Discussion with Robert Prentner and Timothy Jackson 1

This transcript is automatically generated; we strive for accuracy, but errors in wording or speaker identification may occur. Please verify key details when needed.

[00:00] Timothy Jackson: My background, academically and in terms of my actual career path, is I'm an evolutionary biologist and I've been very concerned with the origins of novelty as a fundamental question in evolutionary biology. Because I've had a long-term interest in philosophy and my PhD is partly in history and philosophy of science, this naturally extended for me into a consideration, over the last decade, of not just the origins of novelty within biology but more generally. Specifically, I've done a lot of work on the evolution of toxins in chemical ecology in poisonous and venomous organisms. But the basic form of the question has sprawled out from my empirical basis in that research and become a very generic question about the origins of structure in general. That led me to connect with a lot of different philosophical paradigms. Ultimately I'm attracted to process-relational views as an evolutionary thinker, but there are models in generalised selectionism and all sorts of things. When it comes to this question that Mike's got us together to discuss today — about the existence, although that might be the wrong term especially if we're going to use scholastic terms, but the reality of a realm, again a loaded term, of a latent space, as Mike is calling it, of more or less definite potential, more or less definite possibilities that are not yet actualised — I'm very sympathetic to that. I think we need that kind of account; we need it in evolutionary biology. It's built into the reasoning you see in evolutionary context in generalised selectionism. But there are a lot of devils in the details. There are a lot of different philosophical approaches to this question. It's one of the big questions. It's a very big ancient question. A lot of ink, a lot of words have been spilled about this over millennia. Since Plato, to some degree — even though the question predates Plato — he made a particularly influential attempt in the theory of forms to solve this problem. In more modern philosophy, certainly since Kant in the German idealist tradition, and in its development into the process relationisms of people like Charles Sanders Peirce, Alfred North Whitehead, Gilbert Simondon, Gilles Deleuze, and also in Nietzsche in parallel, we discover this notion of an inverted Platonism. The question is, what is the source of structure? What is the source of definiteness? Do we require a model that is Platonist, in which definiteness is already out there, ready to be taken up, ready to be ingressed by actual evolving systems — ingression being Whitehead's term — or is it something shaped in a relationship of adjacency? We can bring in Stuart Kauffman's notion of the adjacent possible, in a reciprocal relationship, and something Deleuze would call counter-effectuation with the actual. Is definiteness in the virtual, in that latent space, primary, and then is definiteness in the actual formed by the ingression of those forms, or achieved by the ingression of those forms? Or do we have things backwards when we say that? That's why we speak of processual accounts of an inverted Platonism, in which a nuanced notion of actuality comes first. Then definiteness in the latent space evolves within this relationship of adjacency, always relative to some particular local actualized situation. That's some of the desiderata we've come together to talk about today.

[04:58] Robert Prentner: Let me first say a couple of things about my motivation, how I got into this. Very generally, I'm interested in the relation between consciousness and some more scientific or technologically based approaches to its study. To be open about it, I am not a physicalist about consciousness. Consciousness is a fundamental thing. That makes it difficult to find the relation between science, technology, and consciousness. I don't believe that consciousness emerges from complex computations or from putting together some complexity. Platonism or some variant of Platonism, Platonist idealism maybe, is a very suggestive way to think how those two things might be related. That brings me to Platonism; Timothy already said a couple of things about that. When we talk about Platonism, which is the main theme of what I understand we will talk about today, it's important to distinguish three senses in which one could mean Platonism. The first sense, which I would call Platonism 101, is the streamlined, watered-down version that we teach in undergraduate courses, Introduction to Philosophy, where Plato is presented as a dualism between a transcendent realm of ideas and an empirical realm, with the theory of forms. The second sense is what Plato actually believed. That's a very difficult question; scholars are fighting over that. It's not clear he had any coherent doctrine himself. Philologically, that's not clear. The third sense, which I think is the most interesting one, is a kind of extended Platon-like framework. In his later dialogues, like the Timaeus, Plato saw that there are problems to be resolved, so it gets quite complex. His theory of the soul, for example, is quite complex; he lays it out in the Timaeus. Usually that's simplified in Platonism 101. What Timothy mentioned as inverted Platonism—Whitehead's ideas—would count for me as an extended version of Platonism. It still thinks in the framework of having the virtual or the latent space or the ideal, in opposition to the empirical, actual, concrete. Plotinus, whom we now consider a neo-Platonist, considered himself a follower who extended Plato's ideas. He introduced the idea that you need to think about the ideal realm in terms of a hierarchy. You have a stratification: the One, the Intellectual, the Soul, and then the material, with emanation and realizations between those levels. There are many interesting modifications prompted by intrinsic issues of the framework. That's something I would like to explore and which connects to how consciousness relates to the physical, empirical world.

[09:06] Michael Levin: I was trying to take notes. Who did you say was with the stratifications?

[09:11] Robert Prentner: Plotin, Neoplatonism, so that's late antiquity.

[09:15] Timothy Jackson: Plotinus, we normally say in English.

[09:17] Robert Prentner: Plotinus, oh sorry, my German, I, yeah.

[09:21] Timothy Jackson: Trying to help Mike make that connection. Mike, did you want to come back on that or should I?

[09:28] Michael Levin: Please, go for it.

[09:30] Timothy Jackson: I completely agree with Robert that there are very complex questions of scholarship and interpretation when it comes to Plato himself. Famously, Plato hides himself for the most part in this dialogical format. He puts words in the mouths of various interlocutors. Where we might go, and Robert's already mentioned the "Timaeus", is this relationship: if we're thinking about the theory of forms, even though it crops up in a lot of different dialogues and there are lots of little aspects, the "Timaeus" is probably a great place to drill down on just a little bit because it is essentially Plato's cosmology where it's articulated. "Timaeus" is the one who actually articulates it, but then also the relationship between that and, what may or may not be a later dialogue, the "Parmenides", which is also an exploration of the theory of forms. You can read some aspects in the "Parmenides" as problematizing things in the "Timaeus". Different readings of that can be considered, again, in a sort of shorthand way, the basis of some of the trajectories that Robert's talking about. The Neoplatonic trajectory — Plotinus and Proclus and their followers — has a profound influence on the Christian worldview. That Neoplatonic view is the dominant view of Platonism that I think is baked into Western culture. Whereas the "Timaeus" and the "Parmenides" and the tensions between them possibly already admit of slightly different readings. I will agree with Robert that Whitehead is a very important figure — he famously says, "the safest general characterization of the history of Western philosophy is a series of footnotes to Plato." He's putting himself very much in that tradition, but with tweaks; he could point to Plato and say that trajectory is already outlined in the dialogues themselves. Even thinkers like Gilles Deleuze, who are sometimes considered to be anti‑Platonist, are really about an inversion of Platonism. Deleuze would say that when we invert Platonism — completing Nietzsche's project for him, and drawing on Whitehead and Simondon and others — we're not escaping Platonism. It's not really an anti‑Platonism. In fact, we're going to uphold a number of Platonic doctrines. In particular, Deleuze, as with these other process‑relational thinkers, has a very robust and central role for the virtual — what would correspond to your latent space, Mike. We're not, in the way we navigate all of this terrain, looking to escape that. For me, it's what are the dependence relationships here? Given my underpinning question is something like, what are the origins of structure or novel structures, those dependence relationships become really important. But I'm wary we could get lost in certain details. I do think it's important to quickly outline what's going on in the "Timaeus" — here are the set of operators in a Platonic context — and then see: do we need them? Do we want them? How do those interact with contemporary science? But it can get messy; there's a lot going on.

[13:44] Michael Levin: How do you guys see this impacting your work in particular on a practical level? Because I think some of the most interesting aspects of this is precisely when it integrates into the real research that we're all doing. So I've got my favorites, but I'd love to hear from you guys how you see this coming into your work.

[14:06] Timothy Jackson: Robert, do you want to take that?

[14:08] Robert Prentner: In the beginning, as I said, I think it's an interesting kind of framework to think about the relation of consciousness and the physical. Concretely, I've been involved in work on the so-called interface theory of perception with Donald Hoffman. The basic idea is that our perception is not truthful, but tuned to utility or usefulness. What we perceive is what is useful to us and not what's really there. In some sense, we make interface representations of reality, which allow us to act in a useful way. That's evolutionarily based. That's the way the view was produced. On the one hand, there's the computational theory of mind; on the other hand, there's an evolutionary approach. If you marry those, you get something like the interface theory of perception. Here there's an open question: what's actually beyond our interface or what are we interfacing to? One possible way to think about that is let's call it the virtual realm that we are always interfacing with. Here's one connection to some Platonist-like way of thinking that I see from my own research, which I find very interesting. I think that's the main concrete motivation I have for that. I also have a philosophical and historical interest in how things turn up in the history of philosophy. I'm a professional philosopher of science and technology. I'm interested in how it factors in, especially the question that becomes relevant nowadays of how to incorporate some dimension of meaning and sense into what we do in AI and science. The Platonist model might be very useful in thinking about that.

[16:30] Timothy Jackson: I'm also very interested in these historical questions. My interest in the origins of structure does extend to my interest in the origins of thought forms. Studying the history of philosophy as a way of trying to understand the evolution of the conceptual frameworks that we're deploying in different contexts is in itself a very interesting and useful task to me. It feeds back on the ontological project in important ways because it allows us to draw into question certain things that are taken for granted in a received view of the nature of things. If we trace the history, do this historicist move, of that particular idea back to its not necessarily original source: origins have a tendency to recede across transcendental horizons. There's a subtle—this is going to relate to interface theory—but a subtle way that we handle the existence of these horizons, which are boundaries on our capacity to reconstruct or perceive. We always have this lurking question of what's on the other side of that horizon, exactly as Robert was saying. I'm partly interested in this historical approach. I want to know why we think the things that we do so that I can understand what kind of wiggle room we might have when we're working on an ontology. But I do think of all of these questions as coming very directly out of my work on the evolutionary origins of novelty in a slightly more classical molecular biology framework, because there what I'm really interested in is the role of stochasticity, of arbitrary variations in facilitating the origins of novelty. This can be tied into the notion of the adjacent possible: you have a particular already concretized, already actual form, but you have this fluctuating, shimmering all around the edges of that. That can be an abstract or even poetic way of thinking about things at different levels. It's pretty easy to think about actual genes and stochastic point mutations in that way. Duplication propensities can be thought of in that way. If you tie the origins of novel structures in mutational events to the origins of novel functional traits, you have to bring in the role of gene expression, and gene expression has stochasticity inherent to it as well. I always take this term that I first heard Chris Fields use, the analogy of babbling. I think about gene expression as constantly probing an adjacent set of possibilities adjacent to the current regulatory framework that in some sense defines the non-equilibrium steady state of the organism; it's always shimmering, probing around the edges. There's actually a latent space of possibilities that is concretely actualized at one level already; it's there in this variability of gene expression. In order for it to be stabilized, possibly in ways that will be heritable, some confluence with the environment must occur; that will elicit a particular arrangement and stabilize it — a more classic evolutionary story. Embedded in all of that are deep philosophical questions already. When you're really drilling down on that, even the integration of a classic molecular biology story with the role of stochastic gene expression in the origins of novel functions is already a step outside a genetic determinist framework. I feel the honest pursuit of insight into the origins of novel functional traits can take one down this path.

[21:08] Timothy Jackson: So they're all, this maybe is a bit of a flame-flame answer, but they're always already related to me, the one implies the other. And then I become really interested in the, what I think of as the attempt to avoid front-loading the possibility landscape. So I don't want to posit the existence of definite structures if I could potentially explain them. The ideal, in some sense, is that all of the definiteness emerges from indefiniteness. Lots of devils in the details. But rather than saying these things exist, say they are eternal forms, and again, if we get to the Timaeus, we're going to talk about the operation that the eternal forms play in that cosmological dialogue. But instead of reaching straight to something that is eternal, that is timeless, and which this corresponds with the kind of mathematical Platonism that you've been referencing, Mike, the question is always: could we dig a little bit deeper and actually find a way to account for the emergence of those structures themselves? That becomes a research program in the foundations of mathematics, as well as in scale-free principles of self-organization in physical systems, and this historical approach to different philosophical options. All of those things are going to interact in fascinating ways. One of the things you've said, Mike, is that you feel like emergentism is a mysterian thesis. I would grant that as it is articulated, it's often mysterious, and people always caricature it by saying an algorithm proceeds to this point and then the magic happens. I want to do a lot better than that if I'm referencing emergentism. For me, the kind of Platonism that we see in a more or less straight reading of the Timaeus is where the materials themselves are not intelligent. They're ultimately chaotic. The receptacle, which comes before everything, is chaotic and completely disorganized. Plato needs, and in fact in a number of dialogues Plato has Socrates ridicule the earlier pre-Socratic natural philosophers who believe the materials themselves could be intelligent. Plato in the Timaeus needs these independent or transcendent operators of the demiurge, who is the only being that has direct access, without an interface, to the eternal forms and the reflection. The contemplation of the eternal forms that the divine craftsman is capable of is what facilitates his, capital H, his organisation of the fundamentally chaotic nature of matter. I see this as the more mysterian thesis, ultimately, in the sense that it's saying we can't really explain the origins or the genesis of the forms themselves. They come in as a transcendent posit, and they're what we need to explain all subsequent order in the cosmos. But because we are limited finite humans, the forms themselves, the demiurge, are irrevocably on the other side of this boundary; we cannot access the account of their genesis. To me, that's the ultimate in mysterianism. I think emergence, the promissory note of emergence, if we can take it as far as we can possibly take it — and we're going to bottom out somewhere — if we can more rigorously tell these ontogenetic stories about the origins of structure in all these different domains, that is the attempt to minimize mysterianism for me. Positing all the definiteness existing out there in the latent space and just informing the otherwise indefinite matter — which I don't think is the position that you're taking up, Mike — is, in a sense, a very coarse-grained, pun intended, description of what's going on in the Timaeus. Maybe if Robert wants to, or if you want to, Mike, we can drill down into more details in the Timaeus, how that relates to the theory of forms elsewhere and all of that.

[25:53] Robert Prentner: I have a question to Mike relating to that. I think you gave a very nice summary of what's happening here. My question is what I'm interested in, and I think that's an open question: if you pursue a technologically driven or engineering-driven project like Mike's. One important question would be, if you accept this kind of Platonic framework, you have this form or this virtual or however you want to call it, which you can somehow interface to. The question is what that buys you. There seems to be something happening there, and we need to be able to say to some extent what's happening there in order to make that useful. Otherwise it's, like you said, in some sense also a mystery in this position. You just have this eternal form and there are somehow—I don't know how it works. We copy them, we ingress, however we want to call it. But then that happens all the time. We do something. What we really want to have is we want to build something, we want to engineer something, and then we want to understand, to some extent at least, how these transitions to some novel behavior show up — some interesting things which are happening, which seem to point to some interesting dynamics in this eternal realm. In that sense, "eternal" — we could talk about what that actually should mean to say something is eternal, but that's a different discussion. I'm curious about what Mike thinks: what that actually buys you with such a framework?

[27:41] Michael Levin: It's a critical question because I'm committed to having these kind of ideas that actually buy us something in terms of novel discoveries, novel research programs and so on. I can start with this issue of mysterianism that you just brought. Here's the dynamic that I see in our fields. When you point out some interesting property, some generic property of networks or of these kinds of emergent things, when you say, where does that come from? People say that's just the fact that holds. It's just a fact that holds in the world. The benefit of that position is that it's more minimal. People who don't want to posit a non-physical realm want to keep their ontology very, very sparse. But the downside is that you then end up with a catalog of surprises. When we encounter these things, we write them down and then that's that. You just end up with a catalog of these things. For me, the fundamental project of science is an optimistic one where you say it's not random; we can't prove it, it's a metaphysical assumption, but we're going to assume that instead of a random grab bag of stuff that we can just catalog when we come across it, we can instead hypothesize that it's part of an ordered space, that what we need is a systematic research program of investigating that latent space to find out what's in it. This suggests that to some extent, and I agree with you, Tim, that it is a bi-directional relationship. I don't think these things are fixed. But I do think that there are contents there that pre-exist anything else that happens. The research program, I think, is the creation of tools to really map out that space. What are those tools? Mathematicians have their tools to map out their corner of that space. I see Xenobots, Anthrobots, all the hybrids, the cyborgs, all the crazy stuff that we make as exploration vehicles. They're periscopes that we stick up into that space and we say here's what the standard frog genome allows us to see. There's the standard frog embryo. Now we can tweak some stuff and we can invite in some things that are around that and systematically map out the patterns, and I use that extremely generally. These are patterns of behavior, patterns of morphogenesis, patterns of physiological states, and so on. These are patterns that I think evolution makes use of.

[30:47] Michael Levin: Here I think we have to start thinking. It's popular to think about constraints. A number of people talk about how these things are constrained. Sure, they're constraints, but I think more interesting is the fact that they're actually free lunches in a certain sense. I see scenarios all the time where I think evolution benefits massively from the fact that you don't need to go evolve them, create them, you get handed these amazing facts for free that you can then benefit from. So we as engineers need to learn to use these interfaces to know what's on the other side. I see these things very much as pointers. We make these things. The structure of the pointer is important. It isn't generic and it isn't doing nothing. It's absolutely important. But the thing with pointers is you always get more, you get more out than you put in. So you've got a pointer, but what you're pointing to is a rich data structure that you're going to be able to make use of. Now we can say let's make these things and let's work out the mapping. That's what I think is the research program, is to work out the mapping between the pointer, and the pointers can be extremely dumb things, triangular objects and such, and then more and more complex, and you get to dip into this platonic space of patterns. I think the research program is to create tools and then work out the mappings. What are these patterns? Robert, I agree with you in that I think a lot of what's in that space are not just facts about distribution of primes, but high agency patterns that are, in fact, kinds of minds. I think to some extent, that is the consciousness that we see. I don't think we make consciousness. I don't think we make it with AIs. I don't think we make it with embryos. I think we make interfaces to it. That's what we're inviting into the world. I guess the last thing I just wanted to say about the nature of explanation and things bottoming out. I agree with you that it would be nice to be able to say where the patterns themselves come from as a sort of the next step. But I feel like, and I'm no mathematician at all, in mathematics, I think they have a version of explanation that's lateral, things connect sideways. At some point there are things that just are, and there isn't a notion that it has either a historical or some kind of reductive explanation. Some things do, but eventually you get to these really fundamental patterns. That's just what it is. That seems okay because the alternative is an infinite series of why-is-that questions. It's got to bottom out somewhere and then are we going to be satisfied or are we going to say that's no good, what's behind that? I don't know.

[33:55] Timothy Jackson: I agree it's going to bottom out somewhere. One of the questions is whether it bottoms out in an operational schema or whether it bottoms out in structures or preformed objects. That's a very key question in the development of process relational thought. A generic operational schema, which you could take something like the free energy principle as an example of, might be an account of the genesis of structure itself. That's part of what underlies the inversion of Platonism for a lot of the process relational thinkers. I definitely agree that we're bottoming out somewhere, but I also think that minimalism is the goal in metaphysics or in speculative ontology. So we don't want to prematurely declare that we've reached the bottom. It depends on the mathematicians. I'm not a mathematician, but there are research programs in topos theory, category theory, homotopy type theory, where it's looking at what's fundamental being relations and transformations, and where structures, including things like discreteness itself, are emergent. So there is a kind of historicizing or at least a processualizing approach to the foundations of mathematics, as I understand it. I'm supportive and want to express that. I think the empirical research program you are outlining will absolutely bear fruit because there is a latent space of more or less definite potential adjacent to any given actual situation. If you manage to create a novel actuality, you will pull down never-before-seen potentials. But the question is whether those potentials in their relative definiteness pre-existed that novel actuality. There are a lot of ways of thinking about that. One of the other big things in my life, as I think you know, Mike, is that I'm a musician. I'm a very keen improviser. We can talk about the relationship between musical form as an enabling constraint. Constraint and enablement, or constraint and free lunch, is a false dichotomy. A constraint is a form of enablement for the emergence or the achievement of definite outcomes, but they're not determined by that constraint. The constraint defines an arena and there may be any number of ways of traversing that arena. Those possibilities, those paths do not exist as a bunch of discrete paths prior to their traversal. There's lots there.

[37:07] Michael Levin: Can I ask, that's an interesting point. As I said, I agree that these things are changeable. But especially in the case of, for example, synthetic beings that have never been the subject of specific selection, and you make these things and they have very coherent sets of forms and behaviors. If these things didn't exist prior to the creation of the physical interface, you can tell a story. The standard story is the slow and gradual process of evolution. That's what nailed this particular kind of thing. But when they show up just like that, where, if it didn't pre-exist, when does it and how does it show up?

[37:52] Timothy Jackson: It's a vexing question, and I think this is partly why we have the philosophical research program that is adjacent to the empirical research program, and why an examination of Platonism is actually the right thing to do, and it's different: the different sources or the different active principles within that. That's a big subject in and of itself. But if you're looking at something like a differential field as primordial, but indefinite, but unstable, as a kind of primordial active principle, which in Platonic terms, we can start thinking about the fusion of the core and the one. That's where some process philosophers go. That's something we would need to sketch in and of itself. But you can look at the fact that the thing is always this potential inherent to matter itself and is always pressing to overflow these contextual constraints. So the constraints of any particular actual arrangement. If you change the actual arrangement, what's going to happen is that previously canalized potential is going to overflow and reorganize itself into something else. But what's being drawn upon is not really something; it's the combination between the definiteness or the determinacy which exists in the actual and the excessive inexhaustible potential which underpins everything in this sort of view. That determinate pattern is brought into existence by this particular relational nexus, which you can think about in exactly the terms that you're thinking about. You've created a novel actual system. You've said that xenobots are partly the consequence of a long evolutionary history in the sense that those cells have learned a lot through the evolutionary history of multicellularity. So they are doing things that are at least partly predicated on, but not determined by, that particular history. But you're creating a novel actuality, never before seen in the world. But it has this charge of potentiality, which is still canalized; it's still constrained by various actual features of the cells themselves. But it's not constrained by, as you've articulated nicely in other contexts, the population of cells that previously was holding it in place and saying, "you are a skin cell because you're here in this relational network." Now we're out of that relational network, but we're taking a lot of our constraints with us. We've got this charge of potentiality. That's something we can build up. As part of this research program, there's the philosophical aspect, but there's the mapping of this across the diverse domains of science and what someone like Gilbert Simondon would call genetic encyclopedism. The free energy principle shows us partly how that kind of thing can be done, because it is a series of generalizations. That's another aspect here. You've got this potentiality, you've got a novel actuality, and their union is what brings that new pattern into existence. It doesn't pre-exist, predefined in the virtual space. That would be the hard processual view. You can talk about the ways in which Whitehead might feel uneasy about that, but the more thoroughgoing process ontology, maybe of someone like Gilles Deleuze or Gilbert Simondon, which is more minimalist than Whitehead, even though Deleuze takes a lot from Whitehead, is the kind of thing you'd be looking at. That can be rationalized as an extension from Platonism itself. So it's still in that Platonic tradition. It's just a slight difference of emphasis and a rearrangement of the way the operators that Plato defined actually work together.

[42:16] Robert Prentner: So one issue that I think will always be brought up in this context. If you take this research project, Mike's, where you say, what you're actually doing is creating pointers into some ideal realm or virtual realm or however you want to call it. And everything that you do is basically something like that. And then the question is if you then want to talk about this transcendent or virtual realm beyond, how can you do that? And where, also empirically or scientifically, what is the methodology that you need to use there? It seems to be very difficult and it's a Kantian point where you see also the interface theory suggests something similar: our conceptual tools are not made for exploring that realm. But we still want to explore the realm. So there seems to be an obvious answer. And the question is how to get across this issue and how to resolve it. How to make sure that the relations between the patterns, or whether it's a process of emergence or evolution or whatever that happens in the ideal realm, how is that different? Isn't that only a kind of more abstract way of talking about what we empirically find? So when you mention, for example, the free energy principle, how can something which is remotely similar to a free energy principle tell us something about the Platonic space? Why isn't it just the abstraction from some empirical processes that we see? So there's this skeptical question.

[44:00] Timothy Jackson: And there's always an irreducible skeptical question, and I think we will need to be very careful and sophisticated about the way we handle the nature of transcendentals, and that could be pluralized, but those horizons, and also the nature of what we're saying is on the other side of it, as we discussed in e-mail, are we saying that there is an absolute truth or an absolute nature of things that we don't have access to, or are we going in a more relational direction and saying that what is actual, what is real is, at least to some extent, determined in these relations of adjacency. And so we're going more in a perspectivist view, a perspectivist ontology. There's a lot of different options in this space. And I think Mike is showing one way that you absolutely can experimentally, empirically investigate the structure of that Latent space.

[45:07] Michael Levin: Finish with that. I'll...

[45:09] Timothy Jackson: What one might be looking at here is a multi-pronged, multidisciplinary, transdisciplinary research program. Aimed at getting to the bottom of some of these vertiginous philosophical questions and the way they interface with empirical science and empirical reality. On the one hand, a classic approach in physics is invariance. You're still looking for invariant structures. You're trying to look at the horizon and read these invariant structures off it by, as we say in evolutionary thought, reconstructing the ancestral or primordial states, inferring them based on extant variation. You're looking at the empirical part of it, and where something like the free energy principle can potentially be useful is that you are mapping a certain kind of invariant schematism, which again tells us so, and there are very important principled ways to connect the fringe of principle to its own origins, like Hermann Hagen's synergetics, dissipative adaptation, and all of that—these are all things that need to be sketched here. Then the way those principles have transcended their thermodynamic origins: what's going on in quantum information theory, what Chris Fields is working on, and in the more purely mathematical side of things—I gestured towards topos theory and homotopy type theory. What are the analogies there? Can we find invariant operations that are analogically, but operationally so—not just that they look like each other, but that they are formally, operationally, essentially isomorphisms in these diverse domains? You have the mathematics story. What are the accounts of the origins of structure in mathematics that are currently being explored? You have the empirical story. Mike's research program itself is at the center of that, but it's this genetic encyclopedism. It looks at how in diverse domains—from the quantum to the thermodynamic to evolutionary biology—systems self-organize and the origins of novelty in each of those domains. The philosophical story: indeed, as they appear to have, it's striking if you read "Timaeus" that all of these thoughtful forms are pretty much in there, at least in an incipient form. History of philosophy: series of footnotes to Plato and all of that. To what extent have the things that we're seeing in these different contemporary, cutting-edge research programs also been discovered as thought forms in the history of philosophy, and what ontological options might we recover from there? It's at least three-pronged because you're asking deep questions. Interface theory is an account about how we are going to relate to these transcendental horizons. That's also what Whitehead is doing. That's what all these guys are doing. Deleuze calls this philosophy transcendental empiricism. James Bradley called Whitehead's philosophy a transcendental cosmology. These are all post-Kantian philosophers; just as they recognize that we don't want to be anti-Platonic, they're not anti-Kantian, but they wonder to what extent a rigorous study of the empirical world, the mathematical world, et cetera, can enable us to reconceptualize our relationship with horizons and boundaries so that we're not just Kantians per se.

[49:21] Michael Levin: There's a couple of other interesting things that may or may not lead to this in the research program. One is, especially in neuroscience, we're starting to see our understanding of the mapping between the pointer to the thing it pulls down is actually not very good. The reason is that Karina Kaufman, a student of mine, and I reviewed a bunch of those cases where, as an example, there was a guy that had a third of the cortex of a brain volume of a chimpanzee, and he had above normal intelligence and was doing a master's degree in math and things like this. We clearly don't understand. It's not that you can't tell some kind of story of redundancy after you hear about these things, but we don't have a theory that would actually predict that would happen. It's a bunch of epicycles you have to put onto things later to shoehorn this into your assumptions; it doesn't fall out of anything that we know about neuroscience, so in cases like this you seem to get a lot more out than what you put in. Understanding that kind of mismatch in these cases might be very relevant. Other examples are scenarios where you have causal drivers that don't map nicely onto any physical structure. I'm talking about virtual governors and effective networks: when you train these gene regulatory networks that are fixed, the hardware is completely fixed. People say, the reviewers to our papers ask, "Where's the memory?" You can show them what happened after it's—it's a dynamical systems thing and you can show them the causal, the effective network after training. So where is that network? Well, it's not in the hardware. Where is it? Or Darcy Thompson's grids. In his book on "Growth and Form", he shows you take a species of an animal and you put it on a Cartesian grid, and then you do a mathematical deformation to the grid, and what you get is a different species that also exists. Since then—what was that, the 20s, 30s?—where's the grid? What is this grid? Partially in answer to your question of how we describe the contents of that space: these are the kinds of tools, mostly mathematical, but also psychological. I would argue some of the stuff we get out of psychology are probably really what you're talking about—patterns in this space. These mathematical objects are functionally critical not just for explaining what you see, but for driving the next discovery, for making functional changes, for controlling in an engineering context. That's the beginnings of the kind of thing we need to describe the patterns in the space. There are going to be behavioral patterns, there are going to be geometric patterns—those kinds of things.

[52:43] Timothy Jackson: I certainly agree with the assertion that we don't have a good mapping, and I think we've been labouring under an ontology which gives us a history of false problems, certainly in the history of modern science and with the mechanical philosophy. I do think in a way this also goes back to Plato. Whitehead's characterization of the history of philosophy as a lionization of Plato, which it is and should be because he's this towering genius, is also an indictment of the history of philosophy in a certain sense. That's why an exploration of these options is required, even though we may end up in a place that's recognisably quasi-platonic. I'm in full agreement with you there. I think the question again is where the agency is and whether these forms themselves have agency or whether they are just a more classical formal causation, which is a constraint on some fundamental activity or agency. That's the account that I prefer. But again, there's a ton of work to be done. My interest is certainly not saying that we or anybody has the answers, but given that I'm coming out of this evolutionary perspective, as my interests spread out from evolutionary biology into other domains of science and into the history of philosophy, I've become very interested in the general nature of this way of thinking about Genesis onto Genesis, et cetera. How broadly can we apply that framing? An interesting thing is that wherever I look, including work in the foundations of mathematics that's very contemporary, there is recognizably at least a process-based framing being deployed there. I also want to say I very much agree about the psychological. I think all of the philosophers that we've referred to today, even in passing, would very much agree with that. Many of them were trying to connect deep questions about the nature of lived experience with the science of their times or with mathematical frameworks. Certainly, I noticed that you referenced Jung a couple of times in your paper, and that's a whole other conversation. I've done quite a bit of work on Jung, and I don't know if you've seen some of the dialogues that I've done with Matt Siegel on Jung, but you might find some of them interesting. Jung himself admits of quite different readings. He's very frustratingly self-contradictory and inconsistent in many of his expressions. There's definitely a lot to work through there. A philosopher I've name-dropped a couple of times, who I think is my current guy for exploring this to the hilt, is Gilbert Simondon, a mid-20th-century French philosopher. He's deeply influenced by Jung, which is very unusual among serious academic philosophers even at that time. He has many big books, but his biggest book is called "Individuation in Light of Notions of Form and Information." There he rolls out this approach of genetic encyclopedism, where he tries to map an operational schema of individuation that does not presuppose the individual or the existence of a determinate form, because he thinks that's begging the question. He tries to derive a minimal operational schema for the process of individuation that doesn't presuppose the individual by mapping it across domains from quantum mechanics to culture and society. He is very rigorous at the level of detail that he applies.

[57:00] Robert Prentner: I have a question for both of you. Timothy, you said it in passing, but I wasn't sure whether I understood it correctly. Mike, you said somewhere in our conversation half an hour ago something like, when we do AI and people talk about AI consciousness, we don't create consciousness, but we interface to consciousness. That's roughly what I believe. Consciousness is not something that we can create. It's fundamental, but we can interface to it and we can make interesting interfaces. The second thing applies not only to consciousness but also to agency. We talked about agency a lot. It seems to me agency is something that belongs to this virtual ideal realm and not to the empirical realm. That's not how a lot of people think about agency. In the non-reductionist literature, you find this idea that we create these complicated embodiments, and from that you get agents. I would make the distinction between agents and embodiments of agency. Embodiments of agency are just embodiments of agency. They're not agents themselves. Agents are something that belongs to the virtual or the transcendental realm. My question is whether you share this or how you think about this idea. How do you think about agency? What's the role of agency? Is it an empirical notion that belongs to the physical world, or is it something that is also reached by these pointers?

[58:45] Michael Levin: For me, and I certainly will, in the earlier stuff that I was writing, because I was trying to cement some things before getting into this weird stuff. So you can't do everything at once. People just completely turn off. I've done a lot of stuff in diverse intelligence and collective intelligence and so on, where I say, the agent is formed by the parts aligned, where ultimately that was just loose vocabulary. I really think, I agree with what you just said, but I think what we build our interfaces in the agency is elsewhere. But the one thing I'll just describe quickly is we have a minimal model of something that I think is interesting. And that is that I think these pointers are not even necessarily physical objects like embryos or biobots or whatever. We have a minimal model of this in algorithms. And what we have is work on these sorting algorithms. These are dirt simple things, bubble sort and things like this. This is why I get really frustrated when people say, these AIs, well, I make them. I know what they are. I write them. I say, you don't even know what bubble sort is. We make these things. We do not, I don't think we understand at all the things that we're pulling down. What we see about this is you have your algorithm and the system does do what the algorithm says to do, and that's the constraint part, and that's the part that we can't escape either as physical, in physical embodiment. You've got to obey the laws of chemistry. There's no way around it. However, they also exhibit interesting competencies to address the problem you forced them to address in new ways that you actually didn't tell them about. And while doing that, they also have these interesting side quests where they do some other stuff that wasn't in the algorithm at all. I think what's happening there is when you make these things, you get three things. You get the thing you wanted it to do. You get some extra competencies in doing that thing that you actually didn't put in. And then you also get some other stuff that isn't what you wanted at all. It's a side thing that's like a super minimal version of agency, and people say you can make them do some stuff, just like we are forced to do some stuff. But even the dumbest, simplest machine is doing some things, I think, decides what you wanted it to do, what you told it to do. And it's not just complexity, it's not just unpredictability, it's not just random stuff. It's something that we clearly would recognize if we saw a creature doing it. It's something that we clearly recognize as a kind of very simple mind of some sort. So that's what I think about the agency. You can try to force it to do things. If you're a good engineer, you will force it to do some things. But you're going to get more than you put in. You're going to get some stuff that you actually did not put in. And there, that is where we have to start trying to understand the mappings of the pointer to that space and what is in the space. Because I actually think that in building some of the new robotics and some of the AI systems, I don't think you're getting human minds, but I do think we're now fishing in a region of that space that maybe we've never pulled from before. We're getting some pointers to things that, at least on this planet, probably have never come out. That's my take on it.

[1:02:27] Timothy Jackson: I know we're almost out of time, and this is just a huge question. So slightly nomically, I'd say I don't think the forms have agency, but that agency requires forms. It’ll hinge on the way we think about the Cora in the Timaeus and its relationship to Heraclitus. In a sense, you could say that I'm a Heraclitean, a very coarse-grained way of putting it. But what is fundamental is drive: generativity, differential generativity. This would actually flip classical thermodynamics, in the sense that any entropy would not fundamentally be an index of disorder, but actually the source of novel order primarily, initially, and then the dissolution of established order is a secondary immersion process. That's a long story to tell in and of itself. I think we'd be able to find an account, Robert, where my position would dovetail with yours, but some of the operators are going to be almost inverted. Having said all of that, I'm an anti-substantialist in that I think it's just wrong to say things like "consciousness is fundamental" or "matter is fundamental." I think that whatever we're dealing with here in terms of the most generic nature of reality doesn't accord with those descriptions as we normally use them. So using those terms is already bringing baggage into the discussion, and I'd rather start without that. If we're going to really go down to brass tacks, I'd rather start without that. But if pushed, I'd say agency is canalized drive, and drive is fundamental. Agency is driving towards some particular goal, and that requires constraint. That requires organization. But what's really agential, or what's driving a system forward, is something that's more fundamental and is always excessive. So no matter what the constraint function, it is going to deviate from that at some level. Because it's a random dynamical system. It always has the dissipative generative term in the Langevin equation as well as the solenoidal term. So there's always drift. Drift is inexorable. Drift can't be stopped. That's one of the ways that I bring it back to my grounding in evolutionary biology. These are all very related things for me. I noticed, Mike, you've been talking with Brandon and McShay. I do like to talk about the zero force evolutionary law, and the ontological correlate of the zero force evolutionary law is fundamental drive, which is essentially a Heraclitean way or the way Nietzsche would take that. But again, long story.

[1:05:27] Robert Prentner: What's that? The zero — I don't know that.

[1:05:33] Timothy Jackson: The zero force evolutionary law, in its most basic sense, is just saying that change will happen. Things are always, things are always varying and that is grist for the mill of selection. So when you have the generalised selection triad — variation, selection, heritability — which is really a pre-Darwinian thing, Darwin develops it, Charles Sanders Peirce codifies it; it says all logic is in different ways a variation of that. And then Richard Lewontin, of course, famously as a 20th century evolutionary biologist, just describes it in that way. But basically she's saying that variation is fundamental and it's a dynamical term: variation is always accruing. It cannot be stopped. Drift is always happening in a genetic sense.

[1:06:24] Michael Levin: Robert, you got the last word for tonight.

[1:06:27] Robert Prentner: I think it's totally exciting. We're discussing all this abstract and complex stuff, and it's really interesting. But I think this question—how to break it down to concrete, empirical, scientific research programs—that's really the interesting question. It's one of the few research programs that I know that take a non-standard philosophical, conceptual approach and try to make it into something very tangible. I think that's what we need to do at the moment. That's really where we need to go. My hope is that some of the more complex questions will be resolved, or at least we will see some solutions, by going through that. We don't need to endlessly debate and go back and forth. It has happened for millennia now. So I really think if we can manage to break that down into some concrete tasks, empirical tasks, something which I can give to my undergraduate students, that would be real progress.

[1:07:37] Michael Levin: Let's come back in a couple of months; we have a few pretty crazy things cooking. We need to finish this up, and I'll show you some data. We've been doing some wild stuff on the transcriptomics of these novel synthetic organisms because they have an extremely different transcriptome. We're also investigating some more very basic cellular automatons, some things that look minimal, stupid and mechanical. And it turns out that that's not it at all. We'll come back, and I'd love to get your take on some of these things.

[1:08:22] Timothy Jackson: Fantastic.

[1:08:25] Michael Levin: Amazing, guys. Thank you so much. Super interesting.

[1:08:28] Robert Prentner: I thank you very much for your time and great to meet you, Timothy.

[1:08:31] Timothy Jackson: Great to meet you too, Robert. Thanks, Mike. I was chatting to Matt Siegel about this because he's been reading your Platonic Space paper as well. We're going to chat about it tomorrow as you've given us permission to. We both agree that it's wonderful that you are so open to engaging with people from diverse academic backgrounds and wading into this philosophical fray, but precisely as Robert says, with that empirical counterpart. That's rare. I think it is something to be celebrated.

[1:09:09] Michael Levin: Well, thank you so much. I appreciate that you guys are going to go through it. I'm going to listen carefully to that. So getting the terminology. I need to figure out if we are keeping this terminology because it sounds like things people have said or do we actually need new terms. So you guys are the experts. I'll listen to it, then we'll figure it out. Welcome.

Discussion with Robert Prentner and Timothy Jackson 1

Watch Episode Here

Listen to Episode Here

Show Notes

Transcript

Related episodes

“The Unreasonable Effectiveness of the Behavioral Sciences in Developmental Biology and Biomedicine”

Discussion #1 with Katrina Schleisman and David Burke

Conversation with David Resnik 1

Discussion with Robert Prentner and Timothy Jackson 1

Watch Episode Here

Listen to Episode Here

Show Notes

Transcript

Related episodes

“The Unreasonable Effectiveness of the Behavioral Sciences in Developmental Biology and Biomedicine”

Discussion #1 with Katrina Schleisman and David Burke

Conversation with David Resnik 1

Thoughtforms Life Podcast