6 thought provoking questions posed to @awaisaftab (psychiatrist) and myself (brain researcher) and we hit on so much:
The challenge of escaping reductionism. Theories of consciousness. Are mental disorders brain disorders? Why should anyone care about philosophy? Is epistemic iteration is failing? And what bits of brain research are awaiting their Copernican moment?
With nods to @summerfieldlab, @knutson_brain, @tyrell_turing, @Neurograce, @eikofried
and so many more.
Read it all here (and let's discuss)!
https://awaisaftab.substack.com/p/advancing-neuroscientific-understanding
@NicoleCRust @awaisaftab @summerfieldlab @knutson_brain @Neurograce @eikofried
Fascinating discussion! Thanks for sharing this.
But, I can't help myself, I have to engage in my usual refrain here (sorry, lol), since it comes up in the second paragraph:
Brains are *literally* computers and information processing devices, it's not a metaphor!!! 🙂
@tyrell_turing @NicoleCRust @awaisaftab @summerfieldlab @knutson_brain @Neurograce @eikofried
Is that an "are" of strict identity? Or do you believe brains also have functions outside of computing/information processing..?
@WorldImagining @NicoleCRust @awaisaftab @summerfieldlab @knutson_brain @Neurograce @eikofried
Strict identity - if we're using the broad definitions of computing and info processing from computer science, anything else implies magic/spirit.
Cognition is not computation. It is an elaboration of organismic agency, which is not of an algorithmic nature. There is nothing magical about that.
You're fundamentally misinterpreting Turing's theory of computation, which was always about the human act of computing, not about the brain or the world. Turing would turn in his grave if he would hear you.
Some historical context: https://plato.stanford.edu/entries/church-turing.
This may also help: https://arxiv.org/abs/2307.0751.
Longer paper coming soon!
Respectfully, I disagree. I don't think you're accurately summarising computability theory, nor the implications for neuroscience.
As Yoshua Bengio once said to someone at a workshop I was at (paraphrasing):
"Computation just means physics. Saying the brain is a computer is just a way of saying that it is a physical device. I don't even know what a non-computational theory of the brain means, unless you're talking about magic."
Remember, as the very article from the Philosophy Encyclopedia describes, the original goal of both Turing and Church was simply to formalise the idea of an "effective method", i.e. a mechanical way of solving a problem.
If the Church-Turing thesis holds, then any problem that a mechanical (i.e. purely physical) system can solve represents a computable function, and the object implementing it is engaged in a computation.
Did you know that this argument only came up with Deutsch and Lloyd in the 1980/90s? Before that, nobody (including Church & Turing) ever thought that it would be reasonable to apply the term "computation" to physics.
So, no: it's you, not me who's misreading Turing. Effective computation means literally "something a human being can do by rote" (i.e. calculating, scheduling, optimizing). In Church & Turing's 30s papers, a computer is a (usually underpaid, female) human.
I don't really want to engage in an extended debate here, but I will just note this:
Turing and others very much so connected humans being able to do some calculation by rote to the ability to create a mechanical device to do that function.
So, there was always a direct connection to physics/mechanics, because the implication was always that if an effective method existed, then a physical/mechanical system could in principle implement the function for us.
Nobody doubts that you can mechanicize rote human behavior. In fact, that's exactly the point of a universal Turing machine. It is designed to implement any process that can be machanicized that way.
How you get from that to "the brain or the world are a computer/mechanism" is mysterious to me. And it was to Turing and Church who always resisted such an unwarranted extrapolation.
No extended discussion needed, really. There is simply nothing that supports your worldview...
@yoginho @tyrell_turing This thread is a fun read, but I wonder if you are both working from the same definition of computing to ask if cognition or physical processes are computation?
If you each were to independently write down definitions of computation, how operationally similar would they be?
If you have very different definitions, it's a disagreement of assumptions, not one that can be resolved through evidence.
cc @NicoleCRust re: my comment in your thread on jargon.
@debivort @yoginho @tyrell_turing @NicoleCRust This thread, while educational to me, is the opposite of fun. I'm sure @tyrell_turing is used to it for having provocative opinions, but this style of academic attack turns many people off from engaging in discussions in the first place. If you find someone's take "absurd" why not approach it with curiosity and good faith instead? @NicoleCRust is a great example to follow - often disagrees, never attacks.
@beneuroscience @debivort @yoginho @tyrell_turing @NicoleCRust
This conversation is genuinely confusing to me. @yoginho seems to be saying that "computation" means anything that can be mechanized, reduced to an algorithm or program, and the brain is not limited to that. @tyrell_turing seems to be saying that a computation is anything that a physical system can instantiate, and so a brain, being physical, does computations.
I"m confused as to whether there is genuine disagreement here, or just different definitions of computation. @yoginho can you explain what are, or give examples of, things a physical system (a brain) can do that are not computations by your definition? @tyrell_turing, are you stating that the brain, or indeed any physical system, cannot do anything that is not computable, reducible to a rote computation, an algorithm or program?
Again, what I'm trying to understand is: are you using the same definition of computation, or different definitions of computation? Are you disagreeing about what physical systems, including the brain, can do, or only about what to call it?
... no closure to physics (it remains forever incomplete) & is therefore not formalizable in a way the term "computation" suggests.
This is why I find any broad & vague application of the term misleading & restricting. By considering physics as computation, you limit yourself to the syntactic, symbolic realm (a small world where everything is well-defined), but the world we actually live in is full of ill-defined problems, ambiguity, & relevant cues are hard to come by.
Basically, you can have two attitudes towards computation:
(1) You can use computational models as tools to understand the world, with all the limitations of the approach that implies.
(2) You can come to live in your computational model and mistake it for the world. That's what computationalism is. That's why I call it deluded. It mistakes the map for the territory. It commits what is called the equivalence fallacy.
This immediately & radically restricts the questions ...
... you can ask, and the explanations you consider scientific. Our physical theories become complete descriptions of the world. You become trapped in your small world.
That's why computationalists can't even see the point of the argument I'm trying to have with them. The world is large and open-ended, but you can only get a shallow understanding of why that is if you live within your limited model.
Concrete examples:
Stephen Wolfram has a model called the "Ruliad," which he considers to be a model of not just this world, but all possible world. But it's 100% computational. So: nope. It remains trapped in a small world and does not even begin to describe what is possible in the real (large) world.
https://writings.stephenwolfram.com/2021/11/the-concept-of-the-ruliad
Similarly, Sara Walker and Lee Cronin have come up with an interesting computational model of chemical complexification during the history of the ...
... universe, which is called "assembly theory." But instead of seeing this as what it is (an interesting model) they claim it is a new theory of physical time.
https://aeon.co/essays/time-is-not-an-illusion-its-an-object-with-physical-size
In both cases, the models are really interesting, but it is a mistake to claim they *are* the world. This is my main point here.
Computation is a powerful tool to model the world. But I have no idea what someone means when they say "physics *is* computation." Probably, it does not mean anything.
@yoginho @kendmiller I'm not a proponent of the digital physics idea or even know enough physics to really articulate how the universe works, but here's a paper by Hector Zenil that lays out how the universe could unfold algorithmically if it's doesn't do so randomly https://arxiv.org/abs/1109.2237
I'll have to read this more carefully. Thanks for sharing!
The abstract gives me strong vibes of map-instead-of-territory though, and the rotten whiff of the Laplacian demon is also succinctly present.
I've written about why that ugly demon must die (and should never be allowed to come back again) here: https://www.expandingpossibilities.org/4-death-to-the-demon.html.
@yoginho I'm pretty sure the author leans towards believing the universe is fundamentally computational, -- because there are so many things are algorithmically compressible, which are unlikely to be generated by random analog systems -- so definitely keep an eye out for biases and implicit assumptions like Laplace's demon, but it's at least a detailed explanation of that point of view. He mentions it that could be nondeterministic, mostly random and analog with occasional order, but if it is deterministic it has to be algorithmic
p.s., I'm interested to read more of your project. Its focus on agency made me think of Karen Barad's "agential realism" theory https://www.dukeupress.edu/Meeting-the-Universe-Halfway/
@kendmiller @beneuroscience @debivort @yoginho @NicoleCRust
I personally think it is just a semantic disagreement.
Ah. Dismissive non-engagement again.
To clarify whether it is "just a semantic disagreement" we'd have to first know what you actually mean by "computation" and why you think it's useful/reasonable to equate that to all of physics (which, I guess, means the whole of reality).
Why so reluctant to tell us? 🤔
Maybe if you would engage with the actual argument, I'd be less of a pain. Worth a try?
I do get the impression that do like to make big sweeping pronouncements, but shy away from constructive discussion, not just with me...
If you send a few links to work of yours that clearly outlines your assumptions, I'd be keen to read that.
My position on this topic is well-summarised in this paper:
https://www.frontiersin.org/articles/10.3389/fcomp.2022.810358/full
And no, it's not really worth my time to see whether engaging with someone who is using insults and complaints to try and goad me into have a debate with them online would lead to better results.
Note that I have been happily engaging with others on this thread. Maybe it's worth a try to not be a dick? 😘
Thanks! I browsed your paper quickly & find it very useful because it clearly exposes the kind of misinterpretations & unsupported assumptions I'd expect in such an argument. I'll read it carefully, & we can have a genuine discussion.
I wonder: how much energy did it cost you to send this link? How much easier would that have been than your repeated whining about my provocations? Why post glib provocations yourself, if you don't want to engage? It all seems a bit odd to me.
Ok. I read the paper carefully and really like it. It lays our your conceptual framework very clearly, which allows me to precisely pinpoint the gaping holes in your argument.
The fundamental hidden assumption, that is just stated but not justified at all, is that algorithmic computation "can be applied to almost any object in the universe."
This means we can simulate (approximate) anything, like throwing a ball, but the actual ball throw is not literally computation ...
The important difference is between "making a computational model of the ball throw" vs the claim that the ball throw *is* computation, which it isn't, because (a) it has no symbolic content (other than that imputed on it by your model), and (b) it is not necessarily Turing-computable (as you claim without justification; but maybe you realize that classical mechanics, standardly used to describe the ball throw, is not computable because the real numbers are not) ...
The second error is a simple flaw in your logic: human brains can calculate all computable functions (in theory), therefore the brain *is* a computer (equivalent to a Turing-machine).
Copeland calls this the equivalence fallacy. You cite his paper, but apparently have not read or digested that part of it?
The theory of computation is explicitly a theory of what humans can do by rote (as you write in the paper), so it is hardly a surprise that brains can do effective ...
... computation. What you can't see (from within your computational small-world worldview) is that this is by far not all that brains do, nor did brains evolve to do this (most animal brains don't engage in this kind of activity at all, in fact).
The logic error: for the brain to be literally a computer, it would have to do nothing but compute (in the Turing sense). As I said above, you simply assume that this is case, because you assume that the whole world is computable ...
The problem is that this assumption is not supported by any evidence. There are tons of processes and problems in the actual (large) world that cannot be formalized in the sense of Hilbert.
One, for example, is how organisms realize what is relevant in their environment: how they pick out the features of the world that are important for their given situation. If you try to formalize that, you get an infinite regress. Same thing with organismic organization, as Rosen ...
... has shown a long time ago. Life is not computable. Behavioral and evolutionary possibility spaces are not prestatable as clearly defined mathematical sets. Your assumption that every physical process *is* Turing-style computation is making the simple mistake of confusing the map (your model of the world) for the territory (the actual world).
Your computationalist worldview is based on a completely unfounded assumption. And it traps you thoroughly within a small world ...
... as evidenced clearly by your article. Physical systems that are not von Neumann computers (eg a neuromorphic chip) can do a whole lot (in fact, infinitely) more than just calculate the set of Turing-computable functions. Just because you can perform computations with them, or model them with a computational model, does not mean such systems are computers.
But, I guess you can't see the problem, since you have shut any non-computational physical processes out of your world.
Your are right in pointing out that there are various physical implementations of systems that you can use to perform (more or less universal) effective computation. Your brain is one of them.
But you're totally wrong in your thinking that this is all that physical systems (incl. your brain) can do. The world is much richer than you think. But you can't see this, because you literally live inside your computational (small-world) model or the large world you actually live in.
So this is *not* "just semantics." Or better: semantics matter a lot in the large world we live in.
Your Wittgensteinian definition of the meaning of a term is such a cop-out. "We use the term like this, so that's what it is."
Instead, it seems more useful to me to ask, how does the use of a term reflect on the problems we want to solve with it.
"Computation" really isn't helpful in this sense, if you want to understand those aspects of the brain which are organizational ...
...rather than those connected to the subset of brain processes concerned with actual information processing.
In brief, your term "computation" *is* a metaphor & a pretty bad one, limiting you to only a sliver of the world when you should look at a bigger picture.
Your paper fails to address this problem, bcs it fundamentally fails to question its own assumptions, while dismissing those of others as not real.
Well, they *are* real. Just not in your small world. Consider that?
P.S. Church and Turing would have very probably hated your interpretation of their theory of computation. Both, to the end of their lives, insisted that computation is about a certain kind of (human) activity, not about the brain or the world.
P.P.S. If you don't understand what I mean by organizational vs. information-processing aspects of the brain, think about Weizenbaum's distinction between calculation (that's what computation captures) and judgment (that's what it does not, and was never intended to).
Hmm. I was genuinely curious where you get the idea from that all physical processes must be Turing computable, and why you define a brain that can do effective computation (but also many other things besides) as "literally a computer" (when it is also many other things).
I was also interested what you have to say about simulation vs. reality.
But I guess that was it from your side?
@kendmiller
I defined "computation" above as Turing-computation. It's the only useful & rigorous definition I know of.
What does the equality of computation & physics even mean? What is physics? To me, physics consists of a number of large scale models of the world that remain incomplete. Also: if you consider the possibility of radical emergence (new rules coming into play at particular transitions in the history of the universe: eg nucleosynthesis, origin of life, and so on) then there is..