Michael Levin

Scientist at Tufts University; my lab studies anatomical and behavioral decision-making at multiple scales of biological, artificial, and hybrid systems.

[Michael Levin @drmichaellevin]

Wordpress site is up - https://t.co/RuziJul6Si. Register to be notified of book progress, specific events, news, and new posts with photography, essays, interviews, & more. Unlike at https://t.co/nzip6HyaUZ, here I will post ideas not fully baked yet, & academic-adjacent content.

[Michael Levin @drmichaellevin]

1) a standard physicalist, when looking at a system like this and asked why it has this or another kind of look, still has to admit it's because some kind of equation has only 3 not 2 or 4 solutions... Why does it only have that many solutions and not more or less? Go see the math department. The answer won't be in physics, although physics will be affected by the answer.
2) it's not actually "math" that provides the true cause. Math has no monopoly on this, it's just the oldest exmaple. A lot of the interesting causes for physical events look like simple, low-agency patterns and they're studied by a department called "math". Right next to those patterns are ones that sit between math and psychology - truths of cybernetics for example - and those are studied by others. In biology and cognitive science, we have yet more agentic patterns that really are studied by the behavior science department.
3) for real-world therapies. I don't draw any "lines" - I emphasize a spectrum of tools and use whichever ones help best (not whichever ones we're allowed, to avoid ancient "category errors"). I also don't believe in a distinction between "real things and epismetological framing that allows progress" - I don't know anything about ontological status; if it helps make progress, it's as real as it's going to get (for 3rd person science, anyway).
4) But to your biomedical point. Yes, if done well, it yields specific tools and improvements, because: once we understand the contents of the latent space of possible patterns that guide living matter (patterns that are not exhausted by what we must add via selection, training, and design), and we understand what specific aspects of the physical embodiment is the key for facilitating/suppressing specific patterns, and we have tools for detecting what level of communication (where on the cybernetic hierarchy) each pattern is amenable to, the game changes radically. Each of these by itself is important, but together they will make a massive difference to how we predict disease and manage interventions. A given therapy will always go through some physical interface, of course, but how we predict, control, and build is greatly affected by that. See https://t.co/DdEhwS20Fg and https://t.co/L7d8Osw28C and https://t.co/VRRd0n9eUH. I don't mention Platonic Space there (I focused on ascertaining the agentic level of things before designing intervention, ignoring the source - it would have been too much for those venues) but the connection is obvious. You have to understand the source and behavioral level of the dynamical states you're trying to manage and implement in biomedicine.

[Michael Levin @drmichaellevin]

@danfaggella 🤘🤘

[Michael Levin @drmichaellevin]

#naturephotography

[Michael Levin @drmichaellevin]

#naturephotography
(more at https://t.co/zZL1W0NH7v)

[Will Ratcliff @wc_ratcliff]

1/ The origin of multicellularity was one of the most profound transitions in the history of life, underpinning arguably the single largest increase in organismal complexity since the origin of cells themselves.

[Michael Levin @drmichaellevin]

@YaC_osaurus @ClaudeInside @kevembuangga @IlyaVityugov @CankayKoryak @IAI_TV Yep!! We've been doing this for a while. Stay tuned! it's ongoing. But like everyone else, our results won't say anything definitive about consciousness per se - just intelligence and its mechanisms.

[Michael Levin @drmichaellevin]

"human consciousness" - the one with the language and the obvious motion in 3D space, sure. How about the other intelligences in the body? We don't know how to check them, so we shouldn't assume a no. In fact, since they use the *same* mechanisms as our brains, and many of the *same* algorithms, and we share an evolutionary history (https://t.co/BKw7p6FJET, https://t.co/RB83AFW6DG, https://t.co/VpBn8xRfme, https://t.co/A9PLaCRf4A), the null hypothesis should be a "yes until proven otherwise".

[Michael Levin @drmichaellevin]

@HetkeBrian @kevembuangga @IlyaVityugov @CankayKoryak @IAI_TV so in the deep future, they stay apart forever, yes? that's what you see - for a mature species, in both science and spirituality, they're just always separate domains and never integrated? that is not what I see in our future. and if not, we might start thinking of that now.

[Michael Levin @drmichaellevin]

It used to be pragmatically sufficient to do surgery on babies with no anesthesia. Until the mid-1980's - not 1880's, 1980's. It's often true that women and other groups get less anesthesia. In this case, pragmatism (which I love) fails us because WE DO NOT KNOW whether there is experience in those intelligences (which we know exist) which share our bodies with us but cannot speak up for themselves. The "behavioral responses" are motion in 3D space, which we as apes with good vision are obsessed with, but biological beings traversed many spaces long before nerve and muscle appeared - we have lots of papers on this. Many biological structures act in spaces that can't be seen by us directly, and so the fact that we don't consider them when making pragmatic decisions that focus on the welfare of the one with the language and the wallet means nothing. The whole point of my argument is the very systems we're interested in, we have no way of checking, and there's no way the null hypothesis should be "then they don't have it, because we don't know how to check for it".

[Michael Levin @drmichaellevin]

It's a prank, it's not nasty, I don't think. Like math is a prank on physics; like biology is a prank on chemistry; like a moebius strip is a prank on a nice flat strip of paper. Like all good (not mean) pranks, it reveals our blind spots, shocks us into raising out game, pushes us to let go of limiting assumptions and find new pranks.

[Michael Levin @drmichaellevin]

I do agree that mechanisms of anesthesia are interesting and important, but no, we don't know that consciousness is reliably be removed by anesthesia. The consciousness of the one mind in the body that you can realiably talk to - that is removed; and even that is tainted by the short-term memory wipe effects of some anesthetic mixes - you're hearing from post-anesthesia patient, not during-anesthesia patient, that everything was nice and absent during that time. But most of all, you have no idea what other experiencers in the body there might be whom you never heard from because they don't use language and they operate in spaces that are hard for us to visualize. We're working on that btw. So let's remember: "there's no evidence for X" is only as useful as our ability to actually detect X or its absence. https://t.co/zbuxyMsz0f . I'm not diminishing anything, just practicing the useful part of science where you examine your assumptions and ask yourself what would I *miss* using the net I'm using to fish.

[Michael Levin @drmichaellevin]

it's not outside of science. It's outside of 3rd person science as it's been practiced for some time. It doesn't have to be that way, science can expand, but it has to adapt to its substrates, not shoehorn them into old categories. It's pretty good at that in the long run. I'm optimistic.

[Michael Levin @drmichaellevin]

"common properties of beings we assume to have consciousness" - my point precisely: this set is not astronomical because you likewise *assume* the others not in that set do not have consciousness. We have no clue which those are. Once you open the set to ones where we can't rule out consciousness (all things), what are the common properties? This line of thinking is super prevalent, and all of it stands on the smoke and mirrors of people assuming they know what's in and what's out of the magic set, all of which calibrated on human verbal reports as N=1 example. We wouldn't put up with this in any other field of science. We cannot form all of this on an assumption, that btw goes against the one example each of us has. And "brain like" is a whole discussion (what exactly is brain-like? how far back? all cells do what neurons do) and why do you think post-processing is necessary for qualia? all assumptions.

[Michael Levin @drmichaellevin]

testing? how? we have no idea; I'm not sure it's possible. The closest that I know of is what is described in the below figure (https://t.co/lex2pO0JdB). But I am skeptical about the existence of "unconscious systems" at all. Each of us has found 100% of the systems we can actually test (ourselves, only) to be conscious. On the basis of this, you could conclude we don't know (N=1 is very low sample size), or you could conclude that the null hypothesis should be: "everything is conscious to some degree until proven otherwise", but what you absolutely cannot conclude, by any acceptable form of reasoning, is that "most things have no consciousness, and my friends and I are rare exceptions". If we had any 3rd person ways of measuring consciousness, or any tie for consciousness to specific architectures that wasn't calibrated exclusively on 1 example - human verbal reports, then we could perhaps run some sort of Occam's Razor argument, but we don't have those, and so must set our priors in accordance to the results we have. Which, for all of us, is: we have a positive control for "is conscious", we have no positive control for "isn't conscious". I think "I've only ever seen A, but I'm going to assume most of the Universe is B" is dangerous and bad reasoning. I think the minimal, conservative position is "we have absolutely no idea" - not lean to 0, but literally no idea, unflinchingly held. The more balanced position I think is, it's a continuum that has no known lower bound and is likely ubiquitous.

[Michael Levin @drmichaellevin]

AI's are absolutely not a positive control. Positive controls are examples where, to a standard of scientific evidence, we are sure of what the answer is - that's the whole point of a positive control against which to test other cases. I do not believe we have any "clear criteria" for 3rd-person assessments of consciousness that achieve anything like the level of certainty we expect in other fields of science. And I think the issue you're referring to is not the failure of introspection (introspection is the only criterion any of us has): it's the uncertainty that a system's language use is accurately conveying the results of that introspection. The problem again is on the side of 3rd person verification by other minds, it's not an issue with introspection itself.

[Michael Levin @drmichaellevin]

Doesn't mean they are beneficial - absolutely (also, beneficial to *whom*? Are you beneficial for the cells your body allows to die every day?). Also doesn't mean they are smarter. And yes, doesn't imply they have consciousness. But, you absolutely can't assume they don't, either. So let's specify: what evidence would you want to convince you there's consciousness in any given system? What exactly are the criteria? And, what is the positive control - what's an example of a "just agent" that you know, with the kind of certainty we expect in science, doesn't have a 1st person inner perspective?

[Michael Levin @drmichaellevin]

@IlyaVityugov @CankayKoryak @IAI_TV https://t.co/x6lS8QzqmO

[Michael Levin @drmichaellevin]

@WojtekKrzyanow4 @adameykolab Just fyi, that's not me - Michael Levine is a different developmental biologist. Mine is:

https://t.co/eMTFDJu9bM

chapter 2