Conversation with Richard Watson and Leo Caves 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] Michael Levin: Leo, I would love to know more about the Eurythmic stuff. Is that an okay place to start with everyone?

[00:08] Leo Caves: Because I don't know if this connects with you: it's living in this coherentist perspective where I'm trying to balance a whole series of concepts but possibly not found in any particular place. Picking up on this theme of resonance, I'm attracted to Deleuze's notion that concepts themselves can resonate together. In my path through life I'm looking for these things that resonate. Constantly, that's my filter, my sift. Is this notion of where they came from to get to the rhythmic stuff? Is it notions around things versus process and the notion of what is it that we are talking about? What are these entities that we're trying to understand? I did dabble in process philosophy a little bit, just to get an infusion of that perspective. That related to my interest in second-order cybernetics; I'm a big fan of von Foerster and the Biological Computer Laboratory. I guess an amazing environment that I would have loved to interact with Ask and Ashby and the various logicians that were there. Von Foerster's notion of the eigenform: we have patterns or processes that, when we as observers engage with them, there's a reflective, reciprocal, recursive interaction that may stabilize. In that interaction we are observers, not external. We're in there with the phenomena we're interacting with. There's some sort of stabilization. That becomes what he calls a token for an object. So that's the entity.

[03:15] Michael Levin: It's an object when you resonate with it in a way that finds some stable phase.

[03:22] Leo Caves: Yes, stable, yes, stable.

[03:24] Michael Levin: Stable synchronization. And when you don't, it's not an object.

[03:27] Leo Caves: Then it will cease to be a pattern to the observer.

[03:35] Michael Levin: Yes.

[03:38] Leo Caves: Those notions of persistence and things are about how long that thing sticks around and whether, as a function of the observer, we'll spot it or not. I think this relates to Michael.

[03:51] Michael Levin: It's another way of saying that it's an object to me. If I can synchronize with it, it's an object to me. If I can't synchronize it, but you can, it's an object to you. So being an object or not is a relational thing.

[04:05] Leo Caves: Michael has told us; you recollected the science fiction story of the high-density aliens who don't see human beings because we're just diffuse. Or the rest of the matter, unless someone starts to say there might be a little pattern there. It's very observer dependent. I'm very attracted to that notion of "I can form." The few emails I sent Michael were about this notion: how else has this been talked about? There's a notion of the wave structure of matter from Milosz-Wolfe, which is a view that the entities are standing waves in the solution of a 3D scalar wave equation. You can get these standing waveforms, and their work shows that they exhibit many of the properties of things that we, in other terminology, would call electrons.

[05:24] Michael Levin: So the property of a wave that it is a standing wave, we make that observer dependent because we can synchronize with it.

[05:43] Leo Caves: Yeah, I mean, okay.

[05:44] Michael Levin: So it's standingness becomes relational as well. Yeah.

[05:47] Leo Caves: We have to acknowledge that when we step from the world of the second-order side and exodus, where we are recursively coupled with the world, in fact, we co-construct the environment. So there is no out there and there is no objectivity. Forster says objectivity is the delusion that observations can be made without an observer. But when we step into other frames of reference that we may have a more standard out-there view of the world, the notion of the observer possibly goes away. I think there is an undercurrent in this. What we're getting into is alternative interpretations of quantum phenomena, alternative to the historically predominant Copenhagen-type, Niels Bohr–led interpretation, where the observer does come in. There are interpretations that put entities that have properties like waves or particles. One interpretation was in terms of standing waves. The wave equation has the notion of outgoing waves, but also an equivalent solution of incoming waves. So you've got the notion that any entity is in reciprocal exchange; it's influencing.

[07:37] Michael Levin: So it has an internalized motion which produces waves going outwards, but you can also think of it as waves coming inwards which create the entity's motion.

[07:49] Leo Caves: The entity is right.

[07:55] Michael Levin: The entity is the relationship between those two things.

[07:58] Leo Caves: Which is wonderful to switch languages again, because I'm interested in category theory through my reading of Rosen over the years, which goes way back and I'm no expert, but that notion that in categories you can have sets or objects and mappings between them, but you can also equally take that entity out and it is described just by the mappings that come into it.

[08:36] Michael Levin: The entity is nothing but what the entity does in relation to other things.

[08:42] Leo Caves: It might be a set of a whole set of mappings, all waves. We can change that, change our language accordingly. But again, that speaks to some sort of observer-dependent spotting. There may be in this flux of interchange some points in that, whether we call them standing waves or eigenforms or things, that sit there. Okay, so the eurythmic stuff. At a certain point in my life I moved from UK academia into Portugal and was looking for interesting places to interact with. There's a eurythmic school in Lisbon, led by Professor Crocker, who have picked up on the de Broglie–Bohm wave interpretation of quantum mechanics. This goes right back to the early days of quantum theory. In those early days where people were looking for the meaning of this new conception of matter and the quantum states, De Broglie did introduce his wave interpretation of quantum mechanics, but got roundly suppressed by Bohr and colleagues who were more senior to the extent that he almost dropped the program and picked it up later. In that we've got the notion that rather than this complementarity principle of Bohr — that it's wave or particle depending on how you observe the entity — in the formulations my understanding is that you are both, and that the waveform will have a particular peak which in the Lisbon school they call the acron. That particular peak you might regard as a particle entity, and around this a theta wave or pilot wave or carrier wave, depending on — it's called various things — is emanating from that. So that's a guiding wave, and the peak follows the density of that pilot wave. When you are looking at tracking the motion of this thing, the particle entity will be moving and guided by the pilot waves. And when you've got interacting sets of these things, the waveforms or the guide forms start to interact and the waveforms get complex.

[12:27] Michael Levin: The same achrons can either repel or retract depending on the waves, right?

[12:35] Leo Caves: Waves and the wave patterns, the constructive and destructive patterns.

[12:38] Michael Levin: I love that.

[12:40] Leo Caves: They retain a history of previous interactions such that the motion of the aquamide particle is informed by the history of those guiding waves.

[13:04] Michael Levin: The phase it's in, the dynamics of its internal complexity are modified by that history.

[13:12] Leo Caves: The motion is a product. All the previous interactions of that pile wave.

[13:27] Michael Levin: That would mean that electrons are not all the same. That you can scratch them and make them a bit different, right?

[13:33] Leo Caves: For a long time, I've thought about this degeneracy, this assumption that all of these electrons are the same.

[13:56] Michael Levin: Independent things, if it doesn't do that, it's not an electron. All the electrons are the same.

[14:03] Leo Caves: Now, I don't know how much I've conveyed my word picture. What's interesting about the discern wave interpretation of quantum mechanics is there are macroscopic analogs of it. If you come across these hydrodynamic quantum analogs.

[14:28] Michael Levin: Brilliant.

[14:29] Leo Caves: That means nothing like an intuition pump, able to observe these things. Quantum phenomena are reproduced.

[14:40] Michael Levin: I saw that there was a paper by Bohr's grandson saying that the macroscopic hydrodynamics missed something important about real quantum effects. Is that something you're clued in on?

[14:58] Leo Caves: I'm not aware of that.

[15:00] Michael Levin: OK, we'll come back to that.

[15:01] Leo Caves: I thought it might be of interest to you guys that I'm going to sketch or hint at some of the interpretations of the pilot wave, one of which is that it contains this history or memory-like effect. The other is a paper I haven't connected with yet but became aware of, claiming the structure of the pilot wave is Turing complete. There's a kind of computational interpretation of the role of the pilot wave. I've not dug into that yet.

[15:48] Michael Levin: It doesn't mean that it's running an interesting program.

[15:53] Leo Caves: It's a standard framing clip that will get placed on moderately complex objects. Is it Turing complete? It gives you some indication of—depending on which lens, or which set of tools you're using, you can get different insights into the phenomena.

[16:18] Richard Watson: Could I ask a question about this? About this business with the electrons not all being the same, a couple of questions. What do you think is the overall memory capacity of something like that? How many different scratches can you put on it?

[16:41] Leo Caves: Interesting. There's got to be something related to the matching of the intrinsic frequencies of the pilot wave phenomena and whatever it is that's impinging upon that. To the extent that those two things will couple, that must put a limit on the memory effects that are observable or relevant in a given context.

[17:21] Richard Watson: All of this ties into the polycomputing business that Josh Bongard and I have been talking about: if a lot of the onus is on the observer to say what's going on, then the question of how much memory is in the thing itself becomes really weird.

[17:44] Michael Levin: I don't know that the question is well formed. Suppose I were to ask you how much memory can a rock hold. I decided to use the rock's orientation on the table, just by turning it around, to mean different things. If I can discern a thousand different orientations, a thousand different bearings that it can point in, then it holds a thousand different things. If I can't discern that, then it doesn't. Even just one continuous value, if there is such a thing as a continuous value, can hold as much memory as you like. The tricky thing is that we want the information to be pseudo-discrete, quantized. Then the question becomes how many different configurations are there, not just how many different values of a real-valued number are there. That makes it finite. Because of harmonics, it seems to me the answer is: how deep do you want to dig? If you're able to discern higher-frequency harmonics, then you'll find more configurational possibilities within it. Now you're moving in a fractal or recursive space rather than a continuous space. The answer is still: as many as you are able to tune into.

[19:22] Richard Watson: That makes sense.

[19:26] Michael Levin: Leo can tell us if it's true.

[19:30] Leo Caves: It chimes to me that the more that you start to live within a kind of relational worldview, the more you drift towards that, the more you start to realize that this whole idea of the properties being placed within the object starts to shift: it's in the relation. That's what we're talking about.

[20:04] Michael Levin: The answer to how much information can it hold is how deep can your relationship with it be?

[20:13] Leo Caves: The degree of coupling will be related to the degree of information and will reveal different aspects. And so the richer the coupling, the more richness of the phenomenon is brought forth.

[20:43] Michael Levin: So I would like to dig deeper on two particular directions. One is I'd like to know more about connections to cognition that go beyond memory into territory of learning and adaptive behavior. And the other I would like to push on is relationships to biology. Perhaps we'll do that first. It must have occurred to people, Walter von Foerster if no one else, that the background that we're talking about in terms of the nature of matter and otherwise inanimate things also applies to biological individuals. I'm an acron, you're an acron.

[21:34] Leo Caves: Let's not conflate, because I don't think Von Forster was talking very much about electrons, although he did start with quantum. Von Forster had a paper on the quantum mechanics of memory, even before he came to the States. It was his introduction to the Macy group. One thing I appreciate about the von Forster nucleus is that the people around the work they pushed focused on biological systems. Another one of my particular compasses is that one of the things that will guide me on my path is that there's a sort of continuity of the phenomenon that we're talking about. It's sometimes called scale and variance or universality. The phenomenon we're looking at is the same principle and it will manifest differently. Science and disciplines created different languages to talk about these things. I'm always driven by the fact that that's uncomfortable. It doesn't resonate with me that there are different types of phenomena in different languages at different levels. That they are describing different kinds of realities or different kinds of rules. I'm happy that there are languages that, with a certain level of abstraction, give you a set of tools to talk about. I'm always looking for something that sits below that.

[23:29] Michael Levin: And simultaneously above it.

[23:33] Leo Caves: It's hard to get out of that spatially embedded. The notion of autopoiesis really came out that same scene. Recursive.

[23:58] Michael Levin: But what I'd learned about autopoiesis didn't seem to have the same Akron, theta wave nature to it. Did I not read enough?

[24:15] Leo Caves: No, because I did a poor job of separating those two. I was trying to give you some indication, and I've conflated them with different influences; in me there's a center of these things coming in—I'm trying to make the coherence of these. When I brought in the notion of the wave interpretation in quantum mechanics, that was to me a domain where you could interpret that through the lens of the eigenform. But there is something, some entity that persists, has properties. The language is really tricky and slippery when you start using words like "properties"; that's just the way a language is constructed. That was a domain where I saw there's an instantiation of the kind of thing that Von Forster is talking about here: the fact that what is it that we take to be an entity that we start to talk about? Where does that come from? That reciprocal coupling, that stabilization. And so when I started to talk about the wave interpretation of quantum mechanics, I saw it as a domain—here's some instantiation—at a different level from what Von Forster was talking about, because we're guided by what are these underlying principles, not what physical scale it might manifest at. What's interesting is that in certain kinds of domains, different kinds of salience and interpretations become evidence. That's interesting because then if you're taking a more universalist approach, you say that's the language or interpretation that's taking place there—what does that mean in the server scale?

[26:17] Michael Levin: Yeah.

[26:18] Leo Caves: That was the dance I was trying to do.

[26:24] Michael Levin: Is it unusual to take a position where one views the organism as an acron? That's a dynamical activity that produces a pilot wave that maintains its own activity, that maintains its own coherence as an object. Moreover, the life cycle and its lineage are an oscillation bringing the organism into existence and then recreating the conditions for its own origination in the next life cycle. Is that view of an organism part of this lens of the eigenform? Or is that just me?

[27:22] Leo Caves: I don't remember from Von Forster's writing that the notion of the reoccurrence of the eigenform, the dissipation of it and the reoccurrence. That might be there, I don't remember that. But if your cognitive apparatus doesn't undergo any significant restructuring in a given period of time, then it's going to be attuned. If that phenomenon is also of the same characteristic you would expect, that if there was some decoupling at some point, then it might well easily reestablish itself. On some given cognitive plot, we might not notice that it had gone away and come back again.

[28:20] Michael Levin: When you tune into something so that you can perceive it as an object, that's another way of saying I'm going to look at it on a particular frequency where its orientation always looks the same to me. And I'm not going to look at it on the in-between phases where it was doing something else. I imagine it like a spinning object that I'm strobing at the right frequency so that it looks stationary to me.

[28:58] Leo Caves: My initial response to that is there are two different time scales going on here. One is the inherent underlying cognitive processes that we don't control, which give us our experience of our experiences. And the other is the more intentional or attentive mode of how I decide to look at something.

[29:34] Michael Levin: Important, or how something gets my attention, describe it passively.

[29:40] Leo Caves: You start off with, what's the scaling of this view of an entity, a wave-like entity, which may combine with others to form coherent structures. And then one would expect those coherent structures to have their own waveforms.

[30:20] Michael Levin: In a space with a dimensionality that the previous objects couldn't perceive.

[30:26] Leo Caves: Yes, this becomes the part-whole relationship, which is the compositing entities become the larger one, but the larger one then becomes yet another thing in relation to those.

[30:53] Michael Levin: Since we're talking about part wholes, let's talk about harmonics because we've talked about resonance, but we didn't talk about harmonics yet. If you're going to have some parts, they need to have frequencies you can see. And if there's going to be a whole, that needs to have frequencies you can see. If that's not just a coincidence, then the frequencies of the parts need to fit into the frequency of the whole.

[31:22] Leo Caves: To the extent that my nuance would be when there's an interaction with the component parts to build something, of course we can talk about a buildup or we could equally go in another direction. Are they atomic, or in their interaction are both changed? So you're saying, is there preservation of the frequencies of these notional waveforms that start to combine but create a large object with its own harmonic frequency spectrum? I'm not sure that we could necessarily say that it's completely decomposable, that the frequency structure of those independent things is still there. That feels a bit too non-linear.

[32:45] Michael Levin: So that the parts, if we do it bottom up.

[32:49] Leo Caves: Yes.

[32:50] Michael Levin: By convention.

[32:51] Leo Caves: Yes.

[32:52] Michael Levin: But from the sidelines admonishing ourselves for being so conventional about it, if we do it bottom up, then the parts in coming into contact and tuning into each other create a dance of back and forth between the two of them. That is an oscillation at a higher level of organization that they couldn't do on their own, because it's a modulation. You're a modulation to what I'm doing and I'm a modulation to what you're doing. And together that creates something that's a higher-dimensional waveform than the waveform that either of the parts had. In coming into contact in that way, in order for me to see you and for you to see me, we each need to adjust ourselves in such a way that we tune in, that we synchronize with each other. And that means that we've been changed by the contact, which is another way of saying each of us as we were doesn't quite exist anymore, that we've been transformed. But also in the process, we've created something new at a higher-level organization that wasn't there before. Now, one could imagine that it becomes a sort of total homogenization that I don't exist at all anymore, and you don't exist at all anymore. We've just become this greater thing and our previous identities have been obliterated, like two balls of plasticine pushed together and become one ball of plasticine and the old balls are gone. But organisms and cognitive entities, things that have memory and can learn, they're not like that. They're not homogeneous. They don't become homogenized. They don't lose entirely the component structure, but neither is it true that the components retain the entire independence that they had before, if they ever were. Instead, you have something where there's still something going on that maintains the identity of me, and there's still something going on that maintains the identity of you. And there's something going on between us that creates something new at a higher-level organization. And for those three things to be true, the frequencies of my behavior and the frequency of the behavior of the whole must be in harmonic relationship. They must have that musical relation, because otherwise one of the parts would be coherent, but the whole wouldn't, or the whole would be coherent, but the parts wouldn't. An observer wouldn't be able to see both of them if there wasn't a frequency that was common to both of them.

[36:07] Leo Caves: The observer may see the whole, may choose a gestalt figure-ground, or decide to focus on the parts.

[36:24] Michael Levin: Yes.

[36:25] Leo Caves: But they're not seeing both at the same time.

[36:28] Michael Levin: Can I hear an octave? An octave is composed out of F and 2F. If an octave is playing, if that interval is playing, and I resonate to F, then I can see F, I can hear F. If I resonate to 2F, then I can hear 2F; 2F is there as well. But that's not really hearing the octave. In order to hear the octave, I also need to hear the relation between them. The relation between them is the difference tone. The difference tone wasn't there before. Between one frequency and another, the difference tone is a new frequency that wasn't there before. If I can hear the difference tone as well, that's still not really it. What I really need to cognize is that the difference tone tells me about the relationship between the two other frequencies. It tells me about the phase change, the phase relationship between those two frequencies, even though two things at different frequencies don't have a fixed phase difference. It tells me about the nature of the phase relationship between them. Then I'm perceiving the octave. In order to do that, I have to be an octave. The only kind of thing that can perceive F and 2F and the relationship between them is another octave. I have to be a model of the thing that I am perceiving in order to see the whole and its parts and the relationship between them.

[38:04] Leo Caves: That's taking your harmonic analogy, and that's reifying the more generalized version of the coupling between the observer and the observed and what is the form of that coupling, and you're putting a model, you're putting an analogy on top of that: that it's the resonant relationships that take place, that the resonance is the coupling that will bring forth some experience or some knowledge of that interaction; interaction will bring forth something which will be that resonance, that experience. I'm attracted by the resonance as the mode, as a form of coupling, which has lovely generalizability and scale-spanning properties. When you're articulating it through your talks, it was one of those where you get silent because it's something you think, oh, that means when you hear something that resonates. It was in the reflexive loop. One question I might have is, when you were talking about the coherence that would be formed from the resonating entities: so we get the frequency relationships. How are you accommodating the fact that in a complex organism in the biological realm, you could have subsystems or processes that might be decoupled, that might operate at inherent frequencies that might not be in harmonic relationship to a whole? Because if you go with the harmonic resonance all the way through, you're tied to a single set of coherence, which feels like a strong coupling.

[40:53] Michael Levin: Everything has to be in the same key. They're all built from the same fundamental, otherwise they can never make contact with each other.

[41:00] Leo Caves: So what have you got?

[41:02] Michael Levin: I think everything is built from the same fundamental. So in the polycomputing view, the idea that you could have something—two things going on in the same material substrate that are unrelated to one another.

[41:24] Leo Caves: That's a strong term, unrelated to.

[41:27] Michael Levin: Yeah.

[41:29] Leo Caves: I think that's what it means, right?

[41:35] Michael Levin: The only way that two frequencies cannot have any contact with each other mathematically is that they're in an irrational ratio. That means they're going to have contact with each other in the short term, because they might be quite close in frequency. The effect that they have on each other at one moment in time will be the opposite of the effect that they have on each other in another period of time if you wait longer. In the short term, it will appear like they're in contact with each other, but in the long term they will undo each other's effects. The only way their effects don't accumulate over time is if they were in a purely irrational ratio. Does that make sense?

[42:30] Leo Caves: From that harmonic sort of perspective that you're...

[42:35] Michael Levin: I think that the activities of the cells within a multicellular organism can do stuff. This cell can be doing something with some morphogen concentrations and this cell can be doing something with some bioelectric waves. What they're doing appears to be ships in the night that are not communicating with each other at all, to first approximation. But because those cells are made out of the same stuff and the machinery they have inside shares a history, the tuning in one of them has to be related to the tuning in another. They can't really be singing in unrelated keys. Their activities must eventually come back around in a way which accumulates and not just talk past each other forever. But we knew that because that's why we were calling them a multicellular organism and not just some cells.

[43:54] Leo Caves: In your analogy, as an organism, I can't change key in my song, in the song of myself.

[44:04] Michael Levin: So I've given that a lot of thought. It's a question I'm struggling with. I think that configurational changes in the arrangement of notes create different tones that weren't there before in other configurations. And that means that in one configuration, I'm a thing and you can see me, and in another configuration, I'm not a thing and you can't see me, which is to say that the profile of frequencies that I present to the world can change.

[44:54] Leo Caves: But then you need a mode of coupling which brings into play at the appropriate timing.

[45:03] Michael Levin: But that's the relation. The relation is just another frequency.

[45:08] Leo Caves: You're saying that the harmonic, the resonant mode coupling.

[45:14] Michael Levin: Yeah.

[45:18] Leo Caves: I'm attracted by it, but I'm wondering if it's too...

[45:25] Michael Levin: You're afraid of disappearing up your own orifice, aren't you? Don't worry about that.

[45:29] Leo Caves: No, there's too much of a lockstep. My feeling, from my view of a complex biological entity at whatever scale, is that you want to have subsystems that might run at their own tempo. That's not the same thing you were saying about frequency relationships, but be careful in the time domain we're talking about. Not that they can't — you need decisive action. You need things to take place at the right time, even in response to some sort of context that we find ourselves in. So I needed a way of allowing something to operate at its own intrinsic time scale, but when required, not to say that it's completely separate. I'm absolutely buying some overall level of coherence that we're calling the entity. But within that, you want a degree of loose coupling that allows for the quasi-autonomous nature of subsystems to operate, but then can be brought back into the whole when necessary, not always in lockstep to some sort of one fixed harmonic structure. This is my intuition: there might be other modes of coupling that can play a role within. You could say that the hard one is the dominant one; it's very attractive, but is it the whole story to give you the complexity and the features that we're looking for?

[47:28] Michael Levin: 11 extreme is I'm an oscillation that's driven by an internal driver and I can't change my frequency and I do what I do. You do what you do and we don't really have any coupling at all.

[47:43] Leo Caves: You're entrained.

[47:45] Michael Levin: Another possibility is that I'm quite plastic in the way in which I respond to external influences, so that I don't really have a tempo of my own. I'm just more like a shadow than I am like an agent. In between is the territory that's interesting, where I have a tendency to keep a tempo of my own. But my phase, the profile of frequencies that I display to the world, can be influenced by contact with others. But when that contact goes away, I hold onto it. So that it's like a memory. I think that can be done. I'm trying to do it in a way where everything is a nice circular rotation. Maybe it could be done with things that have interrupts and latent periods instead of things which are proper oscillations. I don't know how to say it except a rotational oscillation.

[49:27] Leo Caves: It's interesting that you use the interrupt because I was thinking about a prototypical model that would capture the attrition that I'm pursuing about the need for the autonomous subsystems which are coupled, but not necessarily through a lockstep resonant form. And I thought about a real-time operating system. It's terrible. You always use the analogies of the age that you live in, but where you get interrupts that might bring into play in the time domain and not in the frequency domain, which might give you the occurrence of the submodule calling on it, which might be a time-domain interrupt to bring that in.

[50:16] Michael Levin: Yeah.

[50:17] Leo Caves: And the job of the kernel is to make sure that the necessary operations occur within the appropriate time window so that the overall process, the controller, is controlled.

[50:35] Michael Levin: Right.

[50:36] Leo Caves: Gets to operate properly.

[50:38] Michael Levin: But it needs that latent period as well, which is saying it's a time scale over which it can't be interrupted. So if I can be interrupted at any time, I'm not really a thing. I only do what you tell me to do.

[50:53] Leo Caves: I'm not able to respond on that time scale because it doesn't meet my own natural.

[50:59] Michael Levin: But if I was always responsive, that's a bit like saying I am just you. You just tell me what to do. I'm just a puppet on a string. But if I at least have a latent period, that means that there are durations where I'm sensitive and durations where I'm not. And that's another way of saying there are durations where I'm doing my own thing. And there are durations when I'm being responsive. And that means that I'm bringing my own pattern of attention and not just permanently interruptible cause and effect relationship. Does that make sense?

[51:49] Leo Caves: Yes.

[51:50] Richard Watson: Just to let you know, I'm going to have to go in about two minutes, but I don't want to cut this off. If you're on a roll, I'm just going to leave this up as I go, but it's totally up to you.

[52:01] Michael Levin: Thanks, Mike. How's your timeline, Leo?

[52:09] Leo Caves: I'm flex for a bit, yeah.

[52:11] Michael Levin: Two.

[52:18] Leo Caves: Were you making a point there about one thing: if we go back to this interrupt model, there's some intrinsic timescale of the subprocess that you're calling upon that you need to bring into play. And that will have its own inherent time scale that you need to match or meet in order to get productive. If you hit it too fast, it's not at its natural time scale. It may not even know that you're communicating with it because it doesn't have the capacity to receive that call for action.

[53:08] Michael Levin: I might go so far as to say that if you give it the same instruction but at the wrong time, it would have the opposite effect.

[53:15] Leo Caves: The timing is important. So is that distinction between the frequency-domain coupling, which is dominant in your song approach, and the time domain — oscillating, bringing in the time domain — which might be a way of phase relationship, right? This is when I'm something. But not necessarily. They don't have to be in a resonant relationship. This makes sense. This is the thing that I've been thinking: you might need that extra flexibility to build into the scheme of combining quasi-autonomous but coordinated.

[54:17] Michael Levin: I don't know what to think about it. I feel the same — reluctance, caution. A potential problem you raised is that if everything is a nice smooth rotation, cogs with teeth that intermesh, it doesn't feel like it has enough flexibility for subparts to have their own autonomy.

[55:01] Leo Caves: Surely the subparts are subparts, or any one of the entities that you're postulating, any one of the living entities, they're going to have that quality. So whether you start composing them into some higher-order coordinated body, the same principles of coordination have to be taking place. My feeling is that whatever we're talking about here, we don't escape it; we have to build it in.

[55:35] Michael Levin: Yeah.

[55:36] Leo Caves: And I agree with you. I didn't mean the crude analogy of the operating system and everything. When these systems couple, they're generating things which are not the sum of the parts. It was just an intuition that if there could be something which would allow for looser coupling, because I think that's important.

[56:15] Michael Levin: So when you turn it into an interrupt type system, that gives you the possibility of a linear chain that A happens before B and A causes B to happen.

[56:38] Leo Caves: Doesn't have to be linear.

[56:41] Michael Levin: Doesn't have to be linear.

[56:42] Leo Caves: Say you need a whole bunch of things to happen before something else happens. But the order in which they happen may not be important.

[56:53] Michael Levin: It gives a partial ordering, but not necessarily a linearization. The partial ordering is valuable because you can do things with feed-forward relationships that you can't do with symmetric relationships. But we're going to put this thing, whatever it is, this system of interrupts connected to one another. It has to, at some level of organization or another, become a cycle of activity, and not just a one-shot thing. When it becomes a cycle of activity, the latency periods of those things are going to matter to whether it's able to bite its own tail or not, whether it creates a cycle that's self-sustaining or misses itself and peters out. I have the feeling that it will end up being like the system that we wanted to get away from because the periods of activity and latency will end up meshing in such a way that they create a higher-level sustaining coherent whole, or they won't, in which case it wasn't the kind of thing that we were interested in. You can't really get away from that. It has to be a harmonic and it has to be in contact. But it's not clear to me yet.

[58:38] Leo Caves: I see what you're saying, that if we are compositing the current structures, then each one of those will have some intrinsic timescale where, as a whole, it operates. Now the question is, I accept that, but is it necessary that that recurrence is expressible as a harmonic oscillation? I understand what you mean. There's a closure there.

[59:25] Michael Levin: Yeah.

Leo Caves: We can express that as an oscillation.

[59:28] Michael Levin: Yeah.

[59:29] Leo Caves: What I'm examining for myself is whether necessarily the only way that can couple in then is through the harmonic relationship. That is the root of my question. So the first thing you want to do is break it.

[59:50] Michael Levin: It articulates a concern which has been bubbling for me and I don't quite know how to handle it because it does feel like the activities of living organisms are more like an interrupt-based stimulus-response system rather than cycles within cycles that are all more orderly. I think there might be. I don't know how to resolve it, but you're right that the issue of closure is the thing that forces me away from that and back towards something which is harmonic. But it's not completely satisfactory. The only ***** of light I have there is that one way of imagining it is: how could something look like an interrupt system when it's really not? That would be when the observer matches that kind of structure in an interesting way. I don't know if I can put it together because I don't know if it makes sense or not.

[1:02:13] Leo Caves: What's coming to mind is that you put the observer back into place. We're not just stuck in thinking about this hypothetical entity. I feel like what we're grasping at is, hopefully in words that are congruent with your metaphor, it's like a mode selection.

[1:02:45] Michael Levin: Mode selection. Yeah, indeed. Yeah.

[1:02:48] Leo Caves: How can we get different kinds of operating modes in relation to some different context such that in a particular context where the observed or organism environment, which might be co-determining, co-constituent, is established and sustainable, maybe at that point a harmonic metaphor works because we are in a state, we're in a stable form of becoming at that point. We're reciprocally related. We know how to move and know how to act. If something changes, something about the environmental context, some other entity comes in, there may be a disruption of that congruence and we need to do a mode selection or shift to something else.

[1:04:11] Michael Levin: Yeah.

Leo Caves: To be in a good fit, right?

[1:04:15] Michael Levin: I have the feeling that you could view a system which is purely harmonic, which doesn't have any of that kind of interrupt structure, but you could view it in a way that it looked like it had an interrupt structure, and then it would look like there were events that came out of nowhere with consequences that dissipated away, and then another event happens, and then there would be another cascade of consequences which dissipated away. It would look like a system that wasn't conserved, but that was because we didn't see the whole picture. If we were viewing the same system in a different way, we would see that all of that cascade of innumerable consequences that we were previously calling dissipation were a different part of a cycle which changed the pressure in such a way that it was recreating the conditions for its own origination of the event that started it all. If it wasn't that to some observer, then it wouldn't be a thing. It would just be an event that happened with consequences that dissipated and there's no point trying to make any sense of it because there's nothing to explain about it. You can explain how A equals B and B equals C and C equals D, but that's the boring part. The answer to 'why was it there and why did it do that' has to be because it was always there and it was always doing that. That's the only way to get away from that question. That's because it is all part of one cycle where, although there are many different threads of the melody, there are many different variants on the melody and variations on a theme, and it looks like you get a different one each time, and that they don't have any pattern or sequence to them, that just means you're observing something that's happening in a high dimensional space where you haven't grasped what the pattern is yet. That's another way of saying you and the thing that you're looking at are not quite tuned in with each other, not quite synchronized, and you're drifting. Some parts of what I'm looking at in the system are coherent, but not everything that I'm looking at in the system is coherent, and so they're drifting in a way that doesn't appear to be stationary. Maybe.

[1:07:40] Leo Caves: It's made me think of a book of Voralis called "Ethical Know-How," and in that he introduced the notion of micro-worlds and micro-identities. From this notion of reciprocal creation of an entity in its environment. The relation to ethics is that if we're in a given situation, a given context, and we are attuned to it, then every action is a right action. Every action is spontaneous and will be right or good in an ethical context. There's a match between the micro-identity and what it terms the micro-world. My memory serves me correctly. If there is a shift in context to another micro-world, you don't know how to act because you may not be attuned to it. It might be one that you visited and established a relationship with before, but it may be presented in a new context. So you need to reconfigure or choose a course of action, but you don't know because you're not tuned yet. You don't know if that's the right thing to do. And that would key into your notions of learning. You have a multiplicity of identities and micro-worlds, and the micro-worlds will key in. Extending from your metaphors that we are songs: I'm thinking, are we song books? Or a song cycle? Song cycle sounds too strong and a bit linear, but you've got the idea that there's a repertoire that I can bring forth in relation to a particular context, with the context being a relationship. And that doesn't really address the question of what is the underlying sort of — so what I'm talking about, micro-density could be the mode switch that we're talking about. I'm also thinking we don't have to choose between anything because both ends might be appropriate; there might be both of these things taking place at the same time, and that's not a weakness. That's giving us more options available, which is another principle to operate on in the future.

[1:11:16] Michael Levin: That would raise the question, what's at stake by choosing between those views? What do you lose by taking one view over the other? Potentially, what do you lose by taking a both-and position? Is there something that you get from just deciding one way or the other that you don't get from a multiplicity? I think that being able to learn causal models as opposed to correlational models would require you to be a causal kind of system as opposed to a correlational kind of system. One thing that I might lose by having the harmonic view is there is no portfolio or catalog of songs that I choose as appropriate for a particular environment. There's just one big song. In the environment that I'm in, which is one big song, my song plays out with all of the variations on the theme, responding to the variations on the theme of the environment in an essentially determined way that eventually will cycle around when the period of me and the period in the environment reaches whatever their lowest common denominator is, even though that might be very long. The thing that I might lose by doing that is that I might lose a truly causal model of what's going on in the world. Because when you reduce everything to one big song, it's like saying you describe everything in terms of one big wave. In the limit, it doesn't really have any time. It's just a shape. It's interminable. It doesn't have a beginning or an end. It doesn't change over time. It only changes over time because there is only time and there are only events with causes and effects to the extent that I haven't seen the whole picture yet.

[1:14:07] Michael Levin: If I go to that limit of everything's one big song, then I don't have any time and I don't have any events and I don't have any A causes B. I don't have any true causal structure. It's just a geometry. I think I like that though. Going the other way, if you allow it, it's not just one big song; there's a whole bunch of things going on and A causes B and the rest of it. In this context, I do that song and in this context, I do that song. Then I have a causal model of the world where stuff happens and stuff responds to what happens. There's a difference between A causes B and B causes A and it's not just a geometry. But the price of going in that direction is that there isn't an overall sense to it. There isn't an overall symmetry to it. There isn't an overall reason for the prime mover, that there would be a prime mover which doesn't have any reason, which erupts into the world, to use Tom Froze's language. I think I'm coming around to both ends, actually. Both of those extremes aren't where the juice is; the juicy bit is in between somewhere. I had a couple of questions in my notes here from earlier that I'd like to follow up on whilst I remember what they mean.

[1:16:58] Leo Caves: Yes.

[1:17:02] Michael Levin: One of them was the process ontology versus material ontology. That's always struck me as uncomfortably dualist, that there's only two kinds of things. There are processes and there are materials. Am I right in thinking that the kind of territory we're thinking about now is much more recursive than that? It's much more that something which is a wave to me is a particle to you. And within that there are waves which are containing, and which are particles to a different frequency. It's not like there's an achron with its theta waves, and that's the truth. The achron is made of subacrons and the theta wave is part of an assembly which makes a greater achron and so on. Is that a normal way of thinking about things?

[1:18:13] Leo Caves: So to start with, I think that that notion that there's a substance ontology or some relational ontology or process ontology is that, in giving context back to our observer-observed, it could be that in a given context an as-if approach might work: you can treat something as if it's this or as if it's that, and depending on your purposes and the context, that might be sufficient for you to do the work you want to do. So again, it might not be something that you want to tie yourself to an underlying metaphysical reality; there are these working models that might be appropriate or not. That's the way I look.

[1:19:21] Michael Levin: So I'm saying that the working model that I think is appropriate doesn't like the idea. But I'm fine with saying that it isn't necessary to choose, it isn't necessary to choose purely a process ontology or purely a substance ontology. It's fine that it will be useful to treat it as if it's one in some circumstances, and it will be useful to treat it as if it's the other in other circumstances. My question is more about the sort of recursive, nested nature of all of this, that even if I were to treat things as substance. I will inevitably end up in territory where there are weird relational things going on. And if I started in a relational way, I want to inevitably end up in a territory where there are weird substance-like things going on. And the whole thing is nested and recursive and fractal and not just a duality between substances and their relations.

[1:20:39] Leo Caves: Well, I think you're getting at the nature of language in the sense that there's a recursive interaction taking place about which we may stabilize on something we are coordinating about. We may give that a token, and we may agree with what that means. And that would be a notion of recursively coming to some coordination about what the entity is that we are creating in that exchange. To me, I'm very sympathetic to having a relational process worldview. I find that in my language, the author, how do you express that without dropping always into subject or predicate language, because that's the way we structure it. And there are different languages which have different sorts.

[1:22:01] Michael Levin: Yes, I can't help but say, "wait, is this a relation or not? How many processes do I have, et cetera?" Imagine that we start off with some thingies, and we start talking about the relations between them and notice there's this relation and there's that relation. Now the relationships are things. How does this way of relating interact with that way of relating? We're just talking about a meta-relationship. But there's a way of short-circuiting, which seems initially counterproductive. You say there's the things and there's the relationships between them. And then you make the relationships into the things and you say, what are the relationships between the things? It's the things that we used to call objects. Those are the relationships between the relations. One relation is related to another relation by the objects they relate on. You're just turning the nodes and the edges of the graph back and forth.

[1:23:19] Leo Caves: Graph, yes.

[1:23:20] Michael Levin: Now, if you do that immediately, if you go from things to relations, then you call the relations things and the things that were previously things become the relations. If you do that reversal too quickly, with just one step forwards and one step back, it seems to have undone everything that was interesting about the distinction you were making in the first place. But more interesting is when you go through several levels of the hierarchies from things to relations, to meta relations, to meta-meta relations. And then it becomes a thing. Then it folds back on itself after many levels of recursion. And that gives you space for more interesting **** to happen. That's the space of possibilities where we can talk about systems, not just bags of particles and not just fields, but we can talk about systems which have organizational structure and which change over time.

[1:25:05] Leo Caves: It's OK.

[1:25:10] Michael Levin: If you have those multiple levels of meta, of multiple levels of relating. It creates a configurational space where you can have interesting systems. But often in computational ways of thinking about things, that spins off to infinity in a way that never comes back. It doesn't have any closure. The tricky but interesting thing, the quality of biological systems is that they push out into this. They create a space of dimensionality where interesting configurations can happen. And yet they are also somehow contained. They have a closure that folds back so that things that look like they were linear chains don't go on forever, and they turn out to be cycles.

[1:26:15] Leo Caves: And then things that are nominally very high dimensional actually behave in very low dimensional ways.

[1:26:25] Michael Levin: It just becomes an aggregate. Dennis Noble talks about the boundary conditions. The constraints that come from below look like material properties, and the constraints that come from above look like universal constants.

[1:26:48] Leo Caves: Right.

Michael Levin: But material properties and universal constants are actually connected to each other. There's a constraint on how those things are related: you can't have these material properties if you didn't have those universal constants. I'm going to mention somebody else. In your e-mail, you mentioned Clifford. Is that Clifford as in Clifford algebra?

[1:27:30] Leo Caves: I believe so, yes.

[1:27:34] Michael Levin: The plot thickens because when I came across Clifford Algebra, I was introduced to it by a beautiful, geometric, intuitive presentation from a guy called Steve Lecha, who also thinks about everything in terms of resonance. In Clifford algebra, every shape or form can be transformed into any other shape or form in the space of possible things within a particular geometry through some kind of rotation. So there isn't really any configurational change. Everything is a rotation. And any rotation that you can do, if you do it enough, it gets you back to where you started. Otherwise, it wouldn't be a rotation. There's this notion of the pseudoscalar where the dimensionality of the form is pushed to the overflow of the maximum grade that you pre-decided you were going to use. When that overflow happens, it becomes essentially a scalar again, only a one-dimensional thing. So the Clifford algebra territory felt to me still has the right kind of mathematical structure to deal with the kinds of things that we want to talk about. You were saying that the very high dimensional becomes low dimensional again. I think that's the mathematical territory where Clifford algebra gives you that.

Conversation with Richard Watson and Leo Caves 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

Conversation with Richard Watson and Leo Caves 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