Watch Episode Here
Listen to Episode Here
Show Notes
This is a ~1 hour 5 minute discussion with Amahury Jafet Lopez Diaz ( Carlos Gershenson-Garcia ( and Richard Watson ( Stay tuned for part 2 where Amahury will present his ideas in more detail. Amahury's paper:
CHAPTERS:
(00:00) Common ground and stakes
(06:10) Ontology of platonic space
(14:50) Testing separate realms
(25:52) Acid tests and metaphors
(34:55) Interaction and narratives
(44:20) Formal facts alternative
(52:38) Experiments and surprise
(01:00:29) Closing reframing proposal
PRODUCED BY:
SOCIAL LINKS:
Podcast Website: https://thoughtforms-life.aipodcast.ing
YouTube: https://www.youtube.com/channel/UC3pVafx6EZqXVI2V_Efu2uw
Apple Podcasts: https://podcasts.apple.com/us/podcast/thoughtforms-life/id1805908099
Spotify: https://open.spotify.com/show/7JCmtoeH53neYyZeOZ6ym5
Twitter: https://x.com/drmichaellevin
Blog: https://thoughtforms.life
The Levin Lab: https://drmichaellevin.org
Transcript
This transcript is automatically generated; we strive for accuracy, but errors in wording or speaker identification may occur. Please verify key details when needed.
Main Episode
[00:00] Amahury Jafet Lopez Diaz: OK, so let's start by just facing where we are currently in the exchange, and potentially I will present something later regarding the concept of symbol, because the concept of symbol in biosemiotics is kind of different to the conception that we have of symbol, for example, in computation. So maybe that would be necessary to present later. But for now, I think that you and I, Mike, are already close on four substantive points. First, both of us think that the mainstream gene-plus-environment story is too thin for explaining, for example, morphology and cognition. We both agree on that. Second, both of us think interfaces matter, and that there is a real research program in mapping what kinds of patterns, policies, and organizations become reachable through different interfaces. Third, both of us reject naive billiard ball causation as the only meaningful kind of causation. And fourth, both of us want a framework that is both experimentally fruitful in biology, but also a metaphysics armchair, you know, like good, as I was telling you to reach out at the beginning, good philosophy and good biology should come together. So all these four points that I mentioned are visible in our exchange so far. And I think then that the disagreement that we have is very narrow. Basically, and if not, correct me, Mike, please. As I understand it, Mike wants to live open and sometimes positively lean toward the existence of a structured non-physical space that does explanatory work beyond the in-world constraint architectures I propose. And what I propose, basically, is I want to keep the ontology monist, no ontic dualisms anymore, and treat the extra descriptive power as arising from constraints, closure, model-relative abstraction, but everything within one world. So, to be more precise, I'm not saying that MAD is fake or that the structure isn't real in our universe. In fact, I think that there is a lot of structure in our universe that's real, and maybe later we will discuss this further, but the structure of the space does not automatically imply a second ontological realm. So I will stop there for now. Let me know if there is an additional point from these four that maybe I mentioned and you disagree with, or maybe we have even a fifth point that I didn't mention so far. What do you think, Mike? Is my conception of the current state correct or no?
[02:51] Michael Levin: I think that's very fair. I think we agree on those four points. I think that maybe it just comes down to, at least part of this will just come down to what does it mean to have a separate ontological realm? Because in the end, maybe we're not separate and this is just terminology that maybe it doesn't matter. So I'm not sure what would really help. I mean, we can talk about anything you want, but what would really help me is to see, given that we agree on those four things, is to see your framework and how it works and what kinds of research program, for example, it suggests. Because, as you said, it would be nice to collaborate. I mean, I'm happy to collaborate whether or not we agree on some of these fine points. We can still collaborate. So that would be fine. But I would love to know because mine makes very strange and specific predictions that we are currently testing. And I would love to hear how your framework works and what it suggests.
[04:00] Amahury Jafet Lopez Diaz: Okay, anyone else has. Like additional comments.
[04:03] Richard Watson: I'm pleased to report that after five minutes, I'm still following. Those points of agreement made sense to me. The difference between a structured non-physical space and a modest within one realm thing, it's like the first question on the list that I've heard real philosophers ask about these things is what's at stake, right? So as Mike says, what does that distinction mean? And I think it would help me to know what's at stake if you were to come down on one side of that fence or the other.
[04:39] Carlos Gershenson-Garcia: Yeah.
[04:39] Amahury Jafet Lopez Diaz: Yes.
[04:42] Carlos Gershenson-Garcia: I mean, I would say that it can be classified as a pragmatic pluralist, meaning that it's as opposed to, you could call it platonism, where people assume that there's just one truth through description of the world, and then everything else is mistaken. So I would say that there are many different narratives to describe phenomena, and then the way to judge them is not truth, but usefulness. And then this usefulness can be different in different contexts. So depending on which question we're asking, some narrative might be more useful than others. And this applies not only to science, but also to philosophy and to literature and everything. So for me, the question is not so much whether there are ideas independent of observers and then we just discover them or we create them or whether mathematics are invented or discovered, but what's more useful to understand phenomena? And apart from that, I have my own narratives that I prefer to use. And then we can also discuss why I prefer those narratives. But anyway.
[06:10] Amahury Jafet Lopez Diaz: OK, so I think that maybe the best way to guide the discussion is that Mike answer this question because maybe even if I understand what he means, these different conceptions, different paradigms that we are throwing into our discussion will make sense. So Mike, what exactly is the ontological status of the space that you propose in your work? Is this a metaphor? Is this a model or a mind-independent domain? When you say there are things from this another platonic space influencing the physical space, what do you have in mind, a metaphor, model, or literally a mind-independent realm? What do you have in mind when you describe this platonic space?
[07:06] Michael Levin: Yeah, well, a few things. So, first of all, I'm not a philosopher, right? And I'm not exactly sure what's at stake about ontological, you know, different ontological status. I look at it from two perspectives. One is as an engineer. Ontologically real to me as an engineer means anything that I either have to worry about or I can take advantage of. If this is something I can interact with in a way that is going to matter, it's going to have consequences, either to predict and explain things that already exist or, I think even more interestingly, to guide research agendas going forward, because I think we have a lot of emphasis on prediction and explanation of things that somebody already made, either evolution or another engineer or whatever. But I think more interesting is, given this view, what do you do next? What research program does it allow you to do that perhaps other viewpoints don't allow you to do? So whatever impacts all of that, I think that's ontologically real. That's kind of my definition. And then the mind independent. So I don't think anything is mind independent. I think, in my view, there's so much more mind everywhere you look at, you'll probably find some kind of mind. And so that's a whole other thing that we can talk about. But I think really the minimal version of what I think is happening, just to make it very minimal, and then maybe many ontological frames can be compatible, is two things. One is that I think math, what we call math, is a tiny tip of the iceberg. In other words, I call it Platonism, even though I'm not going along with a lot of traditional Platonist thought, only because it's an anchor point, so the mathematicians at least know where I'm going. But I think it's a tiny tip of the iceberg. And I think that set of patterns contains many things, some things that mathematicians recognize, some things that behavior scientists will recognize, some, you know, AKA kinds of minds. So it's very much that kind of expansive thing. So it's not all about math. I think it's actually much, much broader. Math is just an example of that people already sort of have grappled with, but I think it's much, much broader than that. And then the other thing I think on a very practical level is I think if we can, if we go beyond the idea of constraints, then, and really take seriously the flip side, which is enablements, like what do these patterns offer? So I think, and this is the testable part, and we're testing all this. We have some cool data that I'll share with you in a few weeks when all the manuscripts are ready. But this question of the accounting of what do you get that you didn't expect. So I put in some effort in this world, effort that we can quantify via energy, via computation, and we have accounting that we use in these, the Turing paradigm and the conservation of mass, all these kinds of things. So we have these things, we have some idea of how much effort you have to put in to make certain things happen. And I think what's happening is that the methods that we use to account for all of that are missing a huge part of what's going on. And the part that they miss, evolution didn't skip it, didn't miss it. We're missing it. We're not doing the accounting right. And so what we're, but evolution exploits it and then life exploits it and so on. And it doesn't even have to be life. I think computations exploit it too. But the point is to really understand what we're getting from that, what I call that space. And I think it's definitely more than static patterns, like the value of E and things like that. But I don't know exactly how far it goes. I think there are policies. I think there are sort of cognitive propensities, maybe compute, maybe actual compute that you can do in that space. I don't know. Yeah, we don't know that yet. And so that's it. And if we agree that studying those things is important, and somebody wants to fold them in and say, look, I'm studying the universe with a big U. It includes the conventional accounting. It includes this new stuff, it all under the same umbrella. I'm okay with that, I guess. You know, I'm totally fine with that. But I do think the tools that we have for what we've been doing in physics and computation are leaving a lot on the table. So that's...
[11:52] Amahury Jafet Lopez Diaz: Very cool, very good. Thank you so much. Because as I understand your latest formulation on these Platonic space ideas, and I'm saying it because I usually read your blog where you also put a lot of these ideas, you suggest that a separate realm is whatever, A, cannot be investigated with the same tools as physics, and B, contains features that cannot be changed by varying physical constants. But A and B are not sufficient conditions for ontological distinctiveness. They show at most that methodological directedness and model invariance matter, but neither of those by itself establishes a second ontology. As Carlos suggested or implied earlier, different modes of access do not imply different kinds of being. Likewise, the fact that mathematical truths are invariant under changes in physical constants shows that they are not ordinary contingent physics facts. We can discuss this further, but it does not yet show that they form an independently interacting realm. This is exactly where the debate in contemporary philosophy of mathematics lives, basically how to move it at all from explanatory indispensability and objectivity to ontology. And here I would like to point out that this best explanation criterion that sometimes you highlight is not enough by itself. Sometimes I feel that when you make claims about, for example, this ontological world, let's call it Platonic space, you are kind of like mixing explanatory indispensability. It is true that in math, that in physics, sorry, we use a lot of math and we can see that those mathematical facts in some way are kind of like influencing physics. But at the same time, what you, for example, mentioned Platonism to a philosopher, then automatically, maybe as me, then a second realm comes to our minds. And there is also stuff about causal efficacy. So those are not the same: explanatory indispensability on ontology and causal efficacy are not the same. And the risk here is that maybe we are conflating explanatory indispensability with causal efficacy. And that's maybe what's causing this confusion. The fact that mathematics appears in our best explanations does not straightforwardly entail mathematical objects causally determine physics. This is exactly what the philosophy literature in mathematics contests. So again, I will make a pause. I'm not sure if everyone else, for example, Carlos or Richard or you, might want to say anything.
[14:50] Michael Levin: I mean, the only thing that I will say is, I understand causation is very deep, these extremely deep waters. And so Lauren Ross and I are writing something on this, and she will, I’m sure, catch all the things, places where I’ve gone wrong. But what to me would be really helpful is when you say those things are not sufficient to establish an ontologically separate realm, could you talk a little bit about what would be, like, let’s have a positive control? What would have to be the case for us all to say, yep, we agree, separate realm? What are we missing here? What do you need to make that happen? Because it’s still unclear to me what else is missing here.
[15:32] Amahury Jafet Lopez Diaz: Sure. Basically, I would count something as a separate ontological realm only if it had, one, independently individuated entities or states; two, a principle law of interaction with the physical domain. That's very important because that solves many of the issues that philosophers claim about Platonism. And three, an intervention profile such that, holding ordinary physical front-end conditions fixed, variation in the putative other realm produces systematic differences in physical outcomes. Mathematical truths as such do not meet that standard simply by being studied differently or by remaining invariant in their physical tweaks. So that will be my positive control.
[16:18] Richard Watson: Could you just say those three again, but a little bit more slowly, please?
[16:21] Amahury Jafet Lopez Diaz: So again, I would count something as a separate ontological realm if it had independently individuated entities or states.
[16:30] Michael Levin: Sorry, what does that mean?
[16:34] Amahury Jafet Lopez Diaz: Basically, you have two substances. I mean, of course, since I'm conceiving the universe as monist, I would say that everything is matter, energy, and consequences. But if you have a second, how do you say, kind of substance, let's say in terms of the card, like a soul and body, or mind and body, those are different individual identities. And that's what I mean for the first part.
[17:02] Michael Levin: But isn't that, I'm sorry, I don't mean to bog down on the first one, but isn't that sort of the whole thing we're arguing about? When you say they're separate, are they separate? Because, for example, for the second one, when you say they interact, to me, it sounds like the minute you say they interact, that means you see consequences on the physical world. Somebody's going to say, oh, no, look, it's clearly touching the physics, so it's got to be part of this world. I think that's the whole thing we're arguing about, is what would you have to see to say, Oh yeah, man, that's definitely a separate realm.
[17:36] Amahury Jafet Lopez Diaz: OK, so just let me finish the three, and I can go back to what you just mentioned. So again, the first point, Richard, is we will have a separate ontological realm if we have independently individuated entities or states; we will go back further to what this means, as Mike suggests. Second is a principle law of interaction with the physical domain. And the third point is an intervention profile such that, holding ordinary physical front-end conditions fixed, variation in the putative other realm, like this other substance, whatever that means, produces systematic differences in physical outcomes. I think that this is something similar to maybe what Mike is doing, but again, mathematical truths as such do not meet that standard, simply because they are studied differently or because they remain beyond their physical tweaks. And I would like to say that this debate that we are putting on the table, Mike, going back to your concern, is not new. This is a long discussion traced back to Plato and Aristotle. Plato was an idealist focusing on a transcendent world of forms, and Aristotle was an empiricist who focused on the physical world. So Plato prioritized abstract reason, whereas Aristotle relied on observation and classification to understand the world. And again, if you take a look at the Stanford Encyclopedia of Philosophy, there are different entries there. It's an amazing encyclopedia. One of them is about mathematical Platonism. And you find out that Platonism is the conjunction of existence, abstractness, and independence, and so on, but immediately you will find a lot of concerns or arguments against metaphysical Platonism. And likewise, in the same encyclopedia, you can find an entry on abstract objects, which are usually treated as causally inert. However, as I understand, Plato's forms were not fully inert. And again, I think that the explicit difference that this type of entries in the encyclopedia do is that this debate should be split in two questions. The first one is whether mathematics explains in empirical science and whether it explains within mathematics. Those are related, but not the same thing as proving Platonic ontology.
[20:15] Carlos Gershenson-Garcia: I want.
[20:16] Michael Levin: I'm sorry, go ahead, Carlos, you go.
[20:19] Carlos Gershenson-Garcia: Yeah, I just would like to quote Erwin Schrödinger on this topic. So I'll read: "The world is given to me only once, not one existing and one perceived. Subject and object are only one. The barrier between them cannot be said to have broken down as a result of recent experience in the physical sciences, or this barrier does not exist." So, I mean, he was strongly influenced by Eastern philosophies. And there's this idea that dualism is a misconception, let's say, can be useful in some cases. But the fact that we cannot speak about an ontology without language, which is subjective, kind of suggests that for practical purposes we can distinguish between subject and object. But then when we try to examine what would be an object without a subject or a subject without an object, then they're mutually dependent. So this is inseparability. But maybe we are kind of getting into discussions that are not so relevant if we want to understand phenomena. Because, I mean, I've been starting a program to systematize the limits of science in general. And many of these are related to limits of causality in the sense that, I mean, sensation times, people speak about causality, and there are many different formalizations about it. And, I mean, in medicine, it's an essential concept, but we don't have a good theory or maybe comparing it with the good definition of life. We don't have a good definition of causality. We don't have a good definition of complexity. We don't have a good definition of intelligence, consciousness, and so on.
[23:03] Carlos Gershenson-Garcia: But let's say our tools that we use to understand the physical world, that have been very useful for physics, seem to be limited for biology, for economics, and in many cases just for complexity. So one of these is that when you have interactions, then these interactions can generate new information, and this might imply that your state spaces grow. And we are still kind of struggling with tools that do what we have with fixed state spaces, but for dynamic state spaces so that they change. Another limitation is that just for combinatorics, we face immense state spaces, meaning that there's not enough time in the universe to start exploring them statistically. So then the probability that we will find novelty in those huge state spaces, even when in theory they might be finite, it's non-negligible. I mean, and then there are other issues, like when you have overlaps in datasets, then that limits your classifiability. It's more related to AI and statistics. Yeah. But let's say in many cases, I mean, at the end, state spaces are descriptions and abstractions we make of phenomena. And in some cases, they're very useful. But in some cases, there are details that were not important at some point and then they become important. So then we need to include them. And then those changes in our models or in our description of phenomena limit our prediction because, of course, we don't know what effect they will have until it happens. And it seems that the solution is not so much that, oh, we just need better tools in order to be able to predict this phenomena. It seems to me that inherently their predictability is limited. So then what we need to do is to accept that prediction and maybe to be able to characterize it better, to say like, okay, we can predict this amount and that amount we want. And then to explore alternatives to understand phenomena, even if this understanding is partial, and which will be better than assuming that we are having a complete prediction and then just bumping against the walls?
[25:52] Michael Levin: Amar, do you think that this issue of whether or not there's another realm, do you think that's a logical, philosophical issue or is it an empirical issue? What I mean is, I get the feeling that it's almost impossible that, I mean, you can tell me, do you think it's impossible for there to be a second realm? Like the whole thing is, it's a contradiction and it just can't be done? Or do you think it could be done, but it could be true, but it just happens not to be true? Is it a?
[26:22] Amahury Jafet Lopez Diaz: Yeah, it makes sense. I mean, I think the strongest answer to a challenge in order to say whether it's possible or not is to ask what would be the properties of the second realm. So if platonic space is more than a metaphor, I would ask first, what is its origin? Is it eternal? Or may it just appear at the same time as the universe; it's kind of like a parallel reality that exists after this singularity that we call the Big Bang, whatever. Second, what are its elements? Are they, as Plato proposed, like these eternal, non-causal, non-spatial, temporal entities? What structure does it have? I think that the topology of the space would be super important because in case the platonic space exists, then of course we want to investigate it. And depending on the topology, of course, it's not the same walking, for example, a torus than walking spheres. So, of course, that would be relevant in order to create an experimental program that explores this space. And the most important part, how does coupling between physical and non-physical work, and what intervention or prediction would distinguish this platonic space from any other type of account? So I think that in our discussion so far, Mike, you have said that you think that this space is causal, but not in the naive realism, causation sense. And then you treat the ontology as a research program, that this is still to be specified. I think you're still also developing these ideas. That's why we are having this chat now, right? Because you are developing these ideas and I'm also interested in them. That is philosophically understandable, but it means the ontology right now is still underspecified and it cannot do the explanatory work that sometimes maybe you suggest when you talk about it. So, for example, after reading one of your most recent posts, I think you published this maybe on March 31 or something, I could notice that now you argue that these platonic patterns may be modified by their projections into the physical world, which means that the interaction is kind of like two-way. You also say there that the time and pre-existence of these patterns may be known to be well-posed, and that memories might be partly constructed from that space. But here comes then an important question: if the non-physical patterns are altered by physical embodiment, then what individuates them? What preserves their identity? And what is the interaction law? And of course, at that point, aren't you already closer to something similar to process ontology, relational, organizational accounts than to classical Platonism? Again, I'm not saying that it's maybe impossible to have some kind of like a second realm. But these questions that I mentioned are, how to say, kind of like an acid test, you know, in order to explain this platonic space: what are its elements, what is its origin, what is the metric topology, if any? Is it causal and not causal? How does it constrain the world? I think that if you build an experimental program that reaches those questions, then more people can be convinced about the existence of a second realm. But for now, as I see it, it's more, how to say, like a nice or interesting metaphor that still is under development, but it cannot do the lift, the explanatory lift, when, for example, we're having this type of discussions of whether there is a second realm or not.
[30:26] Michael Levin: Okay, great. I agree with pretty much all of that. So let me say a couple of things. First of all, about metaphors. So I'm very suspicious of the questions about whether something is a metaphor or do you really mean it? I think in science, everything is a metaphor. Pathways, gene networks, like all of it. I think all we have are metaphors. So in that sense, yeah, it's a metaphor, but I think it's like the other scientific metaphors, meaning we can take it very seriously in specific domains. That's pretty much all I think we can do. And I think that sounds consistent with what Carlos was saying at the beginning. So I don't see that binary distinction. The other thing is, so you raised a bunch of good questions about the space. Where did it come from? What is the structure? What are the, you know, is it sparse? Is it dense? What, you know, topology, like all of that stuff. Those are all great questions. Definitely on my list of research for them. So I have, there's a few talks I've given where I've given a slide. Those things are definitely there. I'm not making any strong claims about it yet because A, I don't have data that constrain that. I have no idea where it came from. I don't know if it's sparse or dense. Like these are things I don't know yet. Definitely things we're going to have to figure out. I think that's a pretty normal thing in science, is just to be able to say something without having to say everything. Like, you know, I think it would be worse to start making claims about it that I didn't have any particular evidence for. But I think the important part of the research program that is very actionable right now is the first part that I talked about, which is just to quantify in some meaningful way the fact that it provides something for us experimentally that is not accounted for using the standard methods that we use in this realm, so to speak. And just to even try to nail down, I mean, we've got to go step by step, just to nail down that we are getting something from this that is surprising, meaning that we weren't ready with an accounting for it, that's A, and B, behaves quite differently than anything we've seen before, and sort of as a corollary to that needs us to revise how we do our accounting. So, that's that. The third thing is, yeah, this is, like, on the first page of my preprint on this, I say this is not what I mean is Plato's original theory, right? So 100%, I know I'm stepping away from that. I'm fine with being closer to these perceptual kinds of views. I do borrow a bunch of ideas from that. I think, yes, I do. I suspect, and again, this is just a suspicion on my part because I don't have evidence on that, I doubt that these things are eternal and unchanging. Some of them may be things, you know, things like the value of E, you know, maybe that thing doesn't go anywhere. But the ones that we're interested in, which are the ones, the patterns that I think we see as kinds of minds, for example, or things like that, I suspect that those things are changed by their interaction with the physical world. And so I guess the final thing I would say is about this business of interaction. So I'm still hung up on whether this is a logical problem or an empirical problem, because it feels to me like any kind of interaction, like we know it can't be billiard ball interaction because then for sure you're still in the physical world. We know it can't be that. So it's going to be some kind of weird thing that doesn't look like causation, like naive causation. And then it's very easy to just say, then that's not causation and that it's not causally potent. I feel like there's a logic problem here because, again, without, I think maybe what we should do is sort of like build a fantasy world and specify all those things where we can all agree, yeah, if this were true, we would just say it's a separate realm. Because it seems to me that, like, I think, when I do that in my head, I can't do better than how I think math and physics. I think that is an amazing example of all the things that you've just said. I don't know how we could get any better than that. And so I think the way it works actually does meet all of your criteria. But if it doesn't, I start to worry that nothing could meet those criteria. Because I think it just wouldn't look like the kinds of thing we're used to. And then you can just say, well, that's not what we're talking about. So I think maybe a little world building would be really useful to have a positive control for this case.
[34:55] Richard Watson: My worry is a little bit different, Michael, almost the opposite. It's not that nothing could meet those criteria, but that if it did meet those criteria, it wouldn't be a different realm.
[35:06] Michael Levin: Yeah, that's another way of putting it. Yeah.
[35:09] Carlos Gershenson-Garcia: I mean, playing devil's advocate, I wonder whether we could arbitrarily create as many parallel realms as we wanted, and then it would be just like translating that other realm into our realm, just like we could invent a language and then we would translate from that language into English.
[35:34] Michael Levin: But does that mean, do you think that it's, because what I heard from Amma makes me think we can't have any alternate realms. Are you saying that you think under those conditions it would be very easy to come up with a scenario where we would all agree that it's a separate realm?
[35:48] Carlos Gershenson-Garcia: So it's a question of narratives, in the sense that we can say, oh, well, some people might think that in order for it to be a different realm, let's say Cartesian dualism, then you cannot link both realms causally. Let's say whatever happens in the world of ideas, there is some relation, but then there is some dynamic and then you cannot link it. But then if it's linked, then it might be just a change of description, like, okay, we can describe, let's say, atoms in terms of matter or information. And these are different narratives, but then we could also color these are different worlds. But at the end, they refer to the same phenomena at different levels and with different generalities, in the sense that in a similar way, you can make a translation from natural language into formal logic. And of course, it's not one-to-one translation or from physics into math. And again, it's not one-to-one translation. And then it's a matter of agreeing on whether we call this or these are different or these are part of the same thing.
[37:21] Richard Watson: Yeah.
[37:22] Carlos Gershenson-Garcia: So because at the end it's also like, okay, what do we mean by reality and what do we mean by world? Whether to consider something real or not, or whether we consider part of the world or not in the world. I mean, I think at the beginning we were speaking about universe with capital U. So I guess we can assume that we include everything in there. So we're not speaking about something outside, but let's say, from all the possible things that can happen, which seem to be infinite, how do we partition what we can perceive and what we can describe? And then this, to me, seems to be, that can be as arbitrary as saying, okay, we will do this partition and then we will have these two worlds. But then, of course, being pragmatic, most of those partitions will not be very useful. And then the question is like, okay, does this narrative where we separate in this way these phenomena from these descriptions or these properties from this substrate, does it make sense? Does it give us better insights, some explanatory power, some predictive power? And then that part, the pragmatic aspect, kind of reduces all the discussions about, oh, my description has these colors, and then your description has these ribbons. And it's, I mean, it can be relevant for other contexts, but let's say for engineering, then it's like, okay, which one allows us to control phenomena more efficiently?
[39:20] Michael Levin: I think the interactionism is the key here because again, it feels like it's a logical issue. If I say, okay, they do interact in a causally important way, then one can say, well, then obviously it's part of the same realm because they interact. And if you say they don't interact, then you say, then we don't need to worry about them because they don't interact. So it seems like the argument is that it's logically impossible for this to be the case, right? For there to be another realm that matters. That's what I'm trying to figure out.
[39:59] Carlos Gershenson-Garcia: Is there a definitional impossible? If there are no interactions, then you cannot say anything about them. The moment you can put a label on them, that's already an interaction. So then, in that sense, they are part of our universe with capital U. But still we could say, okay, within that big universe, let's divide into these worlds, like matter and information. And then let's see how we can go from one to the other. And in some cases, it's useful just to stay at the level of matter, in some, just at the level of information in some others, to consider both or see how one affects the other. And the same if we want to speak about life, maybe let's say we want to separate the living from the non-living in some contexts, even if it's not a sharp distinction. And in some others, we don't need to make that distinction.
[41:08] Richard Watson: For you, Mike, is there a kind of interaction that allows them to interact, but still be separate?
[41:21] Michael Levin: So here's what I've said before about that interaction, and I'm a says it's not, it doesn't do the job, but here's what I've got so far. What I want about this interaction, we've already given up the idea that it's going to be a standard sort of interaction where you can do the physics accounting on both ends. So it's not going to be the typical billiard ball thing where the physics tells the whole story. So we already know that's out the window. So what is left? So I like two things about causation. One is, they're interacting if, in order to have an insight, in order to have insight, in order to have insight into what's going on. And insight, I mean, explain, control, and more than, and predict, but more than that, meaning that I'm enriched by the insight in the sense that I will now go do more interesting things. There's a future, it's not just past sort of focus, it's future focus. I have an insightful understanding of what's going on when it leads to new research programs, new better questions, new discoveries, that kind of forward-looking. So there is an interaction and a causation when the thing that, whether it's in the math world or the behavioral layers or whatever, it is indispensable for you to have insight into what's going on. There is no reduction that you're going to do that leaves it out, that still provides the same level of insight. So A causes B in an explanatory, so I think in an explanatory sense, this is critical. It's a good explanation for why something in the physical world happened if you really have to understand what the other side of it is. Okay, so that to me is a perfectly good example of causation. You can say, and I mean, where people don't like it is they say, well, that's, you know, kind of too observer relative. It's not objective causation. And I don't think there is any sort of view from nowhere. I think everything is observer relative. So I'm okay with that. And then the other thing is the kind of the more of the Judea-Pearl kind of thing, the counterfactuals. So you've got, so A causes B if I can tweak, if I could tweak the properties of A, then B would be different, right? There's a dependency here. So the example is like if the distribution of primes were different, then the cicadas would be coming out at different times. The leptons would be doing different things and so on, right? So that, so those, that's what I see. I think we have to already throw out the window the idea that it's going to be conventional causation. And then what you're left with is counterfactual kind of stuff and explanatory. And I think both of those are perfectly good causation.
[44:20] Amahury Jafet Lopez Diaz: Yeah, I think that at some point in your exchange, you also mentioned that, okay, if we have mathematical facts, of course, by construction, those are not physics facts, right? And if they are not random, then they belong to some kind of structured non-physical space. And I think that's kind of a false dilemma because the alternatives are not random or another realm. I, in fact, advocate for a third option, which is these are formal facts, stable consequences internal to systems of definitions, inferential rules, and structural relations. They're objective in the sense that they are non-arbitrary once the formal setup is fixed, but that does not yet make them causally efficacious residents of a second substance realm. For example, the four-color theorem states that any planar map can be colored with at most four colors, and it's validated with a standard Zermelo-Fraenkel axiom of choice set theory. And this means that the theorem remains valid on planes and spheres, but not on other surfaces, for example, like tori. Something similar applies to, for example, the value of pi. In non-Euclidean geometries, its value changes. Basically, if you change the distance, the value will change dramatically. I will give you an example. There is something called taxicab geometry, where the distance is measured like a taxi navigating a grid of streets, like Manhattan. And the unit circle under that metric is actually a square. And in this metric, the circumference is 8, the diameter is 2, and the definition of pi, which is the circumference over the diameter, turns out to be four. So it's not 3.14, 15, et cetera. So this applies basically to any other theorem or mathematical constant, with the exception, of course, of some results in logic and metamathematics, for example, incompleteness theorems by Gödel, because those theorems, of course, are universal across all axiomatic systems. But my point here is that, yeah, of course, they have some kind of value in the physics that we use. But that doesn't mean their indispensability is that they are efficient causes. That's what I'm saying. So of course, you can ask, okay, but then are those mathematical facts inside the world? And I wouldn't say they are inside the world as localized physical posits, like we both agree on that, but neither are they outside as a second substance. Again, I think they are part of the structured relation between formal systems, measurement practices, and invariant organizations that physical systems instantiate. And this is exactly where philosophical approaches such as metaphysical pragmatism and perspective of realism help. I think I mentioned this paper to you, The Variable Thinness of Being: A Pragmatics Metaphysics of Affordances, a paper by Sandra Mitchell, where she argues that what we are justified in treating as real arises from an interaction between representative models, causal experience, experiments, and conceptual frameworks, rather than from purely top-down readout of abstract structure or purely bottom-up ontology of entities. So this is, again, some kind of middle path between reductionist and ontic dualism. So similarly, there is another paper by Marcus Beckstrom called Natural Law, Modeling Relation and Two Roots of Perspectivism. In this paper, he makes a similar move through Rohsen's epistemology. Indeed, there is a structure in the universe.
[48:28] Amahury Jafet Lopez Diaz: Like, for example, the fact that the mass of an electron, the charge of an electron, have constant values implies that there is a structure in the universe. But scientific knowledge is perspectival because system-environment carving and complexity matter, and yet we live in a single realm. So I think this type of alternative philosophical narratives can be helpful in order to ground the ideas that you are developing. I also share with you a book by Paulo Mancosu called Philosophy of Mathematical Practice, where he asks that the obsessive focus on access to abstract objects has narrowed philosophy of mathematics too much, and that mathematical practice, explanation, and representation matter alongside ontology. Another book by Otavio Bueno and Stephen French, Applying Mathematics: Immanence, Inference, and Interpretation. They explicitly argue that even if mathematics is indispensable to science, that does not by itself justify commitment to mathematical entities, and they also deny that the relevant mathematics itself thereby acquires explanatory primacy. So, again, I think that there is a full landscape of different alternatives from philosophy, particularly the philosophy of mathematics, that walk in the middle of purely reductionist and an antic dualism and that can be informative in order to ground your experiment. So, for example, I will give you very quickly my alternative framing, like what is my positive position here. I wouldn't say this flat reductionism. I'm not a reductionist. You can ask Carlos about it. But my perspective would look like, first of all, you have a monistic ontology, you know, like a single realm. You have multiple causal modes and descriptive roles. This was something that Rosen and Pattee, in their work. Rosen, for example, complained that, of course, you cannot explain living systems using just efficient causation, which is like this push billiard balls that, of course, are not sufficient in order to describe what is like. But he encouraged then the segregation of causal roles that works fine for many physical systems, but fails to capture the organizational entailments characteristic of living systems. That's why Rosen adopted these Aristotelian causes. And that's why he developed it there, because he agreed, and Pattee too, that it's need, we need different modes of causation and different descriptive roles. So we have monistic ontology with multiple causal descriptive roles, having symbol-matter complementarity as epistemology rather than substance dualisms. And I would propose organizational closure, like this ability of living systems to regenerate themselves, to sustain in a precarious way, as the main principle behind stable, heritable, and functionally efficacious constraints. So again, these ideas that I'm bringing to the table are not new. There is another nice paper by Michael Turvey and Robert Shaw, Toward Ecological Physics and a Physical Psychology, written 30 years ago. And they argue that this asymmetry of dualisms should be replaced by a duplexity grounded in organism-environment mutuality and reciprocity, not by these two incommensurable substances. And in my opinion, the first one who brought all these ideas since 1969 was Howard Pattee himself, who introduced the concept of symbol-matter complementarity, the relationship between rate-independent symbolic controls and rate-dependent physical laws. And according to his epistemology, living systems cannot be fully described by physical laws alone, but require a complementary, irreducible description of symbolic, informational, or hietical control structures. And of course, that sounds very similar to what you want to achieve, Mike. But I firmly believe that if we combine participistemology with your experimental prowess, these definitely can establish a new paradigm in biology, which is better grounded, philosophically speaking.
[52:38] Michael Levin: Yeah, so, okay, on the one hand, I'm game to try. So if you want to collaborate and do some things from that perspective, let's try it and see what happens. I think what I see happening here is that, I mean, look, fundamentally, I mean, I like monism as much as anybody. I think ultimately, right, if we had the final answer, it's probably one thing, and maybe it's consciousness and whatever like that, that's great. But in the meantime, I don't know how to make practical progress with that. What I see right now is that we can make progress by taking very seriously some crazy things that have not been considered seriously in how we've done both physics and biology. And these intermediate views are, they seem very reasonable to me. I don't think they're that different really from what I'm saying, except that I don't think they take this stuff seriously enough. And my practical reason for saying that is these things have been around, as you say, for decades. But people didn't do the experiments that we are now doing. And I think absolutely you can, after we've done the experiments, you can take them and say, oh, this makes perfect sense on this view, this view, we can interpret it in a different way. I agreed, absolutely agreed. But why hadn't they been done before? And I think it's because, so all I'm saying is I'm taking the kind of extreme view with the extreme sort of predictions and directions that this suggests. And we're going to kind of see what we find. And then maybe we can sort of, in the end, we can sort of smooth it over and say, well, we really can't say if it's a separate realm or not separate realm. And who cares? Because now we have a better accounting of how these things interact. And in the end, maybe it will end up as just another department. You know, it's like another department that you can't reduce to physics or something. I don't know. But I think in order to do the empirical work, you have to take some of these things very seriously. And I think these middle approaches, they kind of, they almost, and I'm not as familiar with all of them, so maybe this is unfair, but they sound like, and I'm a huge Patty fan and Rosen and all of that, like I love all that stuff. But I think it doesn't go far enough. It kind of keeps you conventional in this. I mean, as unconventional as they were to the mainstream, I get it. But it keeps you from doing some of this stuff that you get sort of pushed into when you really take this seriously. So I'm not sure how much of a difference there is, but methodologically, I think there's a difference.
[55:30] Amahury Jafet Lopez Diaz: I mean, to be honest, it's an excellent question. Sorry, if anyone else wants to jump, feel free.
[55:38] Carlos Gershenson-Garcia: I just remember that at the beginning of the conversation, you mentioned that one of the things that you wanted is novelty or no surprise. And it reminded me that in the early days of complexity, let's say late 80s, early 90s, some people were speaking about emergence as something that should be surprising, then some critics said, oh, then emergence is really a measure of your ignorance, not a physical property. And I mean, since then, there have been other less objective accounts of emergence or definitions or notions proposed, which don't run into that area. But from what I understand, you are intentionally introducing this surprise, but then let's say that, or it's just to say something that has limited prediction.
[56:51] Michael Levin: Yeah, what I mean by surprise, and I'm not a big fan of emergence in that sense, although maybe if there are better definitions now, I'd love to see them, so maybe you can point me to some. To me, it always sounds like we got something we didn't expect, but what I'm getting at is not just surprise and not just, I mean, we know complexity and unpredictability from simple rules, like of course that happens. I mean, things that are, for example, interesting thing, things that are behaviors that are recognizable to behavioral scientists. So policies, it's not just, well, it's complex, but like, yeah, it's complex. And it also looks like habituation and surprise minimization and all these things that behavioral scientists know in a way that is not accounted for by current formalisms. So, we know how much effort we put in to make something and what we get out of it is not accounted for by the methods that we're using.
[57:59] Carlos Gershenson-Garcia: But then do you think this surprise is temporary, in the sense that once we have a better understanding, then that surprise will go down?
[58:09] Michael Levin: What we'll have, I hope, so my sort of naive hope is that if we do the hard work of studying the stuff that I'm proposing we study, so this other latent space, then it becomes less surprising in the sense that we will have new tools. We will be able to say, okay, the traditional tools that we had absolutely did not see this coming. But now with these new conceptual, and I don't mean the tools, I don't just mean benchwork stuff, I mean conceptual too. Now with this, now we can understand what's going on. All the questions that Amma was saying, like we better understand the structure of the space, we better understand the relationship between those patterns and the physical interfaces that we build, so now we know why certain patterns come through when we do things. So it's not the surprise aspect, it's not the critical thing here. What's critical is we have a certain set of expectations, formal, again, metaphors that are currently used, and I think we need to push, they're incomplete, they're importantly incomplete. And that's all.
[59:13] Carlos Gershenson-Garcia: No, and I think it's worth trying because independently of what the results are, we would learn something about it. If, let's say, it gives better answers and understanding and explanation, it's great. And if there are some other limitations, then, I mean, because it might be limits of formal systems that, let's say, you cannot have consistency, completeness, and universal computation at the same time. And then you can choose two out of three. So for example, if we want consistency, say, compatibility, then we cannot have completeness, for example. And it might be that, let's say, we can answer some things, but then there are also new limitations that we find. But I mean, whatever the result is, it would be interesting to explore.
[1:00:24] Michael Levin: Guys, I apologize. I have to run.
[1:00:29] Amahury Jafet Lopez Diaz: All right, so I will maybe back, and if you want, we can maybe make a follow-up where I maybe present more of these ideas.
[1:00:36] Michael Levin: I would love to have that.
[1:00:37] Amahury Jafet Lopez Diaz: To be honest, again, I have the same question that you, because these ideas are kind of like being in the air. Since the 90s, you'll find, for example, proposals like Alicia Squarero's *Dynamics in Action*, where she says traditional mechanistic push-pull efficient causation is a deduction-based explanation, what we call covering law models, are inadequate for explaining intentional behavior. She proposes an alternative framework based on complex adaptive systems and general systems theory, which introduces causation via dynamical constraints and explanation to historical narratives. And again, despite that she proposed that in the 1990s, no one followed up that kind of research program, you know? At least as I translate your narrative into this, in-world closure of constraints is basically that the space is the space of releasable organizations or policies given a class of interfaces and dynamics. The distribution of forms is a distribution over attractors, policies, virtual machines induced by constraints, resource bounds, and coarse graining. And this free launch becomes quantifiable as the gap between what is explicitly specified in the interface and what is entailed by the dynamics of the coupled system. This seems to keep the entire experimental program that you have developed intact while avoiding the ontological step that triggers these access causality issues we discussed in the last hour. And again, if your view is that the ontological language is doing extra work beyond that, then the next step is to spell out where the rephrasing fails. And again, with this, I'm not saying that the closure constraint literature has already predicted every one of Mike's experimental findings in detail. It means that there is already a principled non-dualist research program capable of generating such prediction in form. And Mayone's claim to close will be that we-- I'm not saying that we already explain every result first. But Mayone's claim is there is a promising concrete alternative research program here. Platonic ontology is not the only game in town. So I would close with that.
[1:03:00] Michael Levin: Makes sense. I don't think it's the only game in town, but it's the game that's, and then again, it's not just about the Zenobot. So there's a bunch of computational stuff that's going to come out in the next few months that it'd be interesting to get your thoughts on. It's just, to me, it's the game that has led us to do experiments that hadn't been done before for some reason. So in the end, who knows what the ontological, what the philosophers are going to say about the ontology. I'm really not that worried about it. What I want is a framework that pushes to do experiments that for whatever reason hadn't been done.
[1:03:37] Amahury Jafet Lopez Diaz: Again, as I see that, what you conceive as platonic space is nothing more than PATIS, a symbolic rate-independent epistemic mode, and the physical world is this PATIS material rate-dependent epistemic mode. But if you want, in the next session, I can make a presentation where I introduce these different stuff, connect it better with what we discussed, like how we can reframe your program in terms of these ideas that have been in the air since the second part of the last century. And of course, if anyone else wants to say something before we close, it would be great.
