Aaronson is smart enough to understand IIT and work out its implications, whereas most critiques are not even wrong. However, like nearly all computer scientists and physicists, he starts from a misunderstanding of what consciousness is (hint: not intelligence, not complex problem solving, not advanced behavior). No wonder IIT doesn't fit well with "what he is calling consciousness".
@isaacdavid2491
2 жыл бұрын
I mean, the empirical literature surrounding IIT and competing theories is kinda big, and it grows by the day. No matter how welcome curiosity and interdisciplinarity might be, one must always be wary of not peeing outside the urinal.
@NomenNominandum
2 жыл бұрын
@@isaacdavid2491 Even if IIT is right, I don't think it addresses Chalmer's hard problem of consciousness. At best it can adequately quantify consciousness. Yet it seems not even able to "detect" consciousness, unless you are bold enough to accept that all things with Phi>0 are conscious, like a Vandermonde matrix (which I personally find ridiculous).
@isaacdavid2491
2 жыл бұрын
@@NomenNominandum I think IIT not only answers the "what" and "how" in precise statements, but also the "why". If IIT is right, then the answer to the hard problem is: "existence is consciousness", with IIT providing the ontological analysis to figure out what truly exists, not just in quantity but in quality (look up the paper by Balduzzi and Tononi, or the one by Haun and Tononi). To me, that's the only answer that doesn't come across as contrived. I'm cool with not keeping asking "why" at that point, just as we don't normally demand explanations from physics as to why its fundamental laws and constants are the way they are. I accept Phi > 0 is consciousness, both on theoretical and empirical grounds. IIT and its approximations (like PCI) are probably the best tool neuroscientists and clinicians have to detect consciousness at the bedside, and the empirical evidence keeps growing.
@isaacdavid2491
2 жыл бұрын
@@NomenNominandum it would be nice to know why you find the Vandermonde-like system ridiculous. I know Aaronson's objections, but maybe yours provide a different perspective. Quoting Aaronson: > "I’ve shown that my system-the system that simply applies the matrix W to an input vector x-has an enormous amount of integrated information Φ." To that I say, any description of a discrete dynamical system will look like that, regardless of whether it has high Φ or not. Whatever system Aaronson finds convincingly-conscious (some artificial neural network, the actual neural network in his brain), I can model it in terms of matrices, vectors and their multiplication. > "So for example, if n were 10^14 or so-something that wouldn’t be hard to arrange with existing computers-then this system’s Φ would exceed any plausible upper bound on the integrated information content of the human brain." On a typical computer it wouldn't, because it wouldn't implement the causal links of an actual network, and IIT or any other causal analysis would be able to notice it and assign a low Φ. The fact that it's a Turing-complete system allows us to encode and simulate the network, but that's not the same as being the network. But leaving the inaccuracy aside, what's the problem with conceiving engineered systems that are more conscious than human brains? There are only ~1.6 x 10^9 neurons in the cerebral cortex, and only a portion contribute to consciousness. I wouldn't be surprised if a system of size 10^14 was able to attain higher Φ. But here's the crucial and funny part, he doesn't understand what is meant by "consciousness" in the literature. He brings this massive mathematical acumen to scrutinize a strawman provided by sci-fi movies. He conflates it with "intelligence", or implies that consciousness must be a subset of highly-developed intelligence: > "And yet this Vandermonde system doesn’t even come close to doing anything that we’d want to call intelligent, let alone conscious!" > "But that doesn’t imply that every time you start up your DVD player you’re lighting the fire of consciousness." (IIT agrees, the error correction performed in your CPU or embedded uC has very low Φ) Here's my counter thought experiment to show that expecting consciousness to depend on magnificent behaviour is even more ridiculous: any Turing-complete system can in principle implement any function. Even mechanisms simpler than a grid of polynomials: rule 110, SK combinators. On the other hand, a low-IQ person with paralysis, whose only mental contents are wild, meaningless hallucinations of blobs of color on a big visual field, is by definition highly conscious.
@NomenNominandum
2 жыл бұрын
@@isaacdavid2491 I would guess that you can implement a Vandermonde-like system on a computer because you know how these matrices look like. So then the computer running that piece of software should be conscious because it has a large ϕ, shouldn't it? I have been develping software for many years and a claim that a computer running an algorithm is conscious makes no sense to me. I am deeply convinced that there is an explanatory gap when it comes to consciousness and I don't think that science can ever fill it because the scientific method is not applicable in this case (it presupposes consciousness and tries to explain consciousness, thereby running hopelessly circular).
@MikeWiest
Ай бұрын
This is a valid critique despite smug comments asserting that he doesn’t understand IIT or doesn’t understand consciousness. We can’t rely entirely on intuition but you do have to start somewhere and predict the known cases and the most likely cases. Although I agree that IIT as currently formulated is probably wrong for reasons other than what he articulated, I disagree that it doesn’t address the Hard Problem. It does attempt to solve the Hard Problem, unsuccessfully in my view. But it is unique among the classical neuroscientific theories of consciousness in this respect.
@iterate_yt
2 жыл бұрын
There is only one “case” of consciousness for us to compare IIT to, and that is our own. The five pillars of IIT are designed to agree with our experience of consciousness and thus they have met that burden. Building a theory based on whether it agrees with our non-falsifiable notions of what else is conscious is assuredly just relying on intuition. This just sounds like a circular defense of functionalism.
@jay31415
Жыл бұрын
IIS is the string theory of consciousness science.
@suzannecarter445
Жыл бұрын
This may be simple-minded but when I first read "The Axioms" of Integrated Information Theory, I was astonished. . The axioms could apply to anything at all, any object that exists. They don’t say a thing about consciousness itself. Treating consciousness like any object seems opposite to the obvious fact that consciousness is pure subjectivity.
@Self-Duality
2 жыл бұрын
Thanks for posting! Subscribed.
@peterrabbit2965
10 ай бұрын
He spends seven minutes stabbing around the issue, then concludes "you still haven't proved my intuition wrong". The condensed version would be: "waffle, waffle, waffle, I'm right, prove me wrong".
@patrickl5290
4 ай бұрын
Had a feeling, checked the comments. Thanks for saving me 6 min
@NickMirro
2 жыл бұрын
There is confusion about the 'why' of consciousness. e.g. David Chalmers says "Why should physical processing give rise to a rich inner life at all? It seems objectively unreasonable that it should, and yet it does.” Why is it there? is to ask, What makes it occur? That is different than, Why is it there? (its function or benefit). We should start with the latter meaning of why, since biology can get us going. Consciousness has probably-obviously evolved. Following Scott's points about intuition, we expect that consciousness should occur in living creatures. So we can also intuitively expect that, like most everything else about us, it evolved. It evolved because it solves an essential (we wouldn't survive without it) purpose. So with that, I propose a simple purpose: impetus (not will). Consider all consciousness as pain or pleasure. Pain and pleasure are the impetuses for everything we '"willfully" :) do. We should start there since pleasure and pain directly relate to our genetic fitness, and is intuitive. Consider any thought on which you act. It is to some degree pleasurable or painful, or somewhere on the spectrum. And that is entirely why you act. (c) Pain or pleasure is both necessary and sufficient to explain your behavior. It felt good to post this 😂
@victor_rybin
2 жыл бұрын
there are bed-ridden people, who are "doing nothing", and yet they are conscious. the fact that a wall can't be conscious - it's his intuition, he doesn't move away from appealing to intuition 🤷
@mykotron
2 жыл бұрын
but we have reason to believe the person not doing anything is conscious. We can't say the same for the wall. This isn't just intuition this is logical.
@victor_rybin
2 жыл бұрын
@@mykotron true. but then it boils down to saying that "humans are conscious because they are like me, and i'm conscious" - it doesn't provide any insights into consciousness itself
@user-vq6xc6zj5z
7 ай бұрын
If you give up the idea that intuition is valuable than you can give up logic. The reason is that logic works deductive -we see that some things don't seem to be possible so we can deduct -substract- them until we arrive at the last option possible. If you say everything is possible, even the stuff that is completely counterituitive, you request us to do a substraction from infinite ("everything"). And that would mathematically never be able to arrive at the number 1 - ONLY 1 logical possible conclusion, that is necessary for logic deduction to work.
@FunkyJeff22
5 ай бұрын
He explainsmany cases where intuition has been proven wrong. The keyword is proven. IIT hasn't proven anything.
@mattsigl1426
2 жыл бұрын
Aaronson inability to describe IIT charitably before he critiques it demonstrates to me that he doesn’t fully understand the theory. (If you can’t explain what you’re critiquing you can’t really critique it!) I’ve read (or tried to read, rather) Aaronson’s “critique” of IIT and found it almost unintelligible. (While Tononi’s work, while complicated, is always clear.) He goes off on mathematical tangents unrelated to the informational mathematics utilized by the Theory. as for the “conscious grid” example, I think Aaronson’s reservations are serious but not wholly justified. After all, are not the neuronal structures in our head grid like when untangled from their crammed confines in our skulls. Finally the fact that Aaronson doesn’t know that “phi” is pronounced like “rhymes with Pie” (instead of “fee”) is just weird. Outside IIT, the symbol is also used to signify one of the most important numbers in math, 1.618…, aka The Golden Ratio.
@danielsayre3385
2 жыл бұрын
2:20 I think the answer to asking if the technology with high Phi is conscious is.... yes. Yes it is. it sounds counter-intuitive to call something we may view as frozen and inanimate "conscious", but because it is part of a larger interplay of information integration I think that we need to consider the whole loop of effect->information integration->reaction to potentially be conscious itself rather than solely considering the reactive portion of that loop conscious. Something I've said before though not on this channel is that if humans are conscious then so is the environment that we influence by proxy Lol I unpaused and he literally started saying what I just said, even if he disagrees with it I think that biting the bullet is the way to go. 3:40 - 6:45 que the intelligent language that is tangent to the subject but not necessarily adding anything, and is used as a tool to back up a value-driven strawman that a "blank wall is very very conscious". Buddy, relax. You don't like the idea that something potentially inanimate in relation to your opinion of consciousness is conscious. That's fine. If the other guy is telling you something involved in integrating information is conscious itself, maybe argue against why it's not, rather generating language about why it's so foreign to you in the first place that a piece of that process of integrating information could be considered conscious. I think that the opinion opposing yours IS intuitive, though it's not based on the classic template/stereotype of "conscious". Expanding our definition of consciousness sounds like a forward progression. It doesn't have to resonate with your previously held beliefs about the restrictions consciousness has in order to be true.
@NomenNominandum
2 жыл бұрын
What about deep sleep? There is a lot going on in the brain in this situation, so phi supposedly is very high. But where is the consciousness?
@danielsayre3385
2 жыл бұрын
@@NomenNominandum Hahaha it's my opinion that sleep is the brain attempting to integrate and predict information. The effect is internal, the response is internal. It is a loop, but your question itself is fucking hilarious. Unconscious consciousness I did say that word a dozen times lmfao
@danielsayre3385
2 жыл бұрын
(deleted this comment because it sucked)
@connorschmidt5945
2 жыл бұрын
@@NomenNominandum as someone who follows the theory, IIT predicts that in dreamless sleep one is unconscious. this is because while there is a lot of information in the brain, almost none of it is integrated, which is consistent with the theory.
@NomenNominandum
2 жыл бұрын
@@connorschmidt5945 Thanks, that makes sense.
@arpitkumar4525
Жыл бұрын
You are right. Ultimately it's only us who gives meaning. If someone wants to define consciousness as just highly integrated systems then I am sorry that's not what I am interested in when I searched for a theory of consciousness and we are not on the same page
@connorschmidt5945
2 жыл бұрын
It doesnt matter what people "want" to refer to as conscious. People didnt "want" to think that there were giant holes in space that violently sucked in anything that came near it, but they were forced to, because general relativity was mathematically correct. Refute the theory, not the implications.
@ericweiss8264
2 жыл бұрын
Integrated Information Theory can hardly be called a "theory". It's just a proposed formula for calculating how "conscious" a given system is. That's it. Scott's argument is: If this particular attempt at defining consciousness says that a regular grid of XOR gates is highly conscious, do we still believe that this definition is on the right track?
@connorschmidt5945
2 жыл бұрын
@@ericweiss8264 It is not enough to call a theory (or a definition if you want to call it that) wrong because it is "not on the right track." It is true you can refute a theory if it implies something that must not be true, but what argument is there that a grid of XOR gates MUST not be conscious? we cant just take things for a given without logical justification. At the end of the day, we know that the brain is a system of logic gates, and if consciousness is inherently rooted in these types of networks, then there should be all kinds of conscious structures you could build out of logic gates, and the math shows this must be one of them. in order to refute this argument I'm making, one would have to either claim that consciousness is not rooted in the networks of the brain, or that the math of IIT is wrong. otherwise we have no choice but to accept whatever radical implications the theory has, whether that includes dogs being conscious or certain systems of XOR gates in a grid being consious too.
@danielsayre3385
2 жыл бұрын
@@connorschmidt5945 Yeah I think that this concept is incompatible with mind-body dualism, the belief that there is more than one plane of existence and your mind exists independent of this plane of existence, so it wouldn't probably fly well with a dualist (like Christians). But if you're an incompatibilist, in that you believe that free will is incompatible with determinism, and a determinist (think Laplace's demon, the idea that if you could "know" all current phenomena you could predict all future phenomena) then this IIT theory can still hold value and predictions. In my opinion. Some people may not want to take on an opinion like this because of the implications it has on their own perceived consciousness and freewill. That's fine, but if your definition of consciousness is axiomatic (watered down to the fundamentals) then you might get a result from that axiomatic definition that tells you a set of logic gates are equally or more conscious than an individual human. This is probably out of my wheelhouse, so feel free to critique my comment lol
@connorschmidt5945
2 жыл бұрын
@@danielsayre3385 I agree with most of what you say here. first of all, IIT is derived from an axiomatic approach drawn from fundamental observations about consciousness. second, IIT does imply that mind-body dualism can't exist in a strict sense, but if a superior theory contradicts your pre-conceived opinions, then you should cast aside those notions if you can't refute the theory. Lastly, there is research done by the team working on IIT that shows that based on certain definitions of free will, IIT says that there is in fact free will in both deterministic and indeterministic universes. I could find the paper for you if you want.
@arpitkumar4525
Жыл бұрын
IIT fails because it chose the incorrect definition of consciousness. I would define consciousness as the ability of a probabilistic system to be aware of it's own states at every instant and have control over the next move based on the current state. The error correcting code is highly integrated and thus is highly aware of its own states but not at every instant. The error correcting code is not aware that it is aware of its own states so it does not satisfy "at every instant" part of my definition. Also it is not a probabilistic system. By probabilistic system I mean a fuzzy system that works on heuristics and tries to maximize a heuristics based on the probabilities. Like a chess engine. A chess engine is very close to conscious.
@amartinezsilberstein
Жыл бұрын
Aren't you mixing up consciousness with free will?
@arpitkumar4525
Жыл бұрын
@@amartinezsilberstein Haha. Just literally 3 days ago I was wondering about free will and came to the conclusion that the problem of free will is same as consciousness problem. And then I thought it all depends how you define it and what is the scale of your system. But yes I would define consciousness the same way as free will. Both are illusions/software
Пікірлер: 41