Watch Episode Here
Listen to Episode Here
Show Notes
This is a ~1 hour conversation between Alexey Tolchinsky (https://www.psychologytoday.com/us/contributors/alexey-tolchinsky-psyd ), a clinical psychologist and Adjunct Professor at The George Washington University, Center for Professional Psychology, and Thomas Pollak (https://orcid.org/0000-0002-6171-0810), a neuropsychiatrist and researcher working at the Institute of Psychiatry, Psychology and Neuroscience at King's College London, about issues of collective minds, psychiatry, and cognition. An interesting email exchange that preceded this discussion is at https://thoughtforms.life/if-mind-is-everywhere-where-are-all-the-panpsychiatrists-a-neuropsychiatry-focussed-discussion/ .
CHAPTERS:
(00:00) Clinical backgrounds and FND
(01:47) Thoughts, aging, dementia
(07:45) Thought traps and niche
(11:34) Voices and avatar therapy
(16:26) Affect, voices, confabulation
(22:59) Subagents and therapeutic alliance
(30:46) Regeneration, biodomes, anthrobots
(41:44) Collective intelligence and self
(50:45) DID, alignment, real estate
PRODUCED BY:
SOCIAL LINKS:
Podcast Website: https://thoughtforms-life.aipodcast.ing
YouTube: https://www.youtube.com/channel/UC3pVafx6EZqXVI2V_Efu2uw
Apple Podcasts: https://podcasts.apple.com/us/podcast/thoughtforms-life/id1805908099
Spotify: https://open.spotify.com/show/7JCmtoeH53neYyZeOZ6ym5
Twitter: https://x.com/drmichaellevin
Blog: https://thoughtforms.life
The Levin Lab: https://drmichaellevin.org
Transcript
This transcript is automatically generated; we strive for accuracy, but errors in wording or speaker identification may occur. Please verify key details when needed.
[00:00] Alexey Tolchinsky: I'm a clinical psychologist in Maryland. I work with outpatient requests. I'm doing psychotherapy primarily and I also write papers and Michael was kind enough to collaborate on a few of those. I have a lot of interest in Michael's work and integration of it with clinical psychology.
[00:25] Thomas Pollak: I'm a neuropsychiatrist working in London. I'm also a researcher at King's College London. I research immunopsychiatry, the relationship between the immune system and mental illness, mainly focusing on psychosis and autoimmunity. Clinically, as a neuropsychiatrist, I'm seeing patients with neurological disorders who also present psychiatrically. Those categories are problematic. A lot of the patients I see have brain inflammation and immune problems with the brain. It's interesting to see what happens to them. The other patient group are patients with a diagnosis called functional neurological disorder. These are patients who have symptoms that look very much like the symptoms we see in neurology and neurosurgery, but there's no structural organic cause we can find. There is an idea that there's a top-down etiology happening. It touches on some of the nice ideas about dissociation of placebo that I know you've written about.
[01:47] Michael Levin: In particular, I want to get your views on this potential distinction between organic disease and "psychological" disease. There's a concept that I think about a lot, which is this notion of the thought that breaks the thinker. These thought patterns — there's nothing wrong with the system. It isn't broken, it isn't damaged. That also attaches to some of the things we've been doing in the aging field recently. The typical theories of aging are either that it's an accumulation of noise and damage over time, or that it's evolutionarily programmed somehow. But there's also a cognitive perspective on things where certain things can happen and go wrong, not because of any damage or any problem, but as a result of processing certain inputs. What kind of inputs can trigger cognitive dynamics that then cause significant problems? And maybe eventually they do filter down into the physical. That's the stuff I was hoping we'd talk about today.
[03:03] Thomas Pollak: Certainly what you said about the aging process and the cognitive problems resonates a lot with a group of patients that we've started seeing an awful lot recently. Why we didn't see them as much in the past but are seeing them more is one of those mysteries. These people don't have to be old, though often they are in the age to start developing dementia, and they're complaining of cognitive symptoms. There are no neuropsychological deficits when you test them. They don't have MRI scans suggestive of dementia. Even if you manage to get a lumbar puncture and look for those markers, they don't have them. Some of these people are framed as having functional cognitive disorder, a variant of the functional neurological disorder I was just talking about. One way of thinking about it is that the automaticity with which we normally think becomes disrupted by a misallocation of attention. A typical case is when someone meets a person they know very well—a friend or someone in a shop—and suddenly the name goes out of their head. At this point we're beginning to impute what's happening: there's some reaction, maybe emotional—"Oh God, what does this portend? Does this mean my memory is getting worse? Am I getting dementia?" That causes a reallocation of attentional resources away from the conversation, so more gets forgotten and you get a runaway feedback loop that in a short space of time can take over someone's life to the extent that they become convinced they have a degenerative dementing condition. We think the causal dynamics are similar to those in other functional neurological disorders. Actually it's like what you see in the yips in sport or music. It's the loss of automaticity that comes from hyper-focus on something previously over-learned, and then it breaks down. It's like when I was young and trying to walk past a pretty girl at school and suddenly I'm paying attention to the way my legs are working and I can't do it anymore. It's that sort of thing.
[06:05] Alexey Tolchinsky: I had a quick comment about one of the aspects you've mentioned, Thomas, about dementia non-dementia. Sometimes it's a hard question to answer and we need repeated neuropsychological evaluations a year apart, and imaging data to investigate whether there is or there isn't dementia. If we take it a step further into the progression of this condition, one of the viewpoints is that what we see is disintegration. Proponents of the global workspace theory say that you have this unity of consciousness as the sign of health. I just wanted to bring it up because we talk a lot about Michael's viewpoint that all intelligences are collective. I wanted to mention a couple of things in the DSM where it bumps very strongly into prevalent views. In dementia specifically, you start seeing fragments, these islands. I'm quoting Lionel Nakash, a neurologist, and other colleagues. This is also an interesting viewpoint. If we just take one aspect of this conversation of integration and disintegration at different levels from different perspectives in health and pathology, I think that may be a useful starting point.
[07:20] Thomas Pollak: Absolutely. Mike, this idea of the thought that breaks the thinker, I don't know whether the example I was giving is getting at what you were thinking of. I know you've spoken about the way that obsessions can build a niche. I suspect there's a similar thing going on in that example. Is that what you were getting at?
[07:45] Michael Levin: There's two things, both of which I would love to hear you guys talk about. One is what I think you were saying, the dynamics that takes place mostly at the psychological level, which is that you've had a thought and that leads you to a train of other thoughts, which become maladaptive but also difficult to leave. Some very high IQ individuals can get trapped into this stuff where they encounter some sort of existential crisis. It's really quite debilitating for some people where it wouldn't be for others who aren't that deep thinkers. But some people become trapped in this whirlwind, or they hear these Humean ideas. I've actually seen it when I was homeschooling my kid in philosophy; I realized real quick that you have to be very careful with these things because they sometimes hear these ideas from Hume and Kierkegaard. If you're smart enough, it really screws up your whole thing. Climbing out of that can be hard. The other thing I'm really interested in is niche construction. Once you've had certain kinds of thoughts and they start autocatalyzing themselves, do they in fact, and how much do they affect the underlying substrate? Do they actually induce measurable changes in the neural substrate, whatever metric you would use, and do they actually make those kinds of organic changes, which then of course will facilitate their own construction? I'm interested in that from the perspective of some of our work on thoughts as thinkers. This idea that it's the patterns that can be the agents, and trying to look at some of these things from the perspective of thought patterns and memories that maybe have a drive to either persist or to change their environment, or what they're actually driven to. These are the kinds of things I've been thinking about.
[09:47] Thomas Pollak: And in that example, does changing the environment or that niche construction, you're talking about it working on the substrate being the brain, but is there another perspective where the substrate is the behavioral space? So with the guy with the functional cognitive disorder, the behavioral space changes because all of a sudden they're not going to parties, they're not going to that shop, and with somebody with an obsession, the behavioral repertoire changes so that it's constructed a behavioral niche, and that may be the more obvious niche construction that happens before there's a bit of meat that gets changed, or would that still make sense?
[10:29] Michael Levin: All of that is, I think, part of it. In general, I've been playing a lot with trying to map differently the Turing machine metaphor. You've got the hardware and then there's some information patterns that it processes and the data is passive and then the machine is what does the work. We've been doing a lot of thinking about how to flip that and see that from other perspectives. In our work on the developmental side, is it that the body has bioelectric memories that guide morphogenesis, or conversely, are the bioelectric patterns themselves the agent, and the body's a scratch pad. So everything that happens in the gene expression space, in the cell, cell behavior, histologically, it's a scratch pad of what's actually what's primary, which is the working out of these patterns in their own dynamics. So I think figuring out what's the niche and what's the agent is very much not obvious here at all.
[11:34] Thomas Pollak: In some of our disorders, even more impressively than in the ones we were talking about, in schizophrenia and psychosis, the other agent that is perceived as being external at times takes full control over the body. It's doing stuff that even in real time the individual doesn't want being done. Their arm is being moved against their will or they're acting in a way that they know is potentially harmful to them. There have been these great examples. In psychiatry, we can be very deflationary and talk about that agent as a fiction. The point of psychiatry is to bash that perspective out of the patient with medications or therapy. But I think there are other approaches that are more consistent with the way you might be thinking about things: what happens if you engage with it relationally as an agent? At my university there's a trial of something called avatar therapy, where people who have these distressing voices that say awful things to the patient take control of them and have all this psychological baggage with it. The patient sits down with a therapist and they construct an avatar on a computer screen that has the same voice as the voice they hear and looks the same. The psychologist takes control in a very organized and manualized way over six or eight sessions, and the patient enters into relation with this voice. Frequently there's a slightly nasty dynamic that reflects the actual dynamic of the heard voice initially, but then it can get a little less assertive. The avatar starts to question itself and the power dynamics are subtly encouraged to change. The results have been amazing. Of course, there haven't been big randomized controlled trials, but they're already thinking about taking it into OCD and eating disorders. There's been a small study where the eating disorder voice gets externalized as an avatar. I think what's so interesting there is it's essentially an act of imagination. It's not like there's something so creative that you're giving birth to this pseudo agent. And yet, if you act with it like it's a real agent, then it does the job anyway. There's something that I find very fascinating but confusing: that just acting as if seems to be enough in some cases for the therapeutic efficacy to get there. I don't know whether or how that happens.
[14:55] Michael Levin: Amazing. That makes me think of two things. One is there are these ancient traditions where thought forms are a thing, and not only that, but you can bind the thought forms to specific objects. Now, back in the day they didn't have AI avatars, so these would be much more passive objects or dolls. There is that whole thing. The other thing this reminds me of is the therapy for phantom limb syndrome. Have you seen this stuff? You're providing an avatar for the limb that isn't there. You're unclenching through the mirror; by mirroring the other hand, you can unclench the one that's missing. That eventually becomes the new reality. I think that's extremely interesting. Offloading — that's a new and very useful use for AI: VR and avatars as a way to improve our dialogue with systems. It matches nicely our attempted use of AI now as a translator tool to other intelligences that you wouldn't be able to see. We're working on communicating with your organs and with cells, which would normally be much harder to do, but these kinds of technologies are good for that.
[16:26] Alexey Tolchinsky: I was thinking about the problem when it's posed as the thought that breaks the thinker. Part of me wants to say that what we're doing here in terms of modeling is a modular approach, where we break functions into modules. These are thoughts and these are other things. I think that in a systems approach, the thought is never alone. There is affect, there is generalized arousal, there is biology and all of these other things. I think thought is a low-hanging fruit. We're very aware of it in our conscious minds and thoughts are stabilized in working memory. "I want an apple" is a series of three things that I strung together and I maintain it in the working memory in this global workspace. But while I am having it, I have a thing happening in my body, perhaps a low blood sugar. It's a process, it's dynamic, it's up and down. I think that if we take the extreme example of suicidal thought that results in irreversible, catastrophic destruction of a body, it is also not the thought that kills. There's hopelessness, which is the highest correlation with completed suicides. There is helplessness. Nobody can do anything. There is history and memories that drive the predictions forward. They're all black and negative predictions. There's the mood tone. Part of me wants to think that if we model these things, we certainly need to include affects which are structured differently and work differently and are sometimes wordless and homeostatic, more similar than the thought.
[17:56] Michael Levin: I have a weird question. For these voices, as you guys mentioned, that tend to be very negative and destructive, is there ever a positive version of this in the clinical literature? Does anybody say, "I've got this voice and it's amazing. It builds me up, it tells me how great I am, it has good ideas, tells me to do stuff I otherwise was afraid to do, amazing"? Does that ever happen or is it always a negative thing?
[18:22] Thomas Pollak: I think we're just beginning to appreciate how common it is. I think as you'd expect that there's such a continuity to some extent. I get in trouble with colleagues if I say that it's purely dimensional. There are people who hear voices who only experience positive things being said. There are people who once experienced very negative things and who had the right treatment or one day things got better. I see that on a continuity, for example, I'm one of the 10% of people who don't hear thoughts verbally. I have them as images. I assume that there's probably no sharp discontinuity between somebody having verbal thoughts, I'm hungry, I need to go to the gym, and then eventually someone hearing voices, because you can hear first person voices that feel a little like my own, but maybe not. I'm sure that no matter where you look, you will never be able to find a point where there is a sharp discontinuity there. In recent years, there are things like the Hearing Voices movement, which is a powerful and lovely movement, trying to reclaim these experiences away from the medical establishment.
[19:52] Alexey Tolchinsky: Just to add nuance to what Thomas just said, I think that the positive versus negative voice is the affect. It's not the thought. The thought in and of itself is neutral. It has no valence. But in psychosis literature, and I'm quoting Mark Salmes and others, we often see the direction of wishful thinking. For example, a person with Korsakoff syndrome who cannot retrieve memories or find them fluently and confabulates tends to confabulate in the direction of what he wants. When he doesn't remember what happened yesterday, he's not going to say he was working out in the gym. He says, "I saw my brother John and we were just gossiping and having beers." He will tell you in the direction of where he wants to go. Oftentimes we see the same phenomenon in dreams and other things, but this is where the feeling of the gap happens in the direction of what's missing and what the person wants sometimes.
[20:45] Michael Levin: In the cases that you just mentioned where the negative voices become over time more positive, do you see that as providing therapy to the voice itself? Or has the voice gotten better because you've resolved its issues? Is that a way to think about it?
[21:08] Thomas Pollak: I would say that most people in my field really don't engage with the voices or the characters behind the voices in any way at all. I think probably in psychological therapies, there might be a little bit more of that, but I imagine probably not too much either. No, they are seen really as symptoms of an illness. My feeling is that if one day someone's voice is no longer heard as the devil and it's now heard as God and they're saying good things, that's probably interpreted by most people in the profession as a migraine that suddenly got a bit more pleasant; it's just a symptom changing. I don't think that's the right way to think about it. From the clinical perspective, as soon as the voice gets good, it's not your problem anymore.
[22:06] Alexey Tolchinsky: I agree. I think we don't work with a symptom necessarily. In therapy specifically, we're trying to work with a patient. A general principle in psychiatry is "don't argue with a delusion." We're not trying to talk him out of it because when you do, it becomes more entrenched. But we try to make sense of the symptom. We try to come up with a hypothesis where what's happening makes some sense. And oftentimes people come into therapy with an explanatory gap, such as they're saying, "I feel badly, but I don't exactly know what it is and how it came about," or "I'm doing this thing. I keep on doing it over and over, and I know I shouldn't be doing it, but I'm doing it anyway." And part of the work initially is to fill in this gap with the hypothesis to make sense of what's happening, why it's happening, and then to do certain things together over a long period of time. But typically, I don't think the voice itself is targeted. I agree with Thomas.
[22:59] Michael Levin: There's a distinction between arguing with the patient about the delusion versus saying to what extent, how much agency does the delusion itself have? It could be very low, but probably not as high as the patient, but it might not be 0. Normally we focus on the physical machine and we think that if we fix the machine, the information patterns will get better. There might be scenarios where the driving force is the pattern itself and that what your target is not the physical substrate, it's the pattern, whether that be a pattern of stress, an incorrect anatomical bioelectrical pre-pattern, or memory. You're trying to engage with it. These are all empirical questions, but I don't know to what extent treating them as having less reality or as being mechanical and not mini thinkers is the top of what we can do. There might be scenarios where these things are doing their own goal-directed computation and there might be payoff, at least on the biomedical side. On the psychological side, there might be payoff in engaging with them and their needs and the stressors that are driving that pattern. That's distinct from what the larger system is experiencing.
[24:37] Thomas Pollak: That's what got me so excited about some of what you've been writing. You used the phrase "the space of possible minds." At some point the phrase "the space of possible patients" came to me. If we're treating, either we treat people or we treat their neurotransmitter receptors. There's a whole lot in between that could be potential targets. In this avatar therapy example, you've got to give it a little boost. You've got to do animistic magic to make these agents solid enough to interact with. There's this whole set of practices or possibilities that involve granting a degree of, or boosting the autonomy or the reality or the solidity of these sub-agents so that you can then enter into a relationship with them. There are all kinds of ways that one could do that, but they all sound woo-woo. I don't think medicine has a grasp on it yet. I'm not sure what the analogs of that would look like in more basic biomedicine.
[26:04] Michael Levin: One discipline where that kind of thing is completely normal, and that's computer science. Because in computer science, we are very comfortable with having interactions with systems that are not actually physical. You've got an algorithm, the algorithm that's running will have memories, it may have goals, and you can communicate with it, you can provide input, and somebody could say, look, there is no algorithm, there's just silicon and copper and electrons right here. This is a fiction. But we all know in computer science that kind of reduction doesn't actually get you anywhere, that treating the algorithm as the real thing really does make the electron stance important. And so we're okay with then focusing on managing these patterns. Some people obviously work in hardware, but many people don't. At least on the biomedical side, I think we can have models in which we treat the actual cellular substrate as the memory, as the tape. That's the scratch pad. What's really driving it is an algorithm embedded in some kind of information dynamic; it might be some sort of active inference thing, it might be some other thing. But I do think there is a path there that is not woo. We have conceptual and practical tools at the bench to start looking at some of these things. Because of the physiological imaging that's possible now, we can see calcium waves, metabolic states, and bioelectric properties; we can visualize them in an individual and over time. They move around, they do things, they're causal. They spread across bodies. We've published something and have a lot more coming where you have multiple frog embryos in the dish and you can watch these patterns move between them. They belong to the group, they don't belong to the individual. And so I think there is some kind of a path there. How much of that is going to make its way into what you guys do? I have no idea.
[28:18] Alexey Tolchinsky: I think, Michael, what you are focused on, and this is very useful, is the active phase of treatment. You bring up the computer science metaphor, let's say the API, the interface, at what level of the hierarchy do you intervene at the very top, at the very bottom, mid-tier, and what's the interface and how, whether it's bioelectric interface or something. The difference is that in trauma therapy with dissociations, we usually have a phased approach where if the patient is dead, it doesn't matter what technique I use. So the first thing we do is stabilization. What I'm trying to say is the API is the relationship with the clinician. Always. We have a vast amount of literature that without it, nothing happens, no matter what avatars or AIs or psychedelics people use, because they cannot think, because they feel. They're flooded by affect and they can't take advantage of any communication with any interface. One goal is reduction of suicidality, reduction of non-suicidal self-harm behaviors, substance use, basic grounding. After that, we move to perhaps trauma therapy if we see trauma. We need a model of the condition: what is the etiological cause of the symptoms. We don't target the voice or the dissociation; we target the trauma. Then there's a phase of treatment, which I think has to do with what you mentioned, reintegration, where we generally move back toward that unity, the mental, the subjective unity. At that point, there's possibilities for all these things and also in trauma therapy. I just wanted to say that this is the pathway we usually go through. I don't know any way around it. Even in pharmacological psychiatry, we have this interesting thing called the "psychiatrist effect," where the exact same medication and exact same dose is prescribed by two different psychiatrists, and one has higher efficacy numbers because it is human interaction. This human interaction has biology: the pattern system gets activated. We're trying to lead the patient from disorganized, feeling insecure in a relationship, to learn secure attachment. Then from there we move forward. We have to build that foundation first. We never just start with "let's do trauma treatment" on the first or second session.
[30:28] Michael Levin: Is that the same or related to this idea of the therapeutic alliance?
[30:36] Alexey Tolchinsky: Which in and of itself doesn't heal, but it's a necessary condition. Without it, nothing happens. We need to build it first, and that's usually the first stage of therapies.
[30:46] Michael Levin: I think a lot about what that looks like in the regenerative medicine space. So what's the somatic version of this? How do you— one simple way to imagine it is we've got these experiments where we're working on limb regeneration, and there's this wearable bioreactor that fits onto the amputation wound, and then you deliver the payload, whatever that might be. One of the things that I think that does is that the biodome itself, without any drugs in it, actually has some positive effect, not as much as the combination, but it has some. The idea might be that what is happening is that it convinces the cells that there is hope to regenerate. It's a controlled environment. The compounds and everything they're making aren't just disappearing off into the infinite bath; they actually have some control. So they have some agency over the environment. They can tell because you have this autocrine signaling: you release something and there it is. It's almost like step one is to convince the cells and tissues that this is a safe environment, that you're going to be able to complete this task. That's not enough, but it's a crucial part. So that's kind of a wacky way to think about it. But I think it might have important analogies in other aspects of the regenerative notions.
[32:12] Alexey Tolchinsky: Exactly right. I think what you're doing there is you're telling the cell collective or tissue, for example, that I speak your language and it's okay to talk to me. And if you use a different API, a different interface, who are you? You're not me. I'm not going to talk. I'm not going to engage.
[32:34] Thomas Pollak: Mike, you've spoken about your anthrobots. And the notion that in the future they could be used therapeutically as non-immunogenic agents that you can put inside the organism and they can do their thing. I keep trying to see other psychological analogies. It gets a bit weird. But is there something that's made out of the same psychic DNA as us that is much smaller and more stupid, but acting when you've got loads of them, when they're in there, they can do a good job. Which, because it's made from the same stuff as you, doesn't elicit that rejection. And in a way, part of the therapeutic alliance that I guess Alexia is talking about is a smoothing over of that possibility of rejection. The number of times the first time I meet a patient, things are pretty rough. And then over the weeks and months, they might smooth down. I run the risk of caricaturing things because what Alexia does is so subtle and nuanced and I'm coming at it with all these cell analogies and the like. It seems weird that we can think of anthrobots made of cells, but we can't think of similar things made of thoughts or other stuff. At least none that we can use in a systematically effective way.
[34:23] Michael Levin: I think that's exactly right. There are more conventional, practical uses of these anthrobots, as they're delivering some specific reparative intervention. But the bigger picture is that they're vehicles for transplanting behavior patterns or memories. Transplantation of memories is a thing that people have done in various animal models. We've done it too in morphogenesis. Oftentimes it's a very small piece of tissue that you transplant and it somehow takes over the whole thing. In cases where we've done transplants of pieces of tissue from two-headed worms into one-headed worms, or pieces of tissue into tadpoles that otherwise would have had a severe brain defect, in many of these cases you get conversion to the new pattern — either normalization or the two heads carried over. Why? I only introduced a few cells. Why don't they get convinced by the rest of the environment to do that? It's very non-linear. You can't really tell; it's not the size of the tissue that matters. It's how compelling your message is — the signaling you're doing and how likely it is to align the bigger system you're in. We see this kind of battle. In an early frog embryo, if you inject some cells with a particular potassium channel RNA, it makes a voltage equivalent to the voltage spot that sets up eye formation, so you have a few cells that are saying to their neighbors, let's make an eye. This is a message that goes outwards because there aren't enough of them to do it themselves. In cases where you do get an eye, you can section it and see that there are only very few cells that you injected. They convinced all their neighbors to participate. There was a secondary induction effect. You didn't touch the neighbors; they convinced them. Typically there's a battle because those cells are signaling to their neighbors, let's make an eye. Their neighbors, meanwhile, are doing their normal tumor suppression and saying, no, your voltage is crazy; you should be skin. They fight and try to convince each other. In some cases you get the eye. In some cases you get nothing because the skin or the gut cells won and said you're wrong. That's a very basic thing that bodies do to suppress carcinogenic transformations all the time. I think it's quite interesting how you can have these things as carriers or vehicles of information. What's critical is whether that story will take hold, become convincing, and cause alignment within the larger system, particularly with behavior. On the behavior side, these anthrobots come from donated patient tracheal epithelia. You get a sheet of information on the patient. One thing I've wanted to do is get some from smokers and find out if my anthrobots are nicotine addicted. Do the anthrobots that come from those cells actually pursue nicotine? If I were to take those anthrobots and put them into an immunosuppressed mouse, would the mouse inherit some of the propensities? Can you actually pass behavioral propensities that way?
[38:11] Thomas Pollak: Wow.
[38:12] Michael Levin: We haven't done it yet, so I don't know how that works.
[38:16] Thomas Pollak: Have you considered getting the cells from, say, patients with schizophrenia or with a severe self-disorder, because of the polygenic risk for these disorders? We know for a start that it's associated with physical developmental abnormalities as well as psychological abnormalities. But you could almost flip that on its head and say, what's the social behaviour of these anthrops who are derived from patient cells?
[38:49] Michael Levin: If you have access to tissues, we should talk about the collaboration. But yes, we thought about making that. The next thing we're developing is a bunch of more subtle behavioral tests with these things, because so far all we've been watching is just native movement and they have some specific behaviors and we've characterized all that. What are your preferences? What can you learn? What kind of memories can you form? What are you measuring in the environment? Once we understand a little bit about what the assays should be, then we can start screening through patient samples and find out if A, does it reflect current state? Is it predictive of future state? Is it something that you can get out of it as an avatar? Another thing we're thinking of is these things as screening avatars for drugs. When you have multiple options — should you be on this antidepressant or that one or something else — you might be able to screen that on the bots and actually find out that, for this patient, these other three things didn't work, but this one works. Something like that.
[39:56] Thomas Pollak: Yeah. That's wild. That's amazing.
[40:01] Alexey Tolchinsky: I was thinking and listening to you both and I find myself a little disoriented because I'm trying to figure out what level of the hierarchy we are talking at when we're talking anthropots. It seems like mid-tier. We're not at the molecular level. We're not at the very top. If we take your paper on cancer, Michael, I think that you have built the first principles. You have built a high-level theory, which is TAME. And then you have the therapeutic experiments where you're saying, I do this, I restore the communications with the neighbors and the propagation of cancer stops. But in the middle, you have a model of a condition of a specific cancer and specific environment. You're saying one aspect of this cancer is that this cancer cell stopped communicating with the neighbors. That's a model. I'm missing that with application of your work to clinical psychology. This is where we need to put meat on the bones. It's just saying, what is our model? Using TAME is one of the principles; using your work on schizophrenia is very complex. But even if we take one symptom that we try to do — dissociative symptoms, obsessive symptoms and other things — until then, it's very hard to go straight to therapeutics because we go from first principles to what to do. If we take physics as a metaphor, the principles are prescriptive. They're a method of how to do some things. One level down is mechanics. That's the model, which is Newton's laws. One way down from that is the dynamics of Kepler, or how the bodies, the planets move. But it's hard to go from the first principle straight to the therapeutics or what to do until we have the more nuanced understanding of the specific aspect of a specific disorder.
[41:44] Michael Levin: I think that's exactly right. One of the things that we're writing next is a follow-up to a paper that Patrick McMillan in my group and I did on aspects of morphogenesis that are reflections of collective intelligence. Here are the things in development that actually look like intelligence and these proto-cognitive capabilities. What I want to do now is the flip side and do the disorders. I want to take a look at what are all the disorders, the known disorders of cognitive systems and what that looks like in these developmental — everything from optical illusions of perception. Optical illusions, simple stuff, all the way to excessive certainty about what's going on, not enough certainty, loss of agency, loss of purpose. You can stack them up and just see what we already have. Then from there, develop some of these connective mid-level models, as Alexey was just saying, that would show this is going wrong because you've overestimated the importance of this particular signal and thus put yourself in this condition. From there, we can start to develop therapeutics.
[43:10] Alexey Tolchinsky: Super interesting. I think throughout your work, you're saying all intelligences are collective. I think we need to talk about it. I look forward to, Thomas, the high-level paper that you think about how psychiatry would look like with Michael's principles. I think we can talk a bit about it. Let me give you a citation. The definition of dissociation in the latest version of the DSM-5-TR is a disruption and/or discontinuity in the normal integration of consciousness, memory, identity, emotion, perception, bodily representation, motor control, and behavior. The integration is deemed normal in the guiding document. If I quote a neuroscientist, Anil Seth, he says that monadic self is a form of a delusion in a sense that it only exists at the level of subjective belief and not an objective outside reality. Then we have your statement that all intelligence is collective. How do we reconcile these things together? What do we do? I want to give some credit to DSM. It is indeed so that on average healthy individuals subjectively feel, at the level of phenomenology, the continuity of the mind. It goes back to Immanuel Kant's 18th-century idea: "I am the same as I have been before, and I predict to be the same in the future." We bind together things when I listen to you speak into a coherent whole. There's a lot of integration/binding that happens, and a lot of disorders have to do with disintegration. The important thing, I think, is the perspective. Collective intelligence is something that we see from the outside, not from the inside. The way I feel inside my body is, at that level, a coherent whole. The second thing is when we wrote in the paper that if we model self as a nested embedded functional system, then essentially it's a hierarchical system of beliefs. I don't like Anil Seth's word "delusion." I think it pathologizes an important thing. I would say "belief" or "inference," but beliefs can be powerful; they can kill. It's only the higher levels of the self that believe in the unity: the autobiographical self, the core self. Hunger doesn't even have an "I"; it's low blood sugar that compels me to go do something. So this is the higher levels of the hierarchy that have the unified self. This is what we interact with in psychotherapy with a patient. The final thing is the distance. If we have a house and we look at a person moving between different rooms, then they are different and collective perhaps. But if I'm a mile away and I look at the person in the house, it looks like the same system. I want to say that it's not necessarily a paradox. If we qualify at which level of the hierarchy and from which perspective, then we can have unity here and collective there. We also can have, at lower levels of the hierarchy, collective intelligence. Take an ant colony: the intelligence of the colony is unified. At the next upper level of the hierarchy, that's definitionally emergent. Emergent property happens at the higher level. I wanted to bring it here to discuss.
[46:20] Michael Levin: I'm cautious about this notion that, under normal circumstances, the patient feels unified in the sense that we are really reliant on language and observable behavior. This is what happens when I talk about intelligence of organs and so on: people say I don't feel my liver is conscious. The left hemisphere that is talking to me right now will make that claim, and that's great. But your liver, with whom I cannot yet speak, although we're working on it, might have a different perspective on this. Of course, you don't feel it any more than you feel me being conscious. It's clear that in the normal human, there is a voice that feels unified for sure. I'm not super sure that there aren't others that normally don't get to speak. I think we know from the sodium pentothal experiments, when you put the language hemisphere to sleep, and the split-brain studies, there's at the very least one other set of opinions in there and probably many more, but they're generically suppressed. I think we have to be careful about that. I don't know what you guys think about that.
[47:46] Thomas Pollak: No, I agree. I promised I wasn't going to bring Buddhism and meditation into it, but I'm going to now. I think my experience and a lot of people who do meditate is there's this learning curve where you start off, the monkey mind is thinking all the time. You eventually get through that. There's a feeling of maybe some more unification. People use the analogy of the choppy water's at the top and you get down and it becomes more still. But a lot of people then realize that when you really get quiet enough, that's when the weird stuff starts happening. That's when you start hearing strange little voices, then you don't know where they come from, or ideas that are popping into your head that feel like they've just been burped out of nowhere. I think there's a reason that Buddhists put so much on the fact that there is no unified self, because at least the phenomenology of some of these states, which are hard won and take a long time to get into, suggest that really, when you pay attention, there's not that unity. Also, dreams and things that happen to us throughout the day feel less unified than maybe they might.
[49:11] Alexey Tolchinsky: I agree with you both entirely. When I wake up, I notice myself in this non-egoic state that Metzinger is talking about and Thomas, you talk about where things just flow through the mind. There's not even an "I." I don't yet recall that it's Tuesday and 9 o'clock or what I'm about to do, but just little things that flow through the mind is this non-egoic sense of self. But I wanted to leave some space. I'm trying to say that we're not here to come in and say you are wrong when you talk about unity, because if you take the very severe patients, with DID, disintegration causes a lot of distress. These patients suffer tremendously. And so it is, of course, a spectrum. But I think that we do work with reintegration in a specific clinical practice of traumatized patients with DID, DPDR and dissociative disorders. Yes, it is a belief, yes, but that belief is very important to try to restore, because they cannot work, they cannot maintain relationships with their loved ones, and they drink too much or use substances. Part of it is the disruption — the lack of continuity and predictability. Developmentally, a lot of things that happen in children's lives can be metabolized, but when they have a lack of predictability and continuity, this is an issue. It's even divorce — the world changes. So I just want to say that in more severe kinds of pathology, there is space for this. I'm not saying it's hardware, but at the level of software, it's important that Microsoft Word is the same Microsoft Word, isn't it?
[50:45] Michael Levin: And I wonder, we see this right during embryogenesis when we look and say, there's one embryo. Well, what is there one of? Because what you're really looking at, for example, is 100,000 cells. And when you say there's one embryo, what helps you, what you're really counting is alignment. And you're counting the fact that all of these individual cells are aligned towards the same vision of what they're doing. They're building a particular structure. They're going to work like hell to get there despite various interventions. I don't even think these are necessarily distinct things, but there's a way to think about this as reintegration, so that now there really is one thing, but under the hood, what that might be. And again, I'm not making any claims about psychological matters because I don't know any of this stuff. But I wonder if under the hood, what you're really looking at is a more adaptive alignment so that the fragment that's good at particular kinds of interactions handles that business in your life. And that's another fragment that handles something else. If they're not aligned, then of course it's going to be a nightmare. But if they're aligned, then it's fine that you have these specializations that can come out and do specific things. I don't know if that makes sense.
[51:59] Thomas Pollak: It really does. Is there a way, Alexei, that in these extreme cases, dissociative identity disorder, it almost feels there's a fight for real estate. I don't see these patients very often, but it feels that way. There's a fight between the alters at the same level. There's a lot of argument that the alters may not be as rich, but they're both claiming the top agential spot. So the integration at that level is clearly necessary; it would be hard to see how someone could be healthy without integration at that level. When you start thinking about the verticality of it, and I appreciate that dividing it into horizontal and vertical isn't that useful, perhaps, as Mike says, it's more about alignment, everything pointing in the same direction, as opposed to there being one single agent at each level that wears the trousers.
[53:14] Alexey Tolchinsky: I think we're all, yeah, go ahead.
[53:16] Michael Levin: I love the real estate point, because maybe this avatar therapy that you talked about is an example of that. So instead of fighting over the same real estate, what if we give you more real estate? The problem is that there's only one body, and if one wants to sit home and watch TV and the other one has to go to work, then there's an issue. But if you can, there's some version of VR or some other thing, and we're thinking about this in the biomedical case too, there's almost a decoy, you get some real estate of your own. You can inhabit this thing and do your thing, and you don't all have to fight about the same thing. I don't know what that would look like, but I do think that in all of these cases, that fight of patterns for reality, the fight for real estate in which they can manifest is huge. Maybe we can grow the pot in some way. Maybe that's a solution?
[54:13] Alexey Tolchinsky: I think we're pretty much on the same page. Maybe we use different perspectives, but in a jury, there's 12 members, the jury is functionally doing the same thing. They need to do something together. And so I think if we move up at the level, then we can see that they have the same function or the same goal, or these collection of cells are navigating the maze. They're all navigating the maze together. And the DID — one of the prominent theories is that integration was too painful. If I remember, the coherent whole hurts too much. So I'm going to compartmentalize and move into this piece where it hurts less. As Frank Putnam said, one of the experts in the field, it's the escape when there's no other escapes. These are prolonged cases of neglect and abuse when it feels inescapable. And the extreme defense is, I'm going to break my mind into pieces. This is the extreme immobilization, a sort of dysfunction. I'm going to do that because I cannot tolerate the pain anymore.
