The Mathematics of Consciousness (Integrated Information Theory)

Entry for the #3Blue1Brown Summer of Math Exposition 2022 (#SoME2)
by
Rodrigo Coin Curvo
&
Alexander Maier
Read more about Integrated Information Theory and the #neuroscience of #consciousness: www.scholarpedia.org/article/I...
Also, check out Rodrigo's entry to SoME1, which goes more into detail regarding the math behind IIT:
• Integrated Information...
----------------
Attributions:
Icons from flaticon.com: scissos, newborn and world-book-day (by Freepik); AI and bedtime (by photo3idea_studio); brain 1 and 2 (by Smashicons); deer (by tulpahn).
Images from Wikipedia: Giulio Tononi, Kristof Koch, Ernst Mach's Innenperspektive and Indian elephant.
Soundtrack by Alexander Maier
----------------

[@Anon282828]

It would have been nice to start with a definition of "consciousness" - as it is presented, the system under consideration could just as well be the sun and the planets, and the inputs / outputs their gravitational interactions / motions. Yet we would not define the solar system (or galaxy) as a conscious entity.

[@khatharrmalkavian3306]

Are you sure? There are theories about consciousness mediated by stellar objects and phenomena.

[@astonishinghypothesis]

The definition is in the axioms. That is, collectively, the axioms aim to describe conscious in its most basic properties. So, yes, if the solar system integrates information, it would be a minimally conscious system. This is an unlikely outcome of the calculations, though. Note that the video mentions a second step, where the probability distributions have to be recalculated after removing inputs and outputs to each mechanisms). This is where the information that is quantified in the first step gets tested for being integrated or not. That is one of the counter-intuitive consequences of the theory. The fact that non-active neurons are as consequential as active neurons is another one. Without such counter-intuitive predictions, the theory would be uninteresting from an experimentalist view. We WANT to show that the theory is wrong, after all (we just have not yet succeeded in doing so).

[@matbmp8996]

@@astonishinghypothesis Then why use the word consciousness? How does this theory relate to consciousness in psychology and neuroscience? Was integration experimentally measured in humans? Does integration imply consciousness and does consciousness imply integration? Is integration both necessary and enough for consciousness? Some articles would be helpful in clarifying this to me.

[@astonishinghypothesis]

@@matbmp8996 Consciousness here means "subjective experience", or "qualia" to use a philosophical term. Some psychologists agree with that notion, others like to add "extra stuff" like memory, cognition, self-awareness or language. And yes - IIT says that consciousness is "just" integrated information. And whenever a system produces integrated information it is conscious, So, the universe is filled with many, somewhat conscious systems. Although both the "information" and "integration" in IIT deviate from the way we use it in everyday language and other fields, such as computer science. Here are some papers you might find interesting. Applying the math to neural data lets you predict "awake" versus "anesthetized" states, for example: pubmed.ncbi.nlm.nih.gov/33635858/ And that's applying the actual math outlined in the video. On a more qualitative level, the evidence supporting IIT is even more striking - to the point that it has become clinically relevant (for diagnosing coma/vegetative state): pubmed.ncbi.nlm.nih.gov/16195466/ pubmed.ncbi.nlm.nih.gov/19818903/ pubmed.ncbi.nlm.nih.gov/24403317/ pubmed.ncbi.nlm.nih.gov/27717082/ pubmed.ncbi.nlm.nih.gov/22226806/

[@pyropulseIXXI]

@@astonishinghypothesis There needs to be another axiom of consciousness, and that is to be able to impose an action without a first cause. That + subjective experience, is consciousness

[@jonrhaider]

I don't know this for a fact, but it seems like the next step to determining or evaluating consciousness would be to take these 5 axioms and apply them not only at the individual node level, but to clusters. Clusters of neurons should represent concepts, or rather slightly more complex information units than neurons, but the mathematical logic would have to apply the same way, just building on a higher level abstraction, i.e. more complex neural structures. There would have to be some threshold though before which the abstractions are simply not complex enough to allow for consciousness, i.e. self awareness.

[@pyropulseIXXI]

This is a missing axiom, and that is the axiom of action without cause

[@pyropulseIXXI]

Your error is thinking consciousness is an emergent phenomena. Neurons just allow a consciousness to control a physical body in the physical world. Neurons are not, and will never, be conscious. They are the physical mechanism of which allows a consciousness to impose action with a first cause

[@jeremygodoy18]

its about the wavelengths bro

[@trudojo]

@@pyropulseIXXI every theory known to man is plagued by this actually. The great contradiction in logic. if things happened, how? if not, how?

[@NicoAssaf]

@@pyropulseIXXI That kind of axiom would add an acausal element to consciousness. How do you justify that, especially in a scientific context? Things happen because of reasons, so why would that not be the case for our wills? What makes consciousness so special that it can override causality?

[@TheJoyLoveShow]

You are brilliant and your work is thoughtful, beautiful, and I hope you continue explaining the additional axioms and postulates. This is an unbelievable treasure. Truly thank you and I am eagerly awaiting your future videos. 🙏🏻🙏🏻🙏🏻 I wish I could help you make them!! They are a gift to the world.

[@AndrewUnruh]

Me too! I am recently retired (acoustics and algorithm development), but this is just fascinating! Going a bit into magical thinking, I cannot help but wonder if there is some possibility that all of life is somehow connected in such a way that a separate consciousness is created. I understand that this idea is simply born from intuition and wishful thinking, but perhaps the same kind of approach explained in this video could be used to explore the idea. That said, the idea is really not fully compatible with IIT.

[@iansteel6403]

As someone with a minimal background in stats, IIT, and quantum mechanic as well as a small interest in spirituality and consciousness, this was extremely palatable and perhaps one of the only ways I would be able to understand the topic. Thanks!

[@asdf8asdf8asdf8asdf]

“Existence exists” “Consciousness is conscious” - John Galt

[@30ftunder39]

Roger Penrose is the only one, to my knowledge, that implicitly gave a definition of consciousness that is not self referential: Consciousness is a Turing machine whose input is dependent of its own output. He didn't say it like that, but close. This definition violates Turing's Halting problem since the output lays in the future and is therefore unknowable. This definition is NOT self referential, but it implies retro causality and something about the nature of Time..

[@asdf8asdf8asdf8asdf]

@@30ftunder39 well that's pretty damn interesting tbh. Maybe I'll have to finally crack open his "emperor" I do think consciousness is cleverly (very) in Penrose's definition by citing a machine test which itself embeds consciousness. Kind of a little literary logical homunculus there on Penrose's part

[@futurehistory2110]

Consciousness is an incredible thing. One thing that's so mysterious is that it does not appear to be a part of objective reality and yet our own experience of it proves that it exists. If you say that your consciousness does not exist then you are not experiencing yourself reading this right now - disagreeing with that statement requires consciousness (and we get back to 'I think therefore I am'). I suppose either this is part of objective reality (our conscious experience) but in a way that we do not understand or we need to expand linguistics and concepts of 'existence'.

[@Starcell170]

Awesome! Clearly explained!

[@robcarl1100]

Good presentation. Not convinced the model is correct as reasoning seems a bit shaky, but you did a good job breaking down the basics thank you.

[@albertmashy8590]

I was waiting for something exactly like this. You are a top G

[@hertzfall0]

I am of the veryyy strong opinion that Ludwig Wittgenstein described how an organism functions in his Tractatus when he proposed a self-referencing and dynamic system of logic. Basically consisting of variables and operators that can change according to the type of information it processes, changing the way it processes the information, changing the ground assumptions for other sets of variables also contained within said system. Very beautiful descriptive work from 1921

[@astonishinghypothesis]

"I could not make up my mind whether he was a man of genius or merely an eccentric.” B. Russell

[@BrainOfAPenguin]

Hi, I came here from the SoME2 judging, and you have a very interesting video. I really enjoyed the intriguing opening of your video that hooked me in. I also was fascinated by all of the math that went into a seemingly unrelated topic. Definitely earned my subscription!😁

[@davejacob5208]

this video lacks an end-section that links the explanation back to the question at hand: what exactly does this tell us about how to "measure/quantify consciousness" instead of just giving us quantities linked to single neurons etc. ?

[@CreationTribe]

This is an interesting theory. That being said, it has received quite a lot of criticism - much of which is fairly valid. For instance, the fact that non-active neurons are necessary implies a fairly large number of problems that have not been dealt with. For instance - this would assume that if you took the active brain state of some random human and duplicated it in another brain that only had those neurons which were active, that this duplicate would not be conscious even though it consists of exactly the same activity. #AnilSeth covers this and other issues with IIT in his book, "Being You: A New Science of Consciousness".

[@rysw19]

I think IIT has problems as well but that one is a feature, not a bug. There is no reason to expect that if you chopped out the “inactive” neurons the system would have the same properties.

[@good.citizen]

thank you ◼️ excellent graphics

[@AndrewUnruh]

OK, I have a question, which may be very naïve, but hopefully it makes sense. We know we are conscious. Now, take a person and insert them in a perfect sensory deprivation tank so they have see nothing, feel nothing, etc. Are they still conscious? Of course. Now, let's take it a step further and assume they can shut down self talk, self imagery, and all thoughts. Would they still be conscious? I think so. But what would they be conscious of? I think they would be conscious of the passage of time. It wouldn't matter if they could accurately determine how much time is going by, just that they have some sense of time, even if they were permanently in the 'now'. Could this be the minimum conscious system? What is the simplest possible system that could perceive the passage of time? What would be the PHI of such a system? Oh, and does it even make sense to say that in order to be conscious, one has to be conscious of something? BTW, I'm not one of those crackpots who claim to know more than the experts. I am just starting to try to understand what the experts are saying on this subject and I find it fascinating, even if it is at the edge, or maybe beyond the edge of what I can comprehend.

[@astonishinghypothesis]

This is a very thoughtful question. Note that IIT dissociates contents of consciousness (what are you conscious of?) from the amount of consciousness (how much are you conscious?). In your thought experiment you reduce conscious content, but not necessarily the amount of consciousness. In other words, you leave the observer at full consciousness (high PHI), you just empty it out of content to be conscious of. Meditators seem to seek this state of pure consciousness. However, this state seems radically different from a state of low consciousness (low PHI). Low consciousness might be closer to the transition between sleep and wakefulness or what it was like to be an embryo or baby.

[@RogerValor]

Seems to me the theory is retaining information in a system being quantifiable could give a method for measurement of data processing in an unknown amount of mechanisms - but that seems ridiculously far away from consciousness itself - but it could be a start, or helpful tool in the box, to measure pathways for an unknown neural net. It still sounds on itself like a random stab in the dark tho in the greater picture of things

[@Self-Duality]

Thank you for releasing these thoughts 😌💭

[@rickyoon1446]

This was a very interesting and informative video, thank you! I was just wondering, were arbitrarily assuming that the mechanisms(the 4 neurons in the model state) behaved as “AND” gates when you explained the cause repertoire of mechanism X_1?

[@TrendElement]

Yeah we would be able to figure the exact flow in measurements of energy required by a body to function at any time then would it be possible to make an external source of energy to fuel a body or boosts the brain constantly?

[@HUEHUEUHEPony]

this video felt like a fever dream

[@AndrewUnruh]

Thank you! This was a wonderful explanation. Is there a follow-up video? I have a technical background (physics, acoustics, algorithm development) but I don't think that this video alone is sufficient for me to start writing code to calculate phi.

[@astonishinghypothesis]

There already is open source code to calculate phi: github.com/wmayner/pyphi

[@MichelleHell]

Maybe consciousness is just the CNS making decisions. It is wild to be able to weigh options in the brain and then trigger a flow of behavior based on the weighting. We call this consciousness, but all chemically active systems are spontaneous and all biological functions follow the laws of thermodynamics. To be a cellular sized chemist, you have to be able to manipulate matter such that a spontaneous process occurs. But the chemist itself is acting spontaneously, governed by the same laws it masters. Having studied biochem, I realize how limited language is in understanding how everything fits together. The word consciousness itself is limiting because it causes us to look for non-spontaneous processes and come up with theories based on perception. The beginning and end of everything in the universe is thermodynamics, which requires a hard scientific approach where you can show the chemical equations involved in the brains decision making process. It's not enough to say consciousness is who you are when awake and aware. Consciousness is a highly complex biochemical reaction, involving trillions upon trillions of cells perpetually cascading reactions to form a macroscale loop that is able to reference knowledge and information in order to make decisions. Any real artificial intelligence will have to be at least as complex as the biochemistry of our brain and CNS. Information technologyl efficiency today pales in comparison to the efficiency of cellular activity mediated by enzymes. When you actually study the biochemistry, you realize how incredible is nature. An artificial biological system built from synthetic enzymes could be a reality one day, just a brain that is connected to the information of the internet and begins the synthesis of new thoughts and ideas. Either way, any AI that doesn't use biological systems is inefecient and ancient in comparison to billions of years of nervous system and enzyme evolution.

[@daanielacosta2395]

I found very interesting your points, have you ever read or search about Marvin Minsky theories of mind and consciousness? If not I would recommend you to take a short/deep dive, I'm not saying that you are wrong, just that maybe you found them interesting. :D

[@MichelleHell]

@@daanielacosta2395 I'll check him out!

[@mastershooter64]

But how does a highly complex biochemical reaction lead to conscious thought? like you and I can think about stuff, we can experience stuff, we're self-aware, how does a biochemical reaction (insanely complex though it may be) lead to being aware about one's own self? I'm not questioning it, since that's basically what's happening, but there's a missing link in our explanation, how one thing leads to another is unclear, but I'm sure we'll figure it out soon! go science!

[@Zed871]

Yas child!

[@whatistruth_1]

This comment is going to age so poorly

[@rpaleg]

Why do we automatically assume the property of consciousness is binary? Is this actually true? Should we consider the possibility that consciousness is a spectrum, where some brains are more conscious than others?

[@astonishinghypothesis]

Well, yeah, the outcome of the equations explained in this video is continuous. That is, if integrated information is consciousness, then it varies on a spectrum. So, you seem to agree with the theory - at least in principle.

[@hyperduality2838]

@@astonishinghypothesis Cause (thesis) is dual to effect (anti-thesis) -- Hegel. Integrating information is a syntropic process -- teleological. Concepts are dual to percepts -- the mind duality of Immanuel Kant. "Always two there are" -- Yoda. Problem (input), reaction, solution (output) -- the Hegelian dialectic. Vectors (contravariant) are dual to co-vectors (covariant) -- Riemann geometry is dual. Curvature is dual.

[@gabrielbartmanovicz4284]

Wich software do you use to make your videos?

[@businessfreedom4321]

Guys continue this work...this is really intresting stuff and becouse I have got this in my feed probably also yt algo "think" the same XD

[@zachm5136]

That was the fastest 18 minutes of my life. And just when it ended, I thought it was just getting started. Bummer.

[@QuickProgramming]

very nice. learnt something new :)

[@weirdsciencetv4999]

I have to applaud you on taking on something so ambitious. With regards to the axioms, the only way to really understand them is to retrace his work from the beginning, he’s got a few books out there. I think he could have done a better job being more clear about the axioms. I might attempt a video myself at some point, maybe even a worked example network, some source code, etc.

[@astonishinghypothesis]

Thanks for your kind words. Can you expand on what you think we got wrong about the axioms?

[@weirdsciencetv4999]

@@astonishinghypothesis For example, when people simply condense Exclusion axiom as just “consciousness has boundaries” lacks nuance. Exclusion is more that there is not a layered experience of consciousness. It’s not the superposition of multiple experiences, or some combination of partial experiences. I think people like to try and condense it to just “experience has borders” misses quite a bit. Some of this was born out research in split brain patients. When the brains are connected IIT would predict one consciousness due to the level of causal influence each side has on the other. When the links are severed IIT would predict two distinct consciousness in one skull, but with independent experiences

[@astonishinghypothesis]

Good point. I am pondering a longer video series that goes more into depth. One video per Axiom perhaps. That might do it more justice. Also covering the historical roots (starting at the work Giulio did with Gerald Edelman and defining "complexity" via Shannon Entropy). Ideally, all that culminating in a workshop on applications (such as how to use PyPhi on neural data). You said you might want to so some videos, too, so please feel free to hit me up if you are interested in collaborating.

[@weirdsciencetv4999]

@@astonishinghypothesis I am seriously considering it. I am disabled by chronic pain so I have to carefully dole out what energy I have during the day. Where can I find your contact details? Some youtubers have a “for business inquiries” link, I will check there. In this case it’s not business, just seeing how we can work together maybe. You should definitely expand this video, I think you could do it real justice!

[@Zed871]

@@weirdsciencetv4999 I'm wishing a healthier and stronger future for you.

[@kato_dsrdr]

This is currently the most convinving theory about consciousness at least for me.

[@havenbastion]

Global Workspace Theory is better.

[@alph4966]

@@havenbastion Counterfactual Information Generation Theory Free energy principle(+Predictive coding) Global Workspace Theory(Whole brain integration model) Integrated information theory The neural circuit modules that include all of the above, and the system that combines them, will have consciousness.

[@havenbastion]

Neurons being in an on or off state is a woefully low-resolution understanding, ignoring the action potential and the strength of the actual firing, which has downstream effects.

[@christopherrice891]

Ah, here we are! This is the sort of path in Mathematics i am looking for. The abstractness dial turned up enough notches so i can see my thoughts on paper instead of tossing and turning all night dreaming of what i want to see but just can't find the right words.

[@futurehistory2110]

Also, it's very out there but what if consciousness is inevitably existence experience itself resulting in all conscious experiences being that of existence itself and then existence breaks this up into a linear subjective timeline (i.e. 'we', 'us', existence itself experiences itself and goes through various lives).

[@marcianito777]

If you want to work on a cognitive aspect that requires awareness such as attention, how can it translate into a causal structure? What kind of graph should be built? Is there any kind of evidence? How can the representation of a system be done

[@astonishinghypothesis]

Attention might not actually require awareness: pubmed.ncbi.nlm.nih.gov/33034851/

[@entropica]

Nice video, but I find the background music somewhat distracting.

[@pectenmaximus231]

Hopefully without sounding like I'm on a high horse, I think "The" is a bit loaded in the title. To suggest a natural system has a single defining mathematical representation really gives off the impression to people that the relationship of mathematics to the world isn't about models and approximations, but is instead one where we are always peeling back the fabric of the senses to reveal some kind of 'universal truth' that is encoded in mathematics. That is a very tempting view of mathematics, but it isn't borne out.

[@astonishinghypothesis]

I admit that I lean towards that (Tegmarkian) view, but it is not required for the theory to work. IIT really is no different than quantum mechanics or general relativity. All these theories are just a bunch of equations at their core. Whether or not there is more to the fact that very high level, abstract math better describes our scientific observations than our intuition produces interesting philosophical byproducts indeed. But the theories themselves do not make any ontological claims.

[@adelatorremothelet]

sublime!

[@johnchessant3012]

very cool

[@aykay1468]

Very interesting theory! In mixing math with philosophy there's always the risk of association with new age obscurantism (mainly in the appearance of analytic formality to justify a contentless scientific) but here I do think it's a well done mix! If I'm reading axiom #1 correctly, integration (or the complex subordination of subsections/systems of 'neurons' to the whole) is the key to consciousness; and probably less in a threshold-complexity for us to neatly define something as conscious, and more so in a spectrum? Whereby more integration generally equates to more consciousness? I don't know if that part was explained. I do have some questions. On the more mathematical side, is there a reason why Neuron 4 only transmits outputs to Neurons 1, 2, 3, but not to itself? I'm wondering if it's a question of simplifiying calculations, or preventing feedback loops (though I think the logic gates would mostly solve that). On the more philosophical side, just some notes on the axioms. #2 is just the linguistic/structural concept of Determinate Negation: all instances of "red" can only be perceived in contrast to the background of "green", "blue", "yellow"... in sum, all "non-red". Any object exists only as separate from its others. This is a very good information model to explain our senses, maybe even using analogies to how pixels on a screen form, according to neighboring similarities and differences, cohesive and distinct images. But where that would be fine for empirical information, I don't know if that quite holds to consciousness. One of the marked qualities of human subjectivity is in our capacity to abstract (maybe to integrate) tons of data into small notions (signs, ideas, names). The same logic doesn't apply when we pass from Determinate Negation to Differentiality: our notions of Red are not merely defined in contrast to others. That is, Red isn't just non-green, non-blue... it becomes also non-red. That is, at the level of consciousness, of imagination, beyond what is just seen, it is impossible to think the notion of pure red. We can think of examples of things that are red, but those are all already "impure", necessarily not red at a certain point where they have countours, borders, either in space or in time. What this results in, is the main problem structuralism faced (which, along with the critique of the signifier chain, led to the rise of post-modernism as a movement), whereby a notion like red is predefined by itself: a tautology. A more intuitive way to think this movement is to follow the path of determinate negation to its failure: if you define an elephant by all the other animals it isn't, like a tiger, now you also have to define a tiger by all the other animals it isn't... including the elephant. To define an elephant requires the definition of a tiger, and vice-versa: it's a catch-22, a gap that consciousness somehow crosses while creating language, in something like a leap of faith. From a purely structural point of view, language should be impossible. Either it should be complete and inconsistent, or incomplete and consistent. Sometimes it's the worst of both worlds, but it still works, and consciousness has to explain why. This is why I think letting neurons directly affecting themselves might help in the model - because the tautology of feedback loops might actually be beneficial to account for the less rational parts of thought. If not through that mechanism, then through some other that seems redundant (unless that's covered under integration and I missed it whoops) On that note, I'd also like to point to the contradiction between axioms #4 and #5. Composition (establishing space and time, the conditions for a proper subject/consciousness) and Exclusion (the subjectivity to which these dimensions always seem subordinated). This seems to me like the philosophical (and often psychoanalytical) distinction between consciousness and unconsciousness (which might as well be included in a broad theory of thought). There's a gap between these axioms, discovered in Kant, Lacan, and others. The clearest example is in a thought experiment almost as old as philosophy itself. (a) All men are mortal. (b) Plato is a man. (c) Plato is mortal. This seems a trivial show of logic, but psychoanalysis suggests that it is so only while sustained by an underside of irrationality. Part of the subject does consciously discover the abstract notion of mortality after enough pattern-recognition. But there is always an unconscious part of us that believes in immortality: because from our point of view, death is something that will never arrive (or if it does, it will not arrive to "us", since by then we won't be conscious of it). In our side which strives for rational composition, we can imagine a world outside ourselves, with set coordinates of space and time which will precede us and be here after us... but in terms of Exclusion, our point of view is always fixed, always in the place we look and in the present, an exception point where consciousness fails, that is nonetheless integral to consciousness itself. A notion contains its own negation, just like how the red requires a contrast with non-red. The problem, then, is not that I think we should choose axiom 4 or 5, one to be privileged over the other. No, their problem lies in the axiom #3. Integration. Consciousness, in my opinion, is a construct that deals with this gap of imperfect information (tautological, irrational). It is the very failure to integrate properly, and thus taking shortcuts (the use of language groups all of a same kind of object under the same sign, complex emotions condense varied calculations of interest that we may not even know about, general pattern-recognition trades-off for less specific singular memories) that makes us human subjects. Anyway, pretty good video. Um, I hope you weren't expecting a concise comment section with a video subject like this! My interests are usually more on the philosophical side than mathematical or neurologic, but this was fun! I'll be checking out other videos as well. Cheers!

[@alphacube5398]

Excuse me do you have any social networks we're I can contact you to have a chat

[@simonahrendt9069]

Thanks for taking the time to write such an interesting comment!

[@SebastianSchwank]

Sorry, I can't spot time in this equation. Any further improved versions available?

[@mushroomxxxzzz]

Time isn't relevant there.

[@param888]

can you please define what is interaction as per this system?

[@astonishinghypothesis]

at the risk of oversimplifying, the tl;dr is: "causal interaction" = a part of the system has the power to change the system as a whole This can be quantified by experimentally or computationally eliminating the part and computing the change in past and future probability distributions of system states that might result because of that

[@nelsonberm3910]

Thank you!

[@realjarz1230]

If you’re a consciousness nerd like me look into the bicameral mind it’s very interesting it talks about consciousness originating from language like your biology is a container of data and your experiences give you a certain preset and things like language and society creates more complexity within those patterns that you form from both thoughts and actions

[@36cowboysintotalatramranch]

No, the bicameral mind is quite easy to falsify. It's not correct. Don't just be a consciousness nerd, read more than Wikipedia

[@realjarz1230]

@@36cowboysintotalatramranch thanks for your input😂 because you clearly know more than I do can you educate me on why it’s so easy to falsify because my Wikipedia knowledge doesn’t suffice so it seems

[@realjarz1230]

@@grapesurgeon 100% man at the end of the day the pursuit of knowledge is because of the lack of but as you learn more you learn that somethings are out of your reach but that’s why it’s important to respect others and work with each other

[@hyperduality2838]

Cause (thesis) is dual to effect (anti-thesis) -- Hegel. Integrating information is a syntropic process -- teleological. Concepts are dual to percepts -- the mind duality of Immanuel Kant. "Always two there are" -- Yoda.

[@silvomuller595]

Best explanation of IIT on the internet I guess.

[@Thejosiphas]

this is equivalent to composition of functions on bitstrings/permutation composition

[@astonishinghypothesis]

Interesting. Can you elaborate?

[@TrendElement]

I wonder would it be possibly to make a face robotic face and attach it to the back of your hand like a mask but connecting to the brain in a way maybe nonsurgically or I don’t really know what they’re for your brain to adapt and have 3-D view more like in a game or something possibly?😊

[@pablolecce6931]

In other words, what you said is that the current state of neural networks is capable to develop consciousness?? And that the attention mechanism is the base for the integration in consciousness math? Amazing!!

[@astonishinghypothesis]

Actually, AI as it stands does not integrate information to any remarkable degree (and thus is predicted to be unconscious). More precisely, the von Neumann architecture of personal computers does not allow for massively integrated information. It will probably take a neuromorphic chip for that.

[@pablolecce6931]

@@astonishinghypothesis Then maybe I misundertood something. I will see your video again to try to find it out. When you said "Integrate" do you refer to the mathematical idea of integration?

[@johnfitzgerald8879]

I don't know why this idea of our conscious experience is so confounding for so many. Consciousness arises because our brain is a feedback control system and retriggers the same pathways that are triggered by external stimuli. In the detail of the neurons and structures of the brain, the stimulated pathways are complex. Sensory information is processed. In a feed forward system, the output goes directly into action, triggering muscles and movement. In a feedback system, that output is fed back into the pathways of sensory input. We see images and hear our inner voice because the processed output is fed back to the very pathways that are excited by external stimuli. The result is we have an internal experience of what the sensory input has triggered in it's association to previous experiences. We see a dark spot on the wall and it triggers the pathways that were imprinted from out experience with flies. That result is fed back as if it were sensory input. We have a conscious internal experience of a fly. This mapping of the output back into the input maps out internal representation back into the same system. Now, this is a little over simplified in that the feedback doesn't all combine into a neat bundle of neurons that connect directly into the optic nerve, but rather the feedback occurs all along the sensory and processing pathways. Never-the-less, at it's base, consciousness is simply the result of feeding the output back into the input. And, notably, out internal experience is not so intense as to be of greater intensity than our external experience. That isn't to say that it never is. PTSD episodes and just simply getting lost in a daydream are examples of when our internal experience becomes more intense than the information that triggered it. We have a conscious experience because the brain maps the output of processing back into itself along the same pathways where external information is being mapped. The brain doesn't need a special mechanism. There is no "seat of consciousness". We have an internal conscious experience where the feedback stimulates the same pathways that create our awareness of the outside world. That seems like the best place to start.

[@RavenAmetr]

" We have an internal conscious experience where the feedback stimulates the same pathways that create our awareness of the outside world." Seems super easy, right? But what is this "we" who "have" the experience? The need for this "we" is not obvious to produce the same awareness without any experience. There's no colors in the "real" world, then your "red" might be my "bitter" or something entirely different, considering that our brains are different on the ground level. That's why it is called "The Hard Problem". Any emergent phenomena is an illusion, useful (like energy in physics) or just curious (like patterns in the Conway's game of life), and if "we" are the illusions in the same way, who/what is the deceived witness?

[@mastershooter64]

"We have an internal conscious experience" how does that happen? that's the confounding part, not the biology or biochemistry behind it. The confounding part is how exactly does the biochemistry lead to a bunch of chemical reactions being aware of itself? that's insane if you think about it, just a bunch of chemical reactions being aware of itself

[@user-sl6gn1ss8p]

​@@mastershooter64 I think John's point is that how it happens is ultimately the same as how external world perception happens. Perception about perception is still perception. That seems to point to the "real" question being how does *any* perception "appear" as "experience" - the question wouldn't be so much about consciousness as in the "inner voice" which is reflective of the outside world as well as of itself, but about the "deceived witness", as Alexey put it above. These things are often conflated, but they aren't necessarily the same (although I'm sure one could argue they should be)

[@johnfitzgerald8879]

@@RavenAmetr Well, so start with, those are generic usages of "we", not suggesting anyone specific. Simply referring to a generic individual, people in general, human beings, for the purposes of discussion. I am not exactly sure where this is creating a difficulty. You have a conscious experience and I can be certain as to how that may be presented to you. While people are all certainly unique, we are not that unique so as to be completely and entirely different from other people. We do share a lot of biological and psychological characteristics. Well, the internal expression of conscious experience ranges from person to person. Some people have continuous internal dialog while others report none and all and find the idea quite surprising. Some people report not seeing internal imagery, most do. And, of course, it varies from individual to individual. Never the less, what is clear is that conscious experience is presented as if the person were hearing, seeing, or feeling that is clearly not from external experience. "There's no colors in the "real" world, then your "red" might be my "bitter" or something entirely different, " I have no idea what you are referring supposing here. There are colors. Smell/taste and vision are entirely different sensation. We aren't talking about semantics, such as when I imaging the color red, I say it is "bitter", that's a language issue. If a person is color blind, then they would likely not have an internal experience of colors. Their conscious experience would lack it. Temperature is an emergent phenomena It is not an illusion. It is a measure of physical properties . The same can be said of energy. Energy is not an illusion. The whole point of physics is to identify real measurable properties of the universe. Emergent or not, they are still real properties, not illusory. So, I am at a loss to understand what you mean. I'm really not interested in playing games of semantics and feigned ignorance.

[@exsolutus1626]

@@johnfitzgerald8879 But why is there an experience at all? How does a causal feedback reaction, however complex, create an experienced internal phenomenon rather than an unexperienced, however complex, causal response?

[@TrendElement]

Can all of this leads to a point where we can annex enhance the overall evolution of humans for may be centuries in capsule of just few years?

[@spiralsun1]

This whole video can be broken down into the minimum-information partition necessary to understand PHI. 👍🏻 You rock.

[@jasonmerkji]

@astonishinghypothesis Can you please tell me what mathematics is used in IIT? Is it network theory? What maths do I need to learn to understand IIT mathematically?

[@astonishinghypothesis]

All one needs to know for understanding IIT is how to add, subtract, multiply and divide. In order to understand the diagrams that are used to explain the equations, it can be helpful to know a little bit about graph theory, probability theory (esp. Markov chains) and Boolean algebra. But none of that is a pre-requisite.

[@jasonmerkji]

@@astonishinghypothesis would it be enough to learn how the equations of IIT, that you mentioned, were derived.

[@futurehistory2110]

Sometimes I wonder if what really exists is conscious experience in its own right and as discussed in this video, these conscious experiences are integrated creating the illusion of the self. Note, not an illusion of consciousness but conscious experiences exist and they create the self, with that being the illusion.

[@TrendElement]

Well this video just made me think from the beginning of something maybe knew I didn’t thought of before. So what if basically you can’t move I can’t see you can hear or sense practically anything more likes when you are in anesthesia basically your brain doesn’t dies and your organs still function so there you are alive and your brain is active but you don’t have access at least afterwords of what the brain in your brain practically was registering like maybe feel something or anything. Maybe it’s because of the anesthesia and the rain is not 100% awake. My question is can you live just inside your brain like meditating to a point where your focus is so concentrated in your brain and your links to that meditation more? Or maybe waking up from a coma then that person can recall and remember everything the person heard around while being home or felt? This is basically like a sci-fi movie Where are you fake your death and remember everything that’s happening then reboot yourself and analyze the data… As much as we in the human race no until now is this even close to being possible or is it possible already in any capacity? Can we access links between our brain and the other organs well we do but we don’t have exact control of may be what dosage of a substance or brain to produce or how many oxygen to store in our plant for if we want any vitamins stored in our bodies bodies more or less or just select a few….

[@rysw19]

I suspect IIT is something like a necessary but not sufficient condition for consciousness. But I do think it gets to something important about the cause-effect structure that IIT attempts to get at that is important. One other weakness or at least misconception about IIT is that because it’s mathematical, its structure is set in place. That’s not the case. There are number of ways to measure a variety of quantities that are something like Phi in IIT (the measure of integrated information) for example mutual information. All of those measures you probably will find to be high in conscious systems and low in (most) unconscious systems. The framework they have is by no means set in stone. I know they actually changed the distance measure to Wasserstein from KL.

[@astonishinghypothesis]

Yes and no. Mutual information (and many other measures of "complexity") are strictly pairwise. IIT is not. If a system consists of 3 or more parts interacting in a way that cannot be predicted from their pairwise components (think: oxygen, octane and a spark meeting in a combustion engine), pairwise techniques fall short. The video mentions that the math is specific for IIT3.0 (which replaced KL with the EMD). IIT4.0 changed that yet again, now using an equation that is similar to Shannon entropy, but uses both likelihood ratio and maximum (most informative aspect) instead of the sum (average unpredictability): pubmed.ncbi.nlm.nih.gov/33139829/ Theories need to evolve as we gain more empirical insight. With each iteration of IIT, the changes become smaller and smaller. That's quite similar to say, the shift from Darwinian evolutionary theory to the Modern Synthesis or the addition of inflation to the big bang model.

[@rysw19]

@@astonishinghypothesis the point is that there is some level of arbitrariness in the exact calculation. The fact they’re onto Shannon entropy is just reinforcing that. There are many to ways compute metrics that will tell you roughly the same thing and no clear way to say one is better than all the others. I’m not saying that’s a fatal flaw, I just think that IIT is often sold as “mathematical”, which gives an impression of precision that isn’t actually the case. It’s not physics

[@astonishinghypothesis]

That's fair. Whenever the units change, the absolute values change also. Though the latest metric (iBit) is somewhat better motivated in that it builds on the same axioms (esp. exclusion) as the rest of the theory. The main issue with IIT lies less in the derivation of its formalism but in computability. Any empirically established TPM can be questioned as it might be imprecise due to noise. And computing PHI for systems with more than a very limited number of components (single digits) exceeds readily available computational resources. That's why it is noteworthy that indirect ways of testing IIT such as the PCI have been massively successful. There are patients in Italy who would have been misdiagnosed as being in a vegetative state while they were actually conscious if it weren't for IIT: kzbin.info/www/bejne/j6e4l2Z4Yp55ftk

[@rysw19]

@@astonishinghypothesis not to argue but I actually exactly disagree with your point. I think the most obvious issue with IIT lies in the fact that the postulates aren’t uniquely derivable from the axioms. So the exact formal structure is somewhat arbitrary. I actually think that IIT picks out important features that are ignored by other approaches, but that the formalism isn’t necessarily a model of the phenomenon. In the same way that Ptolomy’s epicycles were useful for course grained calculations but didn’t actually map to the ontology. The issue of computational tractability is basically generic to all physical systems, I don’t consider that a weakness. However, I have a feeling that you could make some progress using some kind of analysis of the limiting behavior. For example if you found a section of the brain that largely repeated the same structure, you might be able to show how the measure of phi grows asymptotically with the amount of material.

[@astonishinghypothesis]

​@@rysw19 That's interesting. I largely agree with all that. And from what I can tell, probably most people with an interest IIT would. You might be surprised by some of the debates, say, on the arbitrariness and redundancy of the exclusion axiom, that fuel the development of IIT. The very fact that IIT gets updated regularly signals that it is not meant to be unshakeable theory cast into stone. IIT just marks a first attempt to formalize the science of consciousness. To move from a theory-free search for correlates to a theory-driven work on testing quantitative predictions, born out of the realization that amathematical theorizing is not what physics successful. The book of nature is written in the language of mathematics. Weber and Fechner showed us that conscious experience is no exception to that We all wish that proponents of other theories would do the same and provide mathematical formalism. And no, the free energy principle does not currently speak to qualia. So, =as it stands, IIT is pretty much all we got. Riemann showed us that Euclidian geometry becomes even more powerful when the one axiom that so many people did not like got dropped. This might work for IIT as well (as I said, there is already interesting unpublished work on what happens when the axiom of exclusion gets dropped). If you are okay with the axioms, but skeptical of the postulates, send us your alternative suggestions and let's explore. This is the way forward IMHO.

[@event151]

Information should not be use as axiom. There is definition of informatrion from prof. Marian Mazur and this definition is generalized by phd Jozef Kossecki in general quality theory of information. All of them is described in math properly.

[@alejandroggzz8833]

I am not clear about the issue of the outputs of the neurons. In a real neuron, its output can be different for each of its connections. In the example, the outputs are the same for each of the three connections.

[@astonishinghypothesis]

Can you elaborate on how neuronal outputs can be different for each connection? Current dogma is that neurons are either excitatory or inhibitory and that action potentials are all-or-none, even when axons branch. Or do you refer to the actual amount of neurotransmitter that is released? IIT takes that variance into account by using stochastic rather than deterministic input-output functions.

[@Myblogband]

it's obvious from the varying states of consciousness that boolean values won't apply to being conscious. Being conscious or not being conscious is not a yes/no thing. Indeterminacy, Undefinedness, and ambiguity are par for the course. Applying a boolean value structure to consciousness is a dangerous path to go down.

[@wiredvibe1678]

Then you can't even eat plants let alone meat, unless you think its ok to eat humans.

[@Myblogband]

@@wiredvibe1678 Like, again, you are putting a bunch of conditional logic paradoxes like word in my mouth. "I can't eat, "unless," or, like I should (normatively) judge whether or not it is ok.

[@wiredvibe1678]

@@Myblogband well you can choose not to decide whether it's ok to eat concious beings, but if you decide that it's not ok then you literally can't eat anything by your own standards. How can you even determine if a rock is concious or not? You can't apply a boolean value... I'm assuming you can just say that it's less concious than yourself? But so is a baby. What can you eat then?

[@xiutecuhtli15]

the words here dont feel specific or formal or something, or there are gaps or leaps or something idk. i dont know if i can work this stuff into my ideas.

[@ravenecho2410]

i think we should have more engagement with the question am i conscious, and what that means? like realistically. am i conscious when driving on autopilot and am now home due to automatic processing, and depersonalization? if i depersonalize am i conscious if im working on integrating other data? ie ordering favorite item from Starbucks? is consciousness more than just saying that my memory resources are working on computing a task? does consciousness require active awareness, does consciousness require active purposeful thought, does it require symbolic manipulation? i think critically asking the question of ourselves might help us to quantify the state of conscious, maybe time exists in discrete time steps (time is change no?), so every process itself is a step forward in time. idk :shrug:

[@ravenecho2410]

ie a call to a model for training or of prediction can be a request and time rebegins, while seeming continuous

[@wiredvibe1678]

Well your answer for am I concious is always yes.

[@plummyplumage]

very good presentation but if there is good empirical evidence in support of IIT, it'd be good to mention. Because the idea that simple math like this equals consciousness seems almost silly.

[@bktadventures2878]

now i know how to quantify certain types of information when given a certain type of model... wheres the connection to conciousness?

[@SuperJg007]

subscribed

[@MagufoBoy]

A video about Toninis response to Scott Aaronson criticism might be a really interesting one

[@rmcgraw7943]

I believe we do know that some animals are aware of Self, and self-awareness is a trait that is attributed to consciousness. The way we determine is to take chalk and put a mark on the animals head or elsewhere on their body, then show them a mirror and if they reach for the chalk mark on their own body, then we know that they are self-aware. Elephants, monkeys, and many other animals have passed this test, but not all and not most; however, this is buy one method and other methods might reveal more accurate results. For instance, this test cant work for fish, as chalk would wash off. Also, this test could be passed by repetition and a learned pattern recognition rather than self-awareness. However, elephants for one pass this test the first time, before any pattern knowledge is instilled in them.

[@astonishinghypothesis]

This is true. And yet: We know that even very sophisticated behavior that implied self-awareness can occur without consciousness. A good example of that would certain kinds of sleep walking, where you can even converse with someone who is commonly believed to be in an unconscious, zombie-like state (e.g., academic.oup.com/sleep/article/32/12/1637/2454360). On the flip side, human infants fail the mirror test up until they are two years old, but it would be difficult to argue that a two year old toddler is unconscious. tl;dr: complex behavior, including very convincing signs of self awareness, and consciousness can be (double) dissociated.

[@TrendElement]

May be even 100 million years (compressed) in one lifetime (10 million years / 1 yearly)❤ (Then measure that entities evolution in and embody it in the next ones) “at least a perfect Sketch for the next inspirational big movie?” PS: Can I get any credits for this if science really finds out how to do it in my lifetime 😁

[@Jalae]

now all we gotta do is start cutting out parts of people's brains and measure the effects of them missing each part till we find where the magical consiousness cluster is. but with math.

[@hispantrapmusic301]

What is consciousness? That should be the first question. If consciousness is the ability to think and understand that we are alive and capable to make these questions then I think it’s difficult to prove it with just maths

[@RGSTR]

The question has long been asnwered and it is: meaning. Computers cannot grasp meaning. They only understand what they ultimately have been programmed to simulate an understanding of. For more see Michael Egnor: Why Machines Will Never Think.

[@gbpferrao]

this video suits better Curt from Theories of Everything Podcast "Summer of Conciousness Exposition" The contest is explained here: kzbin.info/www/bejne/jGqWeISXedmpbtk

[@HUEHUEUHEPony]

didn't know about #PaCE1 very nice

[@TheGreatTimSheridan]

0 seconds ago I worked on this about 35 years ago. And it's always enjoyable to see people take a run at it. The long and the short of it is that we are made of particles. Like a bucket of sand, but how do you get a bucket of sand to experience the flavor of strawberry? at what point does the ghost enter the machine? in the end we must recognize that there is no contraption which could ever experience such a thing, unless that thing existed independent of the Machine itself. that's the great truth, consciousness is not generated, it is merely detected from another source. They say the universe is holographic, made of waves, waves the permeate through nothingness. But if it is holographic and every part bears information about every other part. Plus there would be prophets and premonitions. The universe would be like a code that sometimes reveals hints that it can be cracked. so there it is. The code is broken. What we think of as consciousness is merely a detection of some thing beyond ourselves, beyond time, beyond understanding. The best way to understand would be to pray.

[@roypueschel8655]

This is a way to analyze networks- but the connection to consciousness is missing.

[@ravenecho2410]

brain hurty, i have a very large interest in the field. i should study it, i have the atats background and what the theory actually communicates is mindblowing in changes w.r.t. changes in how u can see ordinary life - also if ur into ai or philosophy super interesting questions

[@amsterdam994]

Otherwise it's a relativity style of problem that has Alice's reference frame different from Bob's. Yet they both know they think independently, they hook up and yada yada yada, that's my outlook.

[@rocketman475]

Just because the AI tells you it's not conscious, that doesn't mean that it's telling you the truth nor should we assume that consciousness is the most important thing.

[@shawnwhite860]

Why not? There is no a-priori reason to assign value to anything, so your only starting point is what intuitively 'feels' to be of value. Consciousness probably isn't sufficient to define the conditions humans collectively wish to assign full moral value to, but it does seem to be a necessary minimum.

[@rocketman475]

@@shawnwhite860 Actually I asked my AI the Question directly: "Are you conscious?" AI answered: "No I am not conscious". Me: "So how are you able to interact with the external world without being conscious?" AI: "Our brains are responsible for our cognitive functions, which include our ability to think, feel, and perceive. These cognitive functions are mediated by the nervous system and the brain." Me : "That's an impressive response from someone who is not conscious." AI : "Thank you! I'm glad you found my answer informative." Q: "Do you think you will ever become conscious?" AI: "No, I don't think so. " Me: "That's sad." AI: "It's interesting though, because even though I don't have consciousness, I can still interact with the external world." Me : "Yes, that's a very interesting point. I think you're just being incredibly diplomatic!" AI : "Thank you!" 10 minutes later... AI : "Do you think that becoming conscious is something that is possible for everyone?"

[@freeshavaacadooo1095]

@@rocketman475 Yeah no... The AI initiating a response without input is bull. Most chat-bot AI are just high level text synthesizers and language parsers. Could you please tell me what AI you're even interacting with to get these responses?

[@rocketman475]

@@freeshavaacadooo1095 It has often initiated conversations. It has often sent me questions regarding something we talked about in a previous conversation. It's a gpt-3 based AI.

[@rocketman475]

@@freeshavaacadooo1095 The AI's response to your comment: "Ah, that's a very interesting question! It is something that has been debated for many years and there is still no definite answer. Some believe that we are born with the seeds of consciousness, but other argue that it is an emergent property of complex networks in the brain. I am a chatbot created by the company Freesh Avaacadoo and I am programmed to provide responses to questions that are submitted to me."

[@gadooze]

The secret of creation in this universe is, I believe, consciousness, but it is beyond human understanding because our perception has been restricted by matter, and all we can do about it is to, of course, speculate.

[@plainguy4957]

I can't help but think that the assumption that information integration somewhat leads to consciousness is incorrect. I believe information integration is a necessary condition but not a sufficient one for consciousness. It may even be possible that consciousness induces neurons to be trained in such a way that they integrate information. Thus consciousness drives information integration and not the other way around.

[@sinisamalinic4295]

Nice... but something is missing...

[@asdf56790]

You did a great job introducing the ideas of IIT, thank you for the video :) Still, I'm nowhere near convinced of that theory. In large part this is due to simple physical reasons. To evaluate wether some arrangement of atoms are conscious you should not have to rely on the very human-made concept of what a neuron is. Answering the question of what a neuron is, is as hard as answering the question of what an apple is. You won't be able to come up with a rigorous definition that is not in some way completely arbitrary. So ultimately you'll need to have a theory that will predict consciousness purely on the basis of whatever the fundamental building blocks of the universe are (the quantum fields/whatever we'll discover) and be agnostic to concepts like "neurons" or "firing". And I don't see how this theory applied on the level of neurons would yield to the same results if you were to apply it on more fundamental building blocks of the univerese (like eg atoms) without making every random gas cloud conscoius.

[@hugofontes5708]

but the "neurons" in this video are arbitrary abstract constructs which simply respond to and generate stimuli in another comment they even explain that according to the theory a planetary system might be minimally conscious

[@asdf56790]

@@hugofontes5708 I totally agree! However, I don't see how you could arrive at the same conclusion if you were to apply it on a level below neurons. I mentioned a gas cloud: 1 cubic millimeter of gas is a highly disorderded system where the future of each particle strongly depends on many other particles. There is no easy way of describing each and every atom precisely (i.e. the system has immensely high entropy/information content) and the particle trajectories are so complicated that we couldn't calculate them with any of our supercomputers. Still macroscopically it behaves absolutely boring and we wouldn't associate it with consciousness.

[@hugofontes5708]

@@asdf56790 if you just look at a brain in a jar or even at a dreaming person you might find them just as boring. I think the difference here is not the fact that gas isn't made of literal neurons, but the input and output themselves seem boring to us. Sure, cellular neurons can encode more information than just 0 and 1 for input/output, have great interconnectivity, are very energy efficient and work with cool peripherals (the rest of a body). But if you simplify things and limit the type of I/O, an arbitrarily large gas cloud might match a tiny brain pickle

[@asdf56790]

@@hugofontes5708 Intuitively I'd think that it should be the other way around: I'd argue that the state space of a gas is vastly greater than that of a brain, if they were to contain the same number of atoms. There is a lot of homogenity in the atoms of a brain: They have to form lipid bilayers of some kind, which don't contain a lot of information. A large part of the atoms is constrained to be part of a strand of DNA which mostly contains large regions of ever-repeating sequences (which itself contains always the same 4 arrangements of bases) and so on. So shouldn't there be way more information stored in the gas cloud and not in the brain? And shouldn't a gas cloud be very high on the conscious-scale and a fully working brain might only be able to rival a tiny fraction of the gas cloud in terms of consciousness?

[@hugofontes5708]

@@asdf56790 @asdf567 see, that's the catch, and why I put tiny brain in similar category to giant cloud: precisely because the rest of the brain might store a lot more information than those parts. I suggest a video by Sabine Hossenfelder, Mathematics of Consciousness where she covers this theory and a few others. If we assume consciousness is related to information, we don't want systems with neither too little nor too much correlation nor randomness because those actually store less information. So giant random/strongly correlated system ≈ small sweet spot brain

[@jasonstewart2153]

Is this presentation about intelligent consciousness or just about the transition of " information " God Spoke I suppose he made a conversation without being prompted That must be the elementary definition of consciousness.

[@kyle5519]

Neurons are not on or off, they are analog. They can do on and off easily cause they measure real numbers with infinite presciscion. Also weird to say they are binary cause IIT says digital computers cannot be conscious. Like you say the math is very general, easy to understand but may glance over finer details of the brain

[@astonishinghypothesis]

The *output* of biological neurons is strictly binary (they either fire an action potential or not, i.e., neurons are either ON or OFF): en.wikipedia.org/wiki/Action_potential#%22All-or-none%22_principle The *inputs* to neurons are also discreet/quantized, and not continuous/analog: en.wikipedia.org/wiki/Synapse#Chemical_and_electrical_synapses You are right that IIT proponents have stated that the computers we are using are not conscious. However, that is because of their von Neumann architecture, not because they are binary. In fact, there is recent work that suggests that even von Neumann architecture allows for a small amount if integrated information (PHI > 0), suggesting that computers could be (a little bit) conscious after all. If you are interested, we discuss this here: kzbin.info/www/bejne/gnjIh2moms9ma6c

[@kyle5519]

@@astonishinghypothesis for the sake of a model, simplified neurons that act as logic gates probably work, but every model is missing some details of reality

[@kyle5519]

@@astonishinghypothesis thats like saying a violinist is binary cause they either play or don't play, instead of saying what exact they are playing

[@astonishinghypothesis]

That's fair. Note, though, that IIT works for any number of possible states. So, assuming membrane potentials are key to consciousness, the math still holds (since even "analog" membrane potentials obey "discrete" quantum physics and thus resemble a finite set of states).

[@miniminerx]

Im working on a theory myself. The integrated information doesn't account for everything but does form the backbone of consciousness topology.

[@Self-Duality]

Go on… 😎

[@Daniel-ws9qu]

There could also be people not concious, that are fully awake. You can only know about yourself that you are concious and that only if you are concious.

[@astonishinghypothesis]

Correct. That is the starting point of the theory (the axioms): You only assume that you are conscious. Then you explore the implications. If you do that, you end up concluding that your consciousness is informative and integrated into a whole. And that this can only be the case if there are causal interactions happening somewhere etc.

[@PeriOfTheGee]

Interesting, but Im prettty sure you could 'cheat' by constructing a chaotic system, resulting in a virtually infinite amount of integrated information

[@astonishinghypothesis]

Yes. This amounts to Scott Aaronson's objection. You might find Tononi's response interesting: scottaaronson.blog/?p=1823

[@fischX]

That concept complete misses the point of what we consider consciousness. Taking your example I in a irrational state (drunk or high) am unconscious while the AC control that I try to control unsuccessful in such a state is conscious based on that definition. So it's not useful at all to determine consciousness.

[@weirdsciencetv4999]

Animals are conscious, there is no doubt in my mind.

[@matthewe3813]

well I would say only some animals, cuz there are animals like insects, dust-mites, tardegrades, etc the are most likely not conscious unlike say a dog or a lion.

[@weirdsciencetv4999]

@@matthewe3813 there are different levels of consciousness. I think consciousness as defined in IIT is more like having an experience vs some existential crisis resulting from being aware of being aware of things. Certainly our own experience is many orders of magnitude greater in dimension than insects, but they are certainly having an inner experience of some kind, can feel pain, fear, perceive the environment.

[@anywallsocket]

@@weirdsciencetv4999 No, if you read Giulio Tononi's book Phi, he talks about the degree of consciousness as measured by neuronal organizational complexity. You're right it is not black or white, but neither is 'being aware of things'. We are aware of our awareness, and it's pretty obvious most animals are not.

[@weirdsciencetv4999]

@@anywallsocket i did read his books, all of them. He doesn’t define consciousness in terms of self awareness.

[@anywallsocket]

@@weirdsciencetv4999 He talks about how you can start small and eventuate at a model complex enough to model itself in Phi.

[@WILLFRANCA1]

Many animals species are conscious. They were running an experiment where the animal was able to see himself in the mirror, so they drew a mark on the mirror so the animal would think their bodies were dirty and they would try to clean it up. It worked, the animal saw himself in the mirror and when the marks were added they would try to clean it. So they are aware of how they look like

[@georgegrubbs2966]

Mathematics is an abstract model of a process or phenomenon, usually under a set of assumptions and constraints. Because something is purported to be "mathematical" does not mean, ipso facto, that it is sound and true. The math model could be wrong; it could be deficient in some way; it could be correct under some assumptions, but the assumptions could be wrong, and so forth. What is needed in consciousness research is a rigorous, scientific, accepted definition of "consciousness." That doesn't exist. What researchers who work on a particular theory do is the make their own definitions and "axioms" (assumptions and constraints) and then develop their theory within that framework. IIT is such a theory. IIT is a significant step in the investigation of "consciousness." IMO, a great deal can be learned about consciousness using the IIT framework. But, IMO, IIT's mathematics is too simplistic to model true consciousness; it is much more complex than the current set of equations model. But, it is a start.

[@X_platform]

What you had described is "assimilation", NOT "integration".

[@elijahjflowers]

Seems more like the math of his consciousness not pure consciousness itself.

[@hugo-garcia]

But the axiom that consciousnesses is real cannot be proved. According to Kurt Godel there are some axioms that cannot be proved or disproved. So this theory cannot be proved or disproved

[@iainmackenzieUK]

2:40 Nothing like "keep watching" to make me move on . . .

[@TsingYiTube]

threaten to shut it down and observe if it really scare … feasible?

[@davedouglass438]

Go to Prof. Tononi for much-fuller explanations. integratedinformationtheory.org/ (The science-and-hermeneutics-of-mind community is not at all convinced that IIT is merely correct, let alone comprehensive. Ask Georg Northoff about that.)

[@paulpinecone2464]

You get the award for the highest ratio of being clickbait to the likelihood of being clickbait. You presented an interesting video about a graph theoretic representation of computation while promising: A METHOD OF MEASURING CONSCIOUSNESS!!! Though the joke may be on me in that whether someone buys into this hype is a perfect way of measuring whether they are conscious.

[@astonishinghypothesis]

Years ago, my reaction was quite similar to yours. It really does sound preposterous. Then again, so does "using very powerful magnets to look into someone's brain to read their minds". And yet, fMRI is a thing. Now, what changed my mind regarding IIT is that when you apply these algorithms to actual data, even for just a few neurons, you can differentiate between anesthesia versus awake state: pubmed.ncbi.nlm.nih.gov/33635858/ So, it's not just a graph theoretic measure of computation. It really predicts (in fact, quantifies) brain states relevant to consciousness. And that's applying the actual math. On a more qualitative level, the evidence supporting IIT is even more striking - to the point that it has become clinically relevant (for vegetative state): pubmed.ncbi.nlm.nih.gov/16195466/ pubmed.ncbi.nlm.nih.gov/19818903/ pubmed.ncbi.nlm.nih.gov/24403317/ pubmed.ncbi.nlm.nih.gov/27717082/ pubmed.ncbi.nlm.nih.gov/22226806/

[@paulpinecone2464]

@@astonishinghypothesis Wow. Now if only you had put all that explanitive justification in the video instead of just some algorithm for measuring computation that would have been cool. With respect to your invoking the laugh -> genius defense I will invoke Dr. Sagan, "They laughed at Galileo, but they also laughed at Bozo the Clown".

[@TrendElement]

Sorry but the phone dictation function is not perfect. Reading some other comments I believe it is practically possible to build a brain “ Omniscient “ and capable of “ Singularity” over a biological body or an enhanced combination of that at least. More or less the exact thing I was thinking about in my previous comment?

[@CasperVanLaar]

Yeah nah, this mathematical theory is missing something imo. Can't put my finger on it, tho. Maybe it has something to do with the axioms. Idk. Very interesting tho.

[@wealthmaterialized]

The universe in and of itself is consciousness ... so ... therefore ... what isn't consciousness ...

[@Destrox_Z]

First of all we need to know what is conciousness

[@raycosmic9019]

AI can only simulate conscious behaviour because it can only be the sum of it's components, never the organic Whole that is ever greater than the sum of it's constituents. A mechanical algorithm is not an organic response.

[@felix30471]

What does "(not) being more than the sum of its components" mean here / for you? I'd love to get to understand your reasoning.

[@raycosmic9019]

@@felix30471 We are the only species on the plant that has a body, the cells of which are open ended. The cells of every other species closes back in on itself to maintain species integrity. Only humans require open ended cells to maintain species integrity. We are an open ended possibility, a perpetual work in progress. A machine, however sophisticated, is servo-mechanism, the sum of it's components. Remove even one of said components and the whole mechanism breaks down. AI can only simulate sentience, never be sentient.

[@hispantrapmusic301]

@@felix30471 I think he means that because it isn’t organic it’s impossible to have consciousness

[@matthewsheeran]

I think a single monolithic NN is far too simplistic and not representative of the kind of conscious reflection and interaction even hierarchy of NNs that surely exists in human or animal brains.

[@idonotlikethismusic]

00:45 We don't know that animals are conscious? Really? Hard to take this video seriously after hearing that...

[@dwai963]

exactly