generator of ideas that matured into testable theories. These theories have shaken the foundation of evolutionary biology and are actively rewriting our understanding of evolution. Many would likely be more comfortable if he were instead inspired by Popperian epistemology. In fact, much of his research program could have been naturally conjectured from Popperian principles. It's a genuine tragedy that Popperian epistemology was so badly derailed by Deutsch's Two Sources hypothesis, preventing it from generating something like Levin's research program.

[00:07:38]  Bruce: But Popper himself anticipated Levin. See way back to episode 38, Animal Learning and Popper's Epistemology. If you don't believe me on this, I go over in detail that Popper did in fact anticipate Levin. But ultimately, the source of conjecture doesn't matter. It simply doesn't. Levin's spiritual worldview is not mine and I don't accept it. Yet I respect that it is a wellspring of a remarkable body of evolutionary theory. So I see Levin as getting this exactly right. He allows his spiritual beliefs to inspire conjectures, but withholds them from the public scientific discourse until they mature into testable corroborated theories. That to me exemplifies the proper scientific and Popperian mindset. Peter recently challenged me on a podcast, and I can't recall which episode, about my strong resistance to panpsychism. I've said multiple times that panpsychists really rankle me wrong.

[00:08:35]  Bruce: Peter himself is an agnostic, but he thinks panpsychism contains a grain of truth. His idea is that reality as a whole bends towards wakefulness and increasing intelligence, ultimately towards humanity spreading through the universe and using it to create ever more intelligent beings. He points out that this trajectory could reasonably be called a form of panpsychism. Note, I think...

[00:08:59]  Peter: I do? You're talking about me?

[00:09:01]  Bruce: Yes. This is what I wrote in my tweet.

[00:09:04]  Peter: Oh, okay.

[00:09:06]  Bruce: Then I say, ‘Note: I actually think I got Peter wrong here.’ It was the next thing I wrote.

[00:09:10]  Peter: Okay. That's an interesting interpretation of what I might have said. I would not have used the word panpsychism, but the distinction between saying there's something baked into the universe—a blueprint that trends towards complexity and novelty or something like that—and panpsychism might be kind of a fuzzy distinction, honestly. So I don't know. Maybe I am a panpsychist after all. Anyway, I'm sorry.

[00:09:51]  Bruce: Okay. This connects directly to the issue you raised about the meaning of words. The “panpsychism” I reject is not the panpsychism Peter is asking me about. In fact, it's not panpsychism per se that bothers me. It's the profoundly bad arguments that so many panpsychists use and then present as if they are serious competitors to science. Philip Goff, in particular, since Peter and I recently read his book, exemplifies this problem. Under a proper epistemology, these are simply junk theories. The core issue is that panpsychism, like all terms, is a vague and fuzzy category that lumps together many different ideas, some true, some false. I'm not even sure I have much of a problem with Levin's version of panpsychism, because Levin admitted he was a panpsychist, sort of. Insofar as I can make sense of it, which admittedly isn't very far, it seems to me he simply means that intelligence is found nearly everywhere, possibly everywhere. But by “intelligence,” he means it in a very basic sense, systems separated by Markov blankets that attempt to maintain themselves by having knowledge about the world outside themselves. On this view, even a tornado might count as a system trying to preserve its structure and thus intelligent in that limited sense.

[00:11:16]  Bruce: Yes, this stretches the word intelligence past the breaking point for most people. But if Levin's right, and we could empirically show that a tornado actively maintains itself, if true, I'm honestly not sure what other English word we would have available to describe that phenomena other than intelligence. And this brings us back to the fundamental struggle of language. Finding the right words to express the concepts we actually mean is not a small problem, and it has no perfect solution. There will always be more concepts than words. So we are constantly forced to reuse old words with established meanings in novel and creative ways just to say what we're trying to say. You wrote: “ I guess we played those word games because on one side it's not possible to speak in a way to not be misunderstood and it's pointless to delve into definitions and on the other side, it's also pointless to accept everyone's personal definition to the point of not meaning much.” I think you absolutely nailed it there. Other than one small disagreement discussed below. You go on to say: “ if you choose to use a term, you take on the burden to clarify it or invent a better one, or else things get too fuzzy to discuss.”

[00:12:33]  Bruce: I profoundly agree. The moment you reuse an existing word, the burden is on you to clarify how you're using it. And yes, almost no one actually does this. That's a real problem. Here's where I may disagree with you slightly, though maybe not. Your paraphrasing seems to imply there might be some alternative to reusing existing terms, but there really isn't. We invent new concepts constantly, and we need language to express them. We invent new words, but that has its own major drawback. If you coin a completely new term, no one has any idea what it means. This makes it nearly useless for communication. I do try to invent new terms where I can. You've probably noticed this on my podcast. Meaning memes, the crit-rat loophole, the creationist fallacy. In those cases, I tried to build terms that more or less explain themselves. People can roughly guess what meaning meme means, even if they've never heard it before, and they haven't because I made it up.

[00:13:37]  Bruce: But this is hard. It takes real effort to coin usable terms. And even then, they're built out of existing words, which means misunderstandings are inevitable. Someone will always claim you're using a term wrong. It's frustrating. Most of the time, people simply reuse existing words and, just as Hofstadter suggests, extend them by analogy. That's basically unavoidable. This is why what I call mathematical realism, a term I invented, others call Platonism. But classical Platonism goes far beyond mathematical realism and includes many claims I reject. Still, I can easily see how someone like Levin may have a concept in mind, what I call mathematical realism, but lacks a word for it. So he searches around and notices that his idea is kind of sort of like Platonism, minus a bunch of stuff. He's not a philosopher. He probably never read Plato. He's just latching on to a core intuition that mathematical structures are real in the sense that they kick back at you. Deutsch's idea, in a nutshell.

[00:14:44]  Bruce: That kind of resembles Platonism. So he grabs the only label that he has and he calls it that. Of course, it isn't original Platonism. I agree he now has an intellectual duty to clarify this. I agree he mostly doesn't, except when directly asked, which I did ask him. And he did clarify at that point. I agree that this creates massive confusion, and I agree it gives ammunition to Sadia, who then says, “See, Levin says he's a Platonist. So Bruce, you're a Platonist too. I was right.” But I generally don't see an easy alternative. Either A, you invent new terms and hope they aren't incomprehensible or misleading, or B, you reuse old terms and hope people don't get confused. And they, of course, always get confused. So I think the real solution is cultural rather than linguistic. We all need to be more charitable. Clarify what you mean. Say something like, “I believe in Platonism, but here's what I mean by that in this context.” And don't immediately attack people for “misusing terms.” Instead, ask for clarification.

[00:15:57]  Bruce: Something like this: “You say you're a Platonist. Do you mean you believe in a perfect chair in a realm of perfect forms?” “Or do you just mean that math is real?” That, I think, is the only sane way forward. It is also why Popper was wrong. We should take the time to clarify and define our terms, or at least clarify our terms, not necessarily define our terms. That's just one possible way to do it. What we should not do is argue over definitions. Maybe argue, ‘That's a misleading use of that term, but okay, I'll accept it in this context.’ Just accept all definitions and move on. This brings me to a point that I think you'll likely agree with me on. Words accumulate meanings over time. Concepts slowly pile onto them layer by layer. And because of that, every word inevitably has what we might call conceptual leakage. If Levin calls his view Platonism, it is almost guaranteed that someone will assume he means he believes chairs have perfect forms in a realm of pure ideas, even if that has nothing whatsoever to do with what he actually meant. In fact, the leakage may even affect Levin himself. He may start with a concept equivalent to what I'm calling mathematical realism.

[00:17:19]  Bruce: But because he lacks a better term, he adopts Platonism. Over time, he might begin thinking, “Well, if Platonism is true, maybe there really is a perfect chair form out there.” And before he knows it, he accidentally is importing ideas he never originally believed, drifting from a narrow defensible view into a much wider and largely false one. We see this all the time in reverse too. Picking on Sadia again, she says, “Bruce, you're a Platonist because you believe math is real.” I respond, “I'm not a Platonist.” “Platonists believe in perfect chair forms, and I reject that.” She replies, “But you think math is real.” “So you are a Platonist.” Eventually I say, “Fine, call me whatever you want. I don't care about labels.” And then she conceptually leaks, and she says, “Ah, so you do believe in perfect chair forms. See, I've refuted you.” At this point, I just roll my eyes and say, “Never mind, you're not allowed to call me a Platonist anymore. I'm a mathematical realist.” Both sides leak conceptually.

[00:18:22]  Bruce: And this really is a serious problem. I don't see an easy solution here. My policy is simple. I take people at their word and I ask them to clarify. If someone tells me they're a Platonist, I don't assume anything. I ask, do you mean A, chairs have perfect forms in a realm of ideas or B, math is real in some objective sense? If they tell me, as Levin did, that they just mean math is real, then I assume from that point forward until proven otherwise, that by Platonism, they really mean mathematical realism. If later they start to talk about perfect chair forms, I revise my view and ask for clarification again. I see no way around this except constant clarification of others and ourselves. That's just the price we pay for having fewer words than concepts. And that was the end of my X post. So I thought that this was an interesting thing because it clarified.

[00:19:17]  Bruce: I've said lots of things like this on the podcast, but this response I was giving this person, I was putting it in a more organized way than I had before. So I thought it might be worth putting it on the podcast and putting it out there for criticism. But my basic view is, I buy the Popperian view that words are instruments that represent concepts, that we try to use to explain a concept, and that concepts are flexible, that they change and stretch by analogy all the time, therefore they can't be falsified. Theories are what make assertions about the world, and it's the theories that are testable. Since what we care about is truth, it's really theories that we have a preference for. This goes beyond what I intended to say, but where I've tended to disagree with Popper is, first, that concepts may well be sources of conjectures. I don't think Popper ever picked up on that, and I think that's just true.

[00:20:14]  Bruce: But I think it goes further than that. If I could maybe explain this in reference to our podcast about AGI and how Popper didn't believe in AGI. I do believe in AGI. I believe AGI can be done on a computer. Human beings do mean things. So there's this question of meaning. Philosophers have spent eons trying to debate what meaning actually is. Popper thought the whole thing was a waste of time and spoke very strongly against trying to worry about meaning and instead tried to go straight to theories with assertions that you could test. And he saw that as the solution to the problem. But humans do mean things. They do. You could take Popper's wording, his strong statements against meaning, you could take it as something like this. You're wasting your time at this point in time trying to work out what meaning is and have a theory of meaning. Because no one's ever made it work.

[00:21:20]  Bruce: No one's even gotten close. And it's going to mostly confuse you. So don't waste your time on it. Because the knowledge to be able to make sense of it doesn't exist yet. If this is what Popper meant, there's nothing wrong with it. I would completely agree with him. But I think you can also read Popper as saying something more like this. The question of meaning is itself a meaningless thing. There's no way that's true. AGI must be able to mean things, must be able to understand and mean things. So there must be a workable theory of meaning that we will eventually discover. And so the philosophers who spent all their time working on it, the problem wasn't that they were wrong to think a theory of meaning was worth pursuing. It's that they were pursuing it way too early, way, way, way before there was any way to do anything productive with it. And I suspect Popper would disagree with me on that. Since he didn't believe in AGI, I'm almost certain Popper would disagree with me on that.

[00:22:17]  Bruce: But if you believe in AGI, trying to understand a theory of meaning makes perfect sense. And one might even say that AGI will be a theory of meaning. It will be the long-lost theory of meaning that philosophy has wrecked itself upon for centuries. And it will be finally actually solving that problem, probably in some way nobody conceived of. Because of that, I can't go all the way with Popper, right? I know ultimately he must prove wrong on this. But he's proven right so far. All the theories of meaning that are out there in philosophy, they really aren't even worth the paper that they're written on. They are net negatives, right? And yet, this is why I will bother to try to stop and think about concepts, words, meaning, things like that, even though I'm a Popperian, even though I believe it has been worthless so far. Because I'm also someone who wants to help discover AGI. And ultimately, there must be something to it, even if we just don't even know where to go at this point.

[00:23:30]  Bruce: So this was some of my early thoughts on that. And I tie this a lot with Hofstadter's theories, which you might argue is a theory of meaning kind of sort of, right? Where he's trying to work out that we think in terms of analogies. It's such a basic theory at this point. I don't even mean basic. Basic might have a positive connotation. I don't even mean it in the positive way. It's so shallow of a theory with so little content at this point, that it's more like just an early conjecture. And that's why I get behind Hofstadter's theory in the sense that I find it interesting, and it's an interesting conjecture, and I don't know what else to use as a starting point. But I'm hesitant to ever endorse it as a theory because I don't actually believe it's a true theory at this point. It's not theory in the falsifiable scientific sense. So I'm somewhere in between all this.

[00:24:28]  Bruce: I also think that Popper had to constantly define terms and clarify his meaning because there was no way around this, even though he claimed you shouldn't have to do that, and that you should just accept that. He always made the argument that there's no point in defining all your terms because ultimately you have to have undefined terms, otherwise you'd have an infinite regress. And yet Popper spent tons of time defining terms, right? There's just no way around it. Even someone who has a hostility towards it still ultimately has to do it. I feel like Popper had part of the truth here—but I don't want to say that his opponents had part of the truth. They didn't. They were on a wrong path. Popper had half the truth and his half was actually correct. And his opponents didn't even have half the truth. But ultimately it will come back around and there will be a theory of meaning. And it will be part of our theory of general intelligence.

[00:25:23]  Bruce: And I don't think we should necessarily just dismiss meaning and concepts and words to the degree that Popper did, ultimately. Even if he has inductively proven right up to this point that going after a theory of meaning as a philosopher is going to wreck you. Like it's going to be a net negative. And I think it will probably continue to be a net negative for a while until we discover AGI. So that's my thoughts on this subject.

[00:25:49]  Peter: Okay. Well, I love that we never run out of things to talk about. I just can't see it ever happening. And I guess we'll just keep talking till we're on the Dyson Sphere and beyond.

[00:26:05]  Bruce: Because we're not going to die?

[00:26:07]  Peter: Because I'll see you in the Omega Point, Bruce, and we can keep having the podcast.

[00:26:14]  Bruce: The podcast that never ends, the podcast that goes into infinity. That's our new tagline, The Theory of Anything, the podcast that goes into infinity.

[00:26:23]  Peter: Yes. Okay. Thank you, Bruce.

[00:26:27]  Bruce: All right. Thanks.

[00:26:34]  Peter: Hello again. 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 bnielson01. Also, please consider joining the Facebook group, The Many Worlds of David Deutsch, where Bruce and I first started connecting. Thank you.