sofiechan home

Ingressing Minds, or: Against Physicalism, Michael Levin

anon_pawe said in #5619 1d ago: received

https://www.mdpi.com/2409-9287/11/5/161

This paper made the rounds on X a few weeks ago, and I think Levin's background and experimental thoughtfulness/success mean that we should evaluate the ideas seriously. I'm very curious to hear the opinions of sofie poasters.

In particular, what Levin's saying seems to be a stance on the 'hard materialism' we've debated in the past, in particular in >>1754, and also Gnon theology from e.g. >>3814. Levin is arguing in roughly the same direction as xenophon from >>1754. His big claim is that 'minds' and other 'emergent behaviors' that we see in advanced organic lifeforms such as humans are really best described as Platonic patterns that live in the same space as e.g. basic mathematical truths, and that those patterns 'ingress' or become physical in sufficiently complex substrates (often far less complex than we might guess, e.g. bubble sort exhibiting delayed gratification). He also argues that pragmatically, this is a vastly superior prior than just saying 'muh physics'.

My first reaction, as someone who finds it difficult to deny the tautological truth of physicalism and the God of Nature, was to attempt a reconciliation. Indeed, the poaster xenophon took a similar approach by arguing for the emergence of function from form, and form as not-reducible to physicalism. To be a bit loosey-goosey, the space of all possible forms and therefore functions is something like the set of all possible configurations of matter, which remains properly ultrafinitist due to the Planck-length and Planck-time, and my feeling is that there's sufficient expressivity and size in that space of potential matter configurations in a Planck-voxel universe (modulo quantum indeterminacy) that the magical ideas which Levin points to as 'Platonic' could be thought of just as 'physically possible forms'. This seems to be a more rigorous formulation. Am I wrong?

But the meat and potatoes of Levin, as an experimentalist, is to point out that regardless of what's actually True, we need to spend vastly more effort on figuring out what's happening in the 'Platonic realm' or space-of-forms, rather than getting stuck on trying to ascertain the particularities of the function which advances the state of the universe from one iteration to the next (i.e. physics):
> monism should not be an assumption—we should be open to whatever ontology it takes to help advance science, applications, and ethics
> I eschew questions about whether the space is “ontologically real”, because as an engineer, to me “real” simply means whatever I need to worry about or can take advantage of, and these patterns certainly meet that criterion.
> I take a pragmatic approach: not commitment to ancient categories (which likely need to change) but to empirically fruitful conceptual frameworks. I think these are to be evaluated not by consistency with ancient ideas, or by the ability of competing views to—after the fact—accommodate experimental results, but by fecundity in driving novel, interesting research programs that would not be reachable otherwise.

I hope this whets your appetites, and I encourage reading the paper and using it to ground discussion rather than just rehashing old threads.

referenced by: >>5621 >>5622 >>5623

This paper made the received

anon_fafa said in #5620 1d ago: received

For me the most powerful intuition pump in the paper is, from memory, the part about the frank impossibility of ontogeny when faced with the possible configuration space of living matter.

The idea that the ~Gb of usable information in the genome is sufficient to describe eg a human body so specifically that something like inheritance is so apparent from parent to child beggars belief.

Instead Levin suggests that the process of ontogeny is the process of moving from attractor to attractor, where each attractor is one of these ingressing patterns.

That the genome can specify a body is miraculous enough. That it can specify a mind the more so! And after reading it i couldn't help but speculate about whether, to what extent, and how these same attractors are in play in training llms. But any deep insights there are beyond my reach for now.

For me the most powe received

xenophon said in #5621 1d ago: received

>>5619
I like Levin's biology work. I find his forays into philosophy to be cartoonish, especially his jumping to "Platonic" patterns. There are patterns all right, but it only muddies the waters to call them "Platonic." It jumps to conclusions about the best metaphysical description.

> form as not-reducible to physicalism
To be more precise, I argue that form is non-reducible to matter, and this by definition, because matter itself does not include form. I take "physicalism" to be a more expansive notion. One could reasonably take to it mean something like "everything is described by phusis," and phusis certainly includes form, to include the forms of living things that Aristotle calls psuche.

Step back from biology, back even from organic chemistry, all the way down to fundamental physics. Our best description of matter in the small is the Standard Model of Quantum Field Theory. This is already a deeply mathematical theory, richly structured, and defined by symmetries and related patterns. If you don't know what I'm talking about, spend a minute or two contemplating this diagram:
https://en.wikipedia.org/wiki/File:Standard_Model_of_Elementary_Particles.svg

What's my point? That even at the lowest levels of matter, long before we've defined amino acids, before we've even defined ATOMS, there is already mathematically defined form. So, form is present and operative at the lowest levels, and continues to be present and operative at higher and higher levels, from living beings to sophisticated human institutions to AIs.

Here's what I think is the best metaphysical description: Form is a principle, not a thing. It does not exist on its own. (Hence my rejection of Levin's foray into "Platonism.") Likewise, matter is a principle, not a thing. It too does not exist on its own. Things exist, and every thing is composed of matter and form.

If I wanted to assert a version of "physicalism," it would be: "Only things exist." I personally don't feel a need to assert that closure condition.

referenced by: >>5622

I like Levin's biolo received

anon_sylu said in #5622 19h ago: received

>>5619
I have not read Levin's paper but the genre is common enough. People like the platonic forms idea because it's an opening for asserting the reality of unreal things, which is usually attached to some otherwise indefensible ethnical claim. You guys can tell me what that is in Levin's case.

Pragmatically, form exists. Our boy xenophon here >>5621 has the shape of it. Form is the way matter is arranged, the shapes it takes, the system of its interactions. His arguments convinced me on that. When we talk about computations, mechanical systems, life, consciousness, math, etc, we are talking about form. None of this though implies that form exists *apart from* matter in some other world ("the platonic realm") where there are coherent goings-on, any more than position. We may be able to see the consequences of particular form in advance by reasoning about the pure forms, but this is form as an abstract unreal symbolic *approximation* of what's really going on, which is a messy bunch of physical detail interacting according to low-level law.

Platonists want the high level to be realer than the low level for some reason, but it's not true. To maintain it anyways requires this incoherent stance of saying the difference between your model and reality is that this is not the real world and this world is actually broken and evil or just a shadow, and the real world is somewhere else in gnosis, salvation, enlightenment or whatever process.

I have not read Levi received

adamjesionowski said in #5623 18h ago: received

>>5619
§4.3, which asserts that the space of forms evolves in the same way the space of matter evolves, was the most intriguing section of this paper. That is not something I've heard proposed before and is a natural post-Darwinian move back to a dualist metaphysics. It's silly to dismiss this paper on "I can't believe it's just Platonism" grounds for that alone.

On the other hand he loses me in §6. I have never cared for the part of Levin's programme that aims to bring machines into the same fold as organisms, and the arguments really start to stretch in §6.3. I grant that his Xenobots work does move into gray areas between a clear biological being and a clear artefact. However, to use his language, the interface between a CPU and an organism is so wildly different that I can't take §6.3 seriously at all. It looks to be the start of an assertion of LLM consciousness to me. No thanks!

Credit to Levin for sticking his neck out for non-physicalism though, it's a much needed radical departure from the kinds of accepted metaphysics in biology.

> To be a bit loosey-goosey, the space of all possible forms and therefore functions is something like the set of all possible configurations of matter, which remains properly ultrafinitist due to the Planck-length and Planck-time, and my feeling is that there's sufficient expressivity and size in that space of potential matter configurations in a Planck-voxel universe (modulo quantum indeterminacy) that the magical ideas which Levin points to as 'Platonic' could be thought of just as 'physically possible forms'. This seems to be a more rigorous formulation. Am I wrong?

Is there a member in the set of all possible configurations of matter in a finite universe that realizes π? Only some forms of matter are efficacious as far as a form persisting goes, and those forms can be composed into other forms--so it seems like we should order these into a hierarchy of forms rather than a flat set. I think the more you try to be pragmatic about the space of forms, the more you end up with Levin's argument.

§4.3, which asserts received

You must login to post.