Forms of life, forms of mind

Different aliens may see different math; doesn’t that blow up Platonism?

I don’t claim (and my current view doesn’t require) that all species see the same math. Imagine a landscape of all possible mathematical systems. Depending on where you start (axioms), and how you traverse that space (which can be very species-specific perhaps), you will find different things; some things will be forever unreachable to you. Different kinds of agents will see totally different regions of that space, or maybe see some overlapping things and agree on them. But, that doesn’t mean that the contents of that space aren’t providing something that is discovered, not created: if you start at a particular spot and explore a particular way, you discover (i.e., you find out, get handed, not have a choice about) certain things. Whether someone else found those things or not doesn’t diminish the fact that you got more than you had at the start. Even if the  planaria, or aliens, can’t start with “empty set, and successor of empty set” and then eventually find a shockingly specific value of e, we apparently did. Maybe they got other gifts, or maybe they got more than we did, or maybe much less. That’s ok, compatible with my view. All I think, for now, is that some ways of exploring the space pay off in ways we don’t manage via anything we consider to be within the province of physics.

How does your view compare and contrast to Gordana Dodig-Crnkovic’s Cognitive Platonism?

I agree with most of it, for example:

– abstract forms as causally efficacious

– there is no pre-written script

– emphasis on feedback between hardware and the software that constructs it

– virtual, substrate-invariant causal architectures

– emphasis on anticipation

– no hidden metaphysical world (because, I don’t think it’s hidden, and I don’t think it’s metaphysical 🙂

– morphological computation in biology – yes!!

Where I differ:

– emphasis on constraints. Yes, there are constraints, but I think that framing (“what you can’t do”) is most appropriate for things studied by physics; by the time we call something the subject of biology, or behavioral science, I think it’s much less about constraints and more about enablement – not just new paths made available because old ways got blocked off, but specifically getting more than you put in – free lunches so to speak. This is important because it’s testable and quantifiable (we’re working on this now); I think evolution exploits the heck out of these but also, they themselves are patterns with agendas and competencies of varied kind and degree.

– “biological form emerges, no central controller, no mental representation”. This is an empirical claim, and it needs to be defended for specific instances. There are surely aspects of biology that are well-described by this kind of model – emergence + complexity science; cellular automata for example. But, I know of no reason to think that all of biology fits this open-loop kind of model, and I have written extensively about examples where it breaks down and leaves too much on the table. Models of virtual governors (central controllers, in a way), and homeostatic processes with a biophysically-encoded setpoints (goal representation) are very useful in developmental biology and bioengineering. Let’s ask: what would it look like if there was representation and a process that aligned parts toward a larger-scale goal – would we recognize it? What would we have to see, to give up (for that one case) the commitment to parts that know nothing of the final state and just do their thing while overall form emerges? Well, here are some criteria, modeled after the criteria for recognizing memory engrams in cognitive science:  we’d have to find a biophysically measurable state that (a) precedes the final outcome, (b) has some specificity with the final outcome (i.e., you can decode the state and see how its features map on to the final outcome), and most importantly, (c) can be *re-written* such that the system then goes to a different, predictably controllable outcome. In other words, you know your thermostat is not an emergent system because there is a memory location where you can read and write your wishes about the future state, and it will implement them: that setpoint is indeed the master plan that guides the system and you know it because you can change it without having to rewire the system.  I have published a number of examples that match conditions a,b,c showing that not only do some cases in morphogenesis have representation and top-down controls, but they even have counterfactual representation and some more sophisticated aspects.  To summarize: it is an empirical, not philosophical, claim that “emergence” is sufficient to understand biology, and there are cases where I think we can readily see that this class of models do not match the data and do not offer the full potential for control and discovery. Much more here.

What are some things to be careful of?

– terms like computers, computation, virtual machines – I don’t think anything is any of those things because I don’t think those are actual things. I think they are formal models – metaphors and frameworks for thinking – which certain phenomena can match to a greater or lesser degree. When they are very useful, we start to think they capture reality. I think we have to be very careful and always ask, which aspect of reality are they specifically *not* capturing. I think it’s always there.

The ontological vs. epistemological distinction, which things are ‘real’ vs. metaphysical – I’m not sure what that really means. I’m an engineer; if I have to worry about it, or if I can rely on it to build something new, then it’s as real as real gets (at least, in the context of 3rd-person science and engineering).

The assumption that Patterns arise solely as products of cognitive activity:  the specific value of e, and other truths of number theory and such – is it constructed by cognitive activity (and is thus changeable to suit) or is it discovered by it? I think the latter.

You sometimes say that people send you crazy stuff all the time. Why not publish it?

It’s true that I have often envisioned the comedic value of a book that just contains a selection of emails (and devices, and objects that defy any classification) that I receive – wouldn’t have to edit it at all, it would be great as-is.  My problem is that I don’t know how to define the boundary between “this is super weird, in a funny way” and “psychiatric disorder”. I would never want to make fun of the latter, and since I can’t decide if there’s a real boundary here or how to determine it, for now I’m not publishing it. But that might change in the future, who knows. I don’t promise anonymity or confidentiality if people send me unsolicited stuff, but I also haven’t decided to just make it all public.

You said, the genome controls the ensemble of bioelectric fields. Perhaps I misunderstood?

I was rushed, and probably didn’t make myself clear… The genome is essential for giving each cell some number and types of ion channels. Those ion channels are needed for having any bioelectric patterns at all. So they are needed, and to some extent they constrain and enable what can happen (which is very different depending on whether you have genes for voltage-gated ion channels, for example, which enable new dynamics – feedback loops and such). But, “controls” is a very strong word. It’s kind of like, a computer that’s computing prime numbers. What controls which number comes out next? On the one hand, Maxwell’s equations moving electrons through the CPU. But that’s only true in the most boring sense of “controls”. What really controls it is deep patterns in mathematics that determine the distribution of primes. Same here. The genome sets the hardware, but what animates it are patterns that are not set by genetics any more than the value of e is set by the laws of physics or the history of our genome etc.

You say “A causes B if and only f A is a fruitful explanation of *why* B occurred rather than something else.” My hesitation here is to tie causation and explanation together. I see causation as mind-independent and causal explanations as ways for humans to try to understand the causal nexus. I think A can cause B even if it outside of human cognitive capacities to be able to explain why B occurred rather than something else.

I certainly don’t mean to rely on human capacities (I include all possible observers – cyborgs, alien scientists, AI’s, whoever). But, I don’t know what it means to say that causation works independent of cognitive agents making sense of their world. I have a very operationalist view of science, and to me what “causation” has to mean is a relationship that allows good predictions (by an agent), provides insight (enabling an agent to do better science/discovery in the future), etc.  If A doesn’t explain why B (as opposed to not-B) occurred, then I don’t know how it can be a cause of it.

Why aren’t bioelectricity-based applications in human patients yet?

Think of how much money has gone into biochemistry and molecular biology over the last 50+ years; I don’t know the exact numbers but the investment in bioelectricity has to have been << 0.01% of that – it’s a matter of resources and amount of effort that’s been put into it by the community over time. The other, scientific, answer is that biochemistry and genetics can be done on dead tissue: take cells, fractionate them, sequence everything, etc., the information is still there. Bioelectricity is much harder – it only exists as long as the cell is alive. It’s gone the moment you kill, preserve, freeze, separate, or fractionate cells. The technology to read and write that kind of physiological information in the living state is relatively new. The first noninvasive imaging of non-neural voltage patterns, and the first use of molecular biology to manipulate those patterns, was only in 2002. If you can’t slice organs into pieces without losing the info (cell voltage changes if you separate them from their microenvironment), it mean you have to image voltage fluorescence through tons of other cells. That’s difficult. Think of how long it’s been since the first complete gene sequence was read – 1972 or so? Bioelectricity is an emerging field, but it’s definitely moving forward. A number of commercial efforts are working on human stem cells, mouse regeneration, and electroceuticals for human cancer etc. It’s coming but it needs a lot of tool development up-front to get to the exponential rise tipping point.

A little more. I asked myself this very question, when strategizing about it in the early 1990’s. One problem was that the older electric field data (while very good) made no contact with the dominant paradigm – molecular genetics. No one knew the source of the endogenous bioelectric patterns, and no one knew the targets downstream. Also, no one in the mainstream cared, because they thought they could ignore it – it made no impact on the pathways they were interested in (or so they thought). By 2002, we made the first molecular tools to allow visualizing patterns of Vmem in vivo over time, non-invasively, without having to poke each cell with an electrode, to manipulate the Vmem patterns via ion channel misexpression (so you knew exactly which cells were modified and in what way – very different from prior electric field applications), to find out which ion channels were responsible for the endogenous signals (inverse drug screens), and to identify which gene expression changes (of things like Sonic hedgehog, which everyone cared about) were downstream of the bioelectric signaling. We used these methods to integrate the bioelectric signals into the developmental pathways that the mainstream cared about, and now they needed to pay attention because it impacted on genes and transcriptional networks that were important to them. We used molecularly-tractable model species and released tools to everyone, which allowed mainstream labs to get into it. We developed computational tools (also open source) that enabled predictions integrating the molecular genetic signals with bioelectric changes over time – which were ultimately used to infer interventions for birth defects etc. We characterized transduction machinery – 6 molecular mechanisms by which bioelectric changes control cell behavior and gene expression, so now it was known *how* bioelectrics exerted their effects. Second, prior work focused on ion fluxes and electric fields. It was known that these could direct cell migration. We focused on resting potential patterns, which enabled mapping out at least part of the bioelectric code in inducing the creation of entire organs (ectopic eyes, first) – that had not been possible before. This in turn connected to deep ideas in neuroscience about high-level stimuli propagating through networks to control downstream mechanisms that we don’t need to micromanage. There’s more but this gives an idea of what was different – more powerful concepts, practical tools, and a commitment to making sure we anchored the work to endpoints that the mainstream cared about. It was (and still is) a slog, but the ice has broken. The pioneers of the past – Lund, Burr, Cone, Jaffe, Robinson, Nuccitelli, Borgens, etc. were inspirational – they had it very tough, but they saw the possibilities clearly (especially Burr, back in 1936) and fought tooth and nail so that others had access to the data showing why this is so important.

Computers are better than living matter because they support instant knowledge transfer, perfect copying.

That’s instant data transfer, not knowledge transfer (here’s knowledge transfer). Perfect copying is a dead end for high-agency architectures; it resists polycomputation (one ground truth about what a physical state means), removes drive for improvisational interpretation of memories, loses out on plasticity, etc. Biology leads to an intelligence/agency spiral because it’s dealing with a fundamentally unreliable medium. See: https://thoughtforms.life/self-improvising-memories-a-few-thoughts-around-a-recent-paper/ , https://mdpi.com/1099-4300/26/6/481 , https://link.springer.com/article/10.1007/s00018-023-04790-z, https://cell.com/trends/genetics/fulltext/S0168-9525(25)00077-0

Does every instantiation of a structure from the “Platonic space” have a subjective experience? If yes, does it mean that there is something to be like the sorting algorithm you mentioned? If not, how and why does it happen in only certain instances?

I don’t have a strong claim on this yet (because I don’t have data pushing toward one view toward another and because having an answer to this is not a rate-limiting step for anything we need to do next). But, my personal opinion is that they all do, to a degree (I think that consciousness in general is continuum of degrees and kinds, not a binary yes/no). In the sorting algorithm, there could be 3 different Perspectives: (1) the entire array (morphing over time, but permanently persisting), (2) individual digits, moving around (but also invulnerable and unchanging in their embodiment), and (3) clusters – forming, behaving, and eventually getting ripped apart in their world (very mortal, having a brief life of doing things neither forbidden nor prescribed by the “physics” of their world, until it grinds them down in the inexorable path to their universe being sorted). Also, I’m not sure that “have” is the right word with respect to subjective experience, maybe “are” is better.

Where do bioelectric patterns come from? What button turns it on?

There’s no button, because cells come from other cells, so it basically goes back to the first living thing with a membrane that separated it from the outside world (and thus created a simple charge imbalance).   If you want to track the mechanism, it’s easy: they originate from the dynamics of electric current flow through ion channels. We can track it from the 1 cell egg even before fertilization. It’s like asking where the pattern for “0” comes from when you turn on your calculator – it’s the default dynamics of the electric circuit of the cells, before any external inputs. But there’s a deeper truth. If you ask why a mathematician emits the sentence “7 is prime” – on the one hand, it’s because specific chemical events in her brain and neurons cause this behavior – you can, theoretically, track causation down to the atoms. But in a deeper sense, why she says it is because 7 is prime which is a mathematical truth that has nothing to do with physics, chemistry, neuroscience, or evolution. Same with bioelectrical patterns: there’s the physical explanation, and then there’s the mathematical truth that stands behind electric circuits and why they make specific patterns and not others.

How does interpretation of bioelectric patterns work?

It basically evolved the way all symbolic codes evolve – by a feedback loop between sender and receiver, both incentivized to communicate and save information in a way that compromises optimally between being most powerful encoding, most hackable by internal parts but not external parasites, etc.  There’s also the issue of polycomputing, which means that different observers decode the same physical events in different ways.

Continuity of mind: are your talking about continuity of intelligence only or do you include qualitative experience?  I could see intelligence as present in gene regulatory networks but nothing like qualitative experience until the mind becomes more sophisticated.

Ok. So I do have a big question to resolve that’s relevant to this, which I haven’t decided yet (having to do with how minds scale, given that I think the bodies are collective intelligences but I don’t think a human mind for example is a glued-together collective of neuronal minds). But on the specific question you asked:   I suspect it’s qualitative experience all the way down. I can’t prove it of course, but I hate discontinuities, so I don’t know what “more sophisticated” means – how sophisticated, and more importantly, what gives you a sense of certainty about any opinion you might have on this? Let’s examine ourselves: why do we think that qualitative experience requires some specific substrate and why do we believe that any process/system can have 0 experience? What makes us think that no-experience is the default or even possible? We can crank up the knob very slowly, and what – experience suddenly winks in from 0?  I find that very unlikely despite people talking about phase transitions, “more is different”, etc. – I don’t find any of those examples convincing, because most of them (except for a few quantum physics things, which I can’t really comment on) still allow you to zoom in very slowly and all those sharp curves become nice and gentle slopes again… I think any phase transition has to be argued for very strongly, it can’t be the null hypothesis. And if we say that it doesn’t go from 0, it gradually increases as complexity rises, then fine, why not say it increases gradually from the very beginning?  I think geometric frustration (like in tiling planes, Ising models, etc.)  is real frustration (just an extremely minimal version). And I think qualitative experience is what Platonic Patterns have (not physical bodies), so any pattern can have some (perhaps tiny) degree of it, and when they project through physical bodies, that’s when the rest of us catch wind of it (via observable intelligent behavior of some degree).

Why does physics offer no support for “free will”?

You can always find a perspective from which something is invisible. Physics will never see free will, just as it can’t see inflation of airline prices, primeness, and many other important things. That’s ok; economics can’t see magnetic fields.

Why don’t many things exhibit free will?

There’s a massive difference between having the capacity for it and exhibiting it. I think in many ways “having free will” is very very close to “having the ability to lift weights”. In the following ways. 1) it’s not binary – question is, how much weight, not yes or no. 2) it’s not constant across people – different ones lift different amounts. 3) it’s not constant within the lifetime of a person – at different stages you can lift different amounts and at any given moment, you may be lifting with intent or going on autopilot and just keeping minimal “body posture” (also lifting but automated). 4) it depends on external circumstances – it’s easy to lift on the moon when you’re healthy and fresh, harder on Jupiter, or when you’re sick or when you’ve been lifting all morning. 5) it grows with use and reduces with disuse – keep exercising it, and your capacity increases. 6) other people see the results – what got lifted – they don’t know what it cost you in effort. 7) it’s not an in-the-moment thing – if you want to lift more, the price is time-extended effort in practice (that’s the real meaning – consistent effort, showing up constantly to increase your cognitive light cone and eventually have more free will).

Are you constrained to stay within orthodoxy?

No. My radius is set by 2 practical considerations. First, to have impact (i.e., advance science and help people with applications), you can’t be so far from everyone else that you’re disconnected from smart people, and/or in such deep waters where no meaningful progress can be made right now. Despite what people keep emailing me (that if I wasn’t worried about my job, I’d be talking about the aliens, miracles, whatever), it’s not about the job, academic pressure, or any of that. No one is suppressing me. But I think it’s more fun (and more important) to make practical progress than just talk about things we can’t do anything about. Having a positive impact means being ahead but not so ahead that no one else can build on your vision. It’s a hard balance to find; I do my best. Whatever I may think personally, I set a scope for the work that tries to be frontier but still actionable.

Second. I try to divide things into 3 buckets. A) Things I currently believe because we (or others) have data for it strong enough that I think others should believe them too. I talk about those things publicly – I publish papers, give talks, etc. B) Then there are the things I suspect, and they’re actionable hypotheses and we’re working on them in the lab. I find them interesting enough to commit time and resources to, but the data aren’t there yet to make any conclusions and I usually only say those things when someone specifically asks. Also I try really hard to be clear about the difference between A and B – I always clarify which things we have data for and which are conjectures/hypotheses. C) Then there are the truly Big Questions, which I have no new data on. You know – aliens, ultimate nature of reality, psychedelic elves, meaning of life and death, etc. I typically don’t talk about those. I have personal thoughts but so do many others and if I don’t have new data or new arguments to add, there’s no point in my adding guesses to the sci-fi or new age bookshelves. In personal conversations I can say what I think, but I don’t subject the public to it. There are already enough people pontificating on everything, and it’s important that no one confuses what I say in my official day job, speaking for the combined scientists in my group, the weight of our rigorous work behind it, vs. speculations on areas where my thoughts have no more validity than anyone else’s. But if you’re really into that category, stay tuned, sooner or later I’ll slip up somewhere and run my mouth when I shouldn’t, probably.

Anthropomorphism is a legitimate term – a cognitive error.

No; it’s like “heresy” – it’s emotionally charged, it sounds like it means something, and it effectively holds up research. It used to be a thing but needs to be retired. Anthropomorphism is like that.  Or better yet, it’s actually the opposite error than you think it is: when someone hears “memory” and instead of engaging with the functional data, cries “anthropomorphism”, it’s because they hold a pre-scientific belief that memory, goal-directedness, decision-making, etc. only happens in humans. Anthropomorphism is the pre-scientific assumption that humans are a special category that can do amazing magical things and that seeking those same competencies in other substrates should be banned by philosophical fiat.

We typically do biology with the un-examined assumption that everything other than brains is understandable as mechanical actions of components that know nothing and have no goals. It’s taught as some obvious and necessary pillar of science; but it is an axiom.  Maybe when science was first getting off the ground, it was a necessary assumption. But in the modern age in which we have cybernetics and behavior science, we’re no longer limited to the ancient 2 options of “stupid like rocks vs. smart like humans or angels”. So we can drag this unquestioned axiom into the light and state that ubiquitous hypothesis as:  “everything other than brains can be handled with the lowest rung of the cybernetic hierarchy.”  But, why would we believe this without testing it?  There are other levels of that hierarchy, which are not mysterious, not magical, and used every day by engineers who study everything from self-driving vehicles to (now) the evolutionary precursors of brains (body bioelectricity). Why not explore, using the tools we now have, and ask: all the levels of biology, from molecules to ecosystems, where on the hierarchy of degrees of learning, goal-directedness, problem-solving, etc. does each thing lie? It is most consistent with the spirit of scientific inquiry that we drop un-needed ancient axioms in favor of experimental testing. And it turns out, as a matter of empirical results, that it’s a bad assumption that has been suppressing discoveries and capabilities. It’s time to move beyond it, and also to drop the unscientific but pervasive belief that mind is necessarily mysterious (and for that reason shouldn’t be brought into biology other than, maybe, brains).

Why are you sure it’s not physics you’re discovering?

Well, if you re-define “physics” as “everything that exists”, then ok, I guess it’s physics, but then you still have 2 very different “sub-realms” so nothing is gained. I prefer to keep “physics” as the kinds of things discovered by, and changeable by, the efforts of physicists. Just sticking with math examples for the moment, I don’t think you can fire the mathematicians and hope the physicists will tell you why the complex numbers don’t keep the same promises as quaternions, or why the Feigenbaum constant has the value it has, etc. etc. And I don’t think there’s anything you can do in physics, tweaking the constants at the Big Bang, to make the digits of e be any different. Most telling, even people who say “it’s just physics” or “it’s a human construct, arbitrary” unviersally have very different reactions when I ask them to consider a non-stationary value of the speed of light or gravitational constant (“no big deal”) vs. when I ask them to consider that the value of e changed over time (“impossible”), so no one actually believes they are the same kinds of things…

The conventional framing is that Plato’s philosophy is abstract and concerned with a higher, unobservable domain of Forms, whereas Aristotle’s is empirical, practical, and grounded in the observable world.

First, I’m not actually into keeping up Plato’s philosophy. My view diverges from his (at least as far as I understand what he said), and the only reason I call it (for now) Platonic Space is to make connection to mathematicians who understand things the same way (and then extend it past the objects of math into cognitive science). Maybe I will rename it later. Second, I don’t know about conventional framing and who’s used Aristotle’s ideas to make empirical discoveries, but I have 5-6 people working in my group on experimental work arising from these ideas, so “unobservable” is not correct for my view. It’s unobservable sort of like quantum foam – not obvious, but not totally hidden if you know how to look, and actually having massive empirical implications. Definitely observable, now, which is why I finally stated talking about it last year.

We have 2 hemispheres, why does consciousness feel like one seamless experience?

First, a better question might be: our left hemisphere is made up of huge numbers of cells; why does our consciousness feel like one seamless experience? Your question is good but why stop at hemispheres – why not the stuff inside each hemisphere? Second: feels that way to whom? One agent in the body is good at language, and at telling others that they are the only tenant in there. If your organs (including the right hemisphere, but also other structures) also felt like they had a seamless experience, the left-brain tenant wouldn’t know and would still ask this question (which I think is exactly the situation we have now). I’m extremely skeptical of “unconscious learning” and stuff like that where the “it was unconscious” verdict is obtained by asking the “subject” – the one that talks.

What’s this “what kind and how much” about cognition – don’t we need a definition first?

My point is not that definitions are not needed. It’s that we need not a definition of crisp binary categories (which just land us in pseudoproblems of corner cases), but a definition of the continuum along which our property of interest scales. So for example, as in the paradox of the heap (Sorites): I don’t want to try to define a “heap” precisely. What I want to define is the understanding of how to interact with it, and when to do so with tweezers, a spoon, a shovel, a bulldozer, or dynamite. So whatever definition people want for consciousness (and I agree that we should specify it when discussing the topic!), it needs to be a definition cached out in continuous terms, not “is it or isn’t it”. And I am very careful to define the terms I use (persuadability, intelligence, etc.).

Isn’t your new “free lunches” data all coming from quantum magic?

I don’t have the background to make any arguments about quantum computation and what it might add, but: the reason we study very minimal computational models, in addition to the biologicals (where people can always say there’s quantum biology we haven’t found yet), is that we can show this stuff happening in purely classical systems with no quantum magic. So, maybe QM adds some cool bells and whistles, but it is not essential for the kind of thing we’re quantifying. Newton’s classical world was already haunted and Pythagoras saw it clearly.

Isn’t it critical to know the intelligence status of biobots etc. to settle moral questions?

yes, but we haven’t settled the rights of pigs and cows in factory farms, so I guess we’re not very good at linking rights to whatever we learn about diverse intelligences. If we can’t figure out how to relate to pigs, I doubt the nuances of cultured tissues will make any impact.

What about Rupert Sheldrake?

I read Rupert’s first book as a teen and was motivated by it – I thought, here’s an example of big thinking, across disciplines, not daunted by the obvious avalanche of negativity that would follow. I know Rupert, and I applaud him for having the courage to stick by these ideas for so long. As for the idea itself. I think it is interesting for sure but doesn’t go far enough. What he’s hypothesizing is that sensitization is fundamental to the universe – a cognitive skill (recognizing and reacting more strongly to stimuli over time – a kind of memory) is proposed to be fundamental and acting very widely across substrates. Cool. But why stop at sensitization? It seems plausible to me that all the other cognitive competencies should be sought for as well – habituation, anticipation, etc. etc. I suspect these kinds of things “go all the way down” (and up) as I’ve said. But also, we can’t just think these things, we have to do experiments (as Rupert agrees). So I remain agnostic on everything here except the ones we’ve actually tested (which is a huge minority of the stuff this could be applied to). So I don’t know, but I don’t think it’s impossible and I do think overall that the fundamental nature of things is not physics, and it’s not even “math” as we define it today, it’s the wide range of patterns from the Platonic Space and most (all?!) of them have aspects of mind, so yeah, it can be checked for in the most unconventional places. Rupert picked some things to look for first – bravo. Others can look for related things elsewhere. We will all see what is found.

What do you mean by calling common terms like life, machines, etc. “artificial categories”?

What I meant by it is ancient categories that are “artificial” in the sense that they don’t cut nature at its joints: they give names to categories that sound like they’re radically different things but that’s only because of our limited knowledge. They are imposed by us for convenience or because we don’t know better, not by nature itself. For example, light, radio waves, gamma rays. We gave them names, but we now know they’re fundamentally phenomena on the same continuous spectrum. So if someone claimed they’re distinct essential natural kinds, we could say no, these are artificial boundaries – maintained by ignorance or expedient everyday use (like “adult” in court) but not scientifically supported by what we now know about them. I claim that when we zoom in and try to define the edges of many terms, we don’t see sharp categories with good definitions – we see convenient labels that obscure a deeper scientific unification.

One example I’ve talked about it is the “life” vs. “machine” distinction. Now that we know about bioengineered chimeras and about mixing evolutionary strategies with rational design in any degree, I argue that “life” vs. “machine” is an artificial binary distinction – there are no 2 such essential categories and trying to maintain the distinction doesn’t help us understand the true nature of this continuum. Of course we can come up with items from the far ends of the spectrum which look very different, but every position on that spectrum is occupied by some kind of system that is very hard to classify in those 2 pure camps, and we don’t need to. More on that here and here. But aside from that contentious opinion, I also argue the same kind of no-binary view for many other things we’ve gotten used to: thought vs. thinker, data vs. machine, and lots else.

I’m not arguing that there can’t be joints. There almost certainly are. But, 1) I don’t like it when people assume it – I think continuum should usually be the null hypothesis and if one thinks there is a phase transition or whatever, they’ve got to argue for it, 2) I am skeptical of “emergence” (not for the usual reductionist reasons – I think it doesn’t go far enough); 3) sharp categories are awesome at generating pseudoproblems (hard corner cases) that only get resolved decades or centuries later when we scrap the whole distinction and realize it’s a spectrum; 4) I personally have gotten way more advances from asking “what might the spectrum be here, and what lies between A and B” than from looking for new sharp categories. And, since the “binary” vs “continuum” distinction is, itself not binary, I don’t “disagree” with you, I disagree somewhat 😊

Dualism is not testable

Strict “testable” is as hard for materialism as it is for dualism (a crisp experiment that would cause adherents to drop the model), but if you mean it has no detectable consequences that can be experimented on, or that it generates no new research agendas, I beg to differ. We are doing experiments and analyzing primary data relevant to those questions. I’m only interested in the kinds of frameworks that drive you to do new experiments no one had done before, not mysterianism. And I’m not interested in any axioms that are not allowed to be questioned (e.g., materialism). It’s all up for investigation, and every model eventually meets its limits.

Can your technologies help Trans people?

My answer is in two parts. First, I have to be realistic and say that we don’t currently have any technologies that are ready for human patients of any kind – we are still at early stages in model systems. But second, I firmly believe that bioelectricity, and other tools, will lead to something I have called “freedom of embodiment” in my lectures – complete control over growth and form (probably in your lifetime, perhaps even in mine).  Once we know how to repair standard organs, we will be able to ask cells to build whatever we want. I think this will go far beyond what is possible now, and I think the next generation will think of us the way we think of primitive hominid ancestors and their myopic, limited lives. I think it will seem ridiculous to them that we had to live our entire life in one body, which we were randomly handed at birth. And they will shrug their wings and wiggle their tentacles and wirelessly share their bemusement with each other, whether under the sea or in the air, over the fact that such things as mere sex change caused so many controversies and conflicts. Anyway, we will keep working as fast as we can toward the time when everyone can be whatever they want to be – in mind or body.

Causality – is there anything not derived from physics?

“Derived from physics” is a tricky concept. Can the *reason* for airline prices going up or down be derived from physics? Can *the best next move in chess* be derived from physics? Could the idea of making a computer in the Game of Life CA be derived from its physics? I think looking backwards after something happens and being able to tell a physics-compatible tale of what happened is not at all the same thing as claiming the physics level is the most powerful or insightful level at which to understand a phenomenon and then do new things (not just explain backwards). I think it’s not, and I think there are lots of causes that trace their way back to facts of mathematics (for example) that can’t be derived from physics (is there a physics story for why quaternions don’t behave just like octonions?). But math isn’t the only thing like that, I’m claiming; and, just like some situations in the physical world trace their most insightful, powerful explanation for why they happened back to a non-physical pattern from a field we call math, others trace their most insightful explanation back to a pattern from a field we call psychology.

Continuum of life forms, ok, but why don’t you talk about how people are different?

No one needs me to be talking about how “people” are special. Every philosophy, cognitive science, linguistics, etc. book for the last how many thousands of years have taken this view. It’s hardly useful for me to echo all that – if you want that, plenty of others to read. But, the workers in those fields who take developmental biology and evolution seriously have also talked about how beings who start as a single cell *become* people and gradually, slowly, acquire (along, you guessed it, a continuum) some or all of the features that we mean when we say “people” in legal contexts and other practical, non-science settings. If I knew of a truly sharp phase transition that was useful for any of the problems I care about, I’d be talking about it. I’ve never heard of a convincing one. Ways to understand transformation and up-scaling though – those are useful and it’s what I work on (even though I spend most of my time on the left end of the spectrum, again, because plenty of folks in child development, evolution of language, etc. are already doing the other side).

And, as I’ve asked 1000 times, when someone throws around the word “people”, exactly which people are meant – you want to dust off your phrenology calipers for a cyborg neighbor coming home from the hospital with some kind of smart implant to see if he’s still a “people”? We got to use that word (until recently) with impunity because it just so happens that the other kinds of hominid ancestors no longer exist. It didn’t have to be that way – easy to imagine a planetary history where more of the lineages survived, with massive differences occupying the spectrum between non-people and people. What would you say then? Of course it’s tendency to use all kinds of minor differences to say that a population “isn’t really people, like us” – history tells us how much we love to do that and the consequences of that approach. But perhaps at least those trying for a scientific perspective, and an ethics for a mature humanity that is consistent with the science, wouldn’t be using unsupportable binaries and complaining about continua if some of our ancestors were still here and obvious. But no worries, the coming diversity of technologically and biologically possible beings, some of them very close to modern humans but very different in some key ways, are going to *really* shake up that snow globe for the traditionalists and for all of us. I’m just thinking out loud, with others, in case anyone else wants to think about it in advance and adjust the pre-scientific categories we’ve inherited to the reality that’s pretty apparent by now.

Michael Levin’s critics frequently attack his ideas of “cellular intelligence” because they sound mystical

Right, they claim that any talk of intelligence is mystical. I think this would be of great surprise to cybernetics, the neuroscience department, or machine learning communities.  Intelligence being mystical went away over 100 years ago… At this point it’s a superstitious, pre-scientific worldview to say that intelligence = mystical. Time to catch up. Also, scientists aren’t supposed to make decisions based on how things “sound” – they’re supposed to think rationally and do experiments.

They reject the idea of talking as if tissues possess an active mind seeking a goal.

That’s silly and insulting to neuroscientists. Everyone accepts one example of a set of tissues that truly seek goals: brain tissue.   So, let’s steel-man this to a claim that’s at least not silly; maybe they mean: brains are unique in seeking goals. What would this claim be based on? And, where did brains get this amazing property, during their evolutionary and developmental histories – it had to come from somewhere. We point out that the mechanisms, and algorithms, brains use to do it are in fact evolutionarily ancient and very well-conserved to other tissues in the body.

A collective of cells does not contain an image of the final body

In at least some cases, this is factually incorrect because we can literally visualize (just as neural decoding can visualize it in brains) the pattern memories in tissue (not surprisingly, encoded bioelectrically) the “image” that serves as their anatomical goal state. Thermostats encode setpoints, brains encode goal states, and tissues encode anatomical setpoints too – we can see it, we can rewrite it, and the cells build to spec. We’ve published all this many times in different contexts; for example, reviewed here.

---

Featured Image by GPT-5