This is a pretty accurate and comprehensive quick sketch of what we know so far about cognition. So far so good. The treatment of IIT is likewise a quick sketch, but as to whether it's "good" or not, that's harder to evaluate at this point. One one hand, this is a field which has often produced simplistic accounts of cognition that did not pan out. That's okay, but it's sobering. On the other hand, the idea of building a richly dynamic model of the richly dynamic system under study is one that makes fundamental sense. And finally we have the means to implement it, at least to a preliminary degree that should serve as proof of concept. So, I think it's a promising line to pursue. What it will yield remains to be seen. Of course. That's how science works. I'd like to suggest that if it works, it should prove to be a general treatment for neural networks of all kinds, not just biological ones. Indeed, if it doesn't work extremely well for the relatively easy case of artificial neural networks, then there must be something wrong with the theory. It's conspicuous that this consideration wasn't raised during the talk, but it was a quick sketch after all.
I have read Francis Crick's book "The Astonishing Hypothesis". It's a good introductory short book about the brain and the neuroscience of consciousness. Also good are Christof Koch's "The Quest For Consciousness" and Jeff Hawkins' "On Intelligence".
How does the II theory deal with connected conscious systems? Take two people who can communicate but are looking at two different views. If you sever the communication link, each person’s awareness is decreased yet there seems to be no subjective reduction in either person’s level of consciousness. What does this imply in integrated information theory? Does it mean that IIT cannot explain the subjective nature of consciousness or maybe that systems of consciousness have a separate consciousness that is not subjectively perceived by the components of the system?
Two people interacting does not change the Phi value (and thus the level of consciousness) of either brain. Now, if you have billions of people interacting, similar to neurons in a brain, that could produce a new meta-system that is conscious according to IIT. None of the individual people would notice or experience a change in their own consciousness, but collectively they would create a new mind. See: en.wikipedia.org/wiki/China_brain
@@astonishinghypothesis thank you! I have no formal training in neuroscience or neural networks but I was an algorithm developer before recently retiring. I had a conjecture that this could be the case and now I have a paper I can read on the subject.
In all the theories of consciousness that I have been exposed to, there seems to be an underlying assumption that the brain is the seat of consciousness. This is not unreasonable. However, what precludes the brain from being an antenna, of sorts? Something that integrates some property or field or whatever with data from our senses along with memory? I totally get how this could be considered a long shot idea as we have seen no proof that there is a property of matter or some field that is consciousness but it does seem strange that the idea isn’t even mentioned.
The issue with that notion is conservation of energy. If something would interact with the brain, we should see changes in the brain that cannot be explained deterministically by other brain processes. And that is just not the case. It is not like neuroscientists have not looked. The only potential way around that involves quantum physics: Since all macroscopic neuronal processes we observe behave like a clockwork, there is no room for outside forces to change its course. EXCEPT for the fact that there is also randomness. This randomness is largely limited to microscopic (synaptic) processes due to quantum effects. And, well, it is random. Or so it seems. But what if it is not? Nobel Laureate John Eccles and philosopher Karl Popper (yes, that Karl Popper) once proposed that there could be "fields" or whatever that ever so slightly manipulate the probability distributions of these random effects - hence giving it some direction. This could lead to something else "steering" the brain, as you suggested. Needless to say this is very fringe. Most people never heard of that theory and the main book is out of print: en.wikipedia.org/wiki/How_the_Self_Controls_Its_Brain
From this very abbreviated presentation, it would seem that a large enough artificial neural network would be conscious according to IIT. Am I getting this right? And can something be said to be conscious without having a subjective experience? If so, it seems that this view of consciousness is shallow.
It depends on how the artificial neural network is implemented. According to this theory, only physically interacting objects can evoke consciousness, and a piece of software cannot. Conventional computers (using von Neumann architecture) only allow for very small values of Phi, if at all. But neuromorphic computers might be a different story. IIT equates subjective experience with consciousness, so according to this theory there is no such thing as consciousness without experience.
This talk is not helpful to viewers wanting to learn this area of neuroscience. The talk uses presentation tricks & techniques to dazzle viewers new to the topic. Ultimately it does a poor job of explaining IIT, which is a controversial theory. This is why viewers needed to watch this video multiple times, because _he doesn't explain core ideas_ and _he doesn't connect the dots_ (he glosses over important ideas with only a terse explanation, & he jumps across ideas without linking them).
One parlor-trick technique occurs @18:05 when he asks "Do you need your eyes to be able to see?" Just a few seconds before he talked about "information of [from] the eyes", so his question **seems** to be asking if we need _information from our eyes_ to be able to see. In fact he is not asking that, as we learn at the end of his line of questioning. He chose to use a very vague word: "see". He is exploiting the many meanings and vagueness of this word, to perform his intellectual parlor-trick. He first presents "seeing" as "sensation" (this specific meaning of the word) - that our sensory system obtains photonic information from the eyes and sends it to the primary visual cortex (V1). Then he switches the game (he changes the meaning of "see", using it as a synonym for "visualize"), and he says we can dream without information from our eyes. Using different meanings of "see" is a word-game trick, since it fools viewers into making assumption (that he caused). Most viewers know we don't use our eyes to visualize, but do use our eyes to get sensory input. However he tricked viewers to play this game, by using a vague word. A careful neuroscientist would not use a vague word like "see", to refer to multiple processes: sensation/perception (of sensory information from the eyes), visualization (which occurs during dreaming and involves only recall/processing of information), and imagination (which can occur during visual-creative dreaming). He was only asking this much simpler question: "do you need your eyes to be able to visualize?", which has a simple answer. However he intentionally used a more confusing way to ask the question -- throughout his talk he uses confusion (induced in the viewer) to dazzle us (by showing so many tiny tiny graphs, fMRI scans, equations, tables, etc. -- without explaining most of them). A good presenter puts few key ideas on slides, and references nearly everything they put on a slide.
@@whoknew4722 > Then he switches the game (he changes the meaning of "see", using it as a synonym for "visualize"), and he says we can dream without information from our eyes. When I visualize Brad Pitt, I definitely don't see him in the sense of sensing photonic information, however when I dream I do see in a way that is indistinguishable from the process of sensing photonic information, at least from what I recall from the dream state. Is my recollection wrong according to neuroscience?
You think you can understand IIT in a spoken 45 minute lecture? lol.. guess what the function of the talk is?! to attract people and collaborators! wow! its a pitch, really?! "Oh I get it now!".. seriously what planet are you on not to pick up the plainly obvious. go read the book over several days & think about it over a few months. youll find it compelling.
Fabulous talk. Watching it the second time. If you find it difficult to follow, slow the speed to 0.75.
Damn! Watching this for the third time!
This is a pretty accurate and comprehensive quick sketch of what we know so far about cognition. So far so good.
The treatment of IIT is likewise a quick sketch, but as to whether it's "good" or not, that's harder to evaluate at this point. One one hand, this is a field which has often produced simplistic accounts of cognition that did not pan out. That's okay, but it's sobering. On the other hand, the idea of building a richly dynamic model of the richly dynamic system under study is one that makes fundamental sense. And finally we have the means to implement it, at least to a preliminary degree that should serve as proof of concept.
So, I think it's a promising line to pursue. What it will yield remains to be seen. Of course. That's how science works.
I'd like to suggest that if it works, it should prove to be a general treatment for neural networks of all kinds, not just biological ones. Indeed, if it doesn't work extremely well for the relatively easy case of artificial neural networks, then there must be something wrong with the theory. It's conspicuous that this consideration wasn't raised during the talk, but it was a quick sketch after all.
stumbled upon this randomly but it's very cool, thanks.
The feeling of life itself?
I have read Francis Crick's book "The Astonishing Hypothesis". It's a good introductory short book about the brain and the neuroscience of consciousness. Also good are Christof Koch's "The Quest For Consciousness" and Jeff Hawkins' "On Intelligence".
Who is the presenter?
Prof. Maier (Vanderbilt University)
scholar.google.com/citations?user=QBoOiHMAAAAJ&hl=en
How does the II theory deal with connected conscious systems? Take two people who can communicate but are looking at two different views.
If you sever the communication link, each person’s awareness is decreased yet there seems to be no subjective reduction in either person’s level of consciousness.
What does this imply in integrated information theory? Does it mean that IIT cannot explain the subjective nature of consciousness or maybe that systems of consciousness have a separate consciousness that is not subjectively perceived by the components of the system?
Two people interacting does not change the Phi value (and thus the level of consciousness) of either brain.
Now, if you have billions of people interacting, similar to neurons in a brain, that could produce a new meta-system that is conscious according to IIT. None of the individual people would notice or experience a change in their own consciousness, but collectively they would create a new mind. See: en.wikipedia.org/wiki/China_brain
@@astonishinghypothesis thank you! I have no formal training in neuroscience or neural networks but I was an algorithm developer before recently retiring. I had a conjecture that this could be the case and now I have a paper I can read on the subject.
In all the theories of consciousness that I have been exposed to, there seems to be an underlying assumption that the brain is the seat of consciousness. This is not unreasonable. However, what precludes the brain from being an antenna, of sorts? Something that integrates some property or field or whatever with data from our senses along with memory? I totally get how this could be considered a long shot idea as we have seen no proof that there is a property of matter or some field that is consciousness but it does seem strange that the idea isn’t even mentioned.
The issue with that notion is conservation of energy. If something would interact with the brain, we should see changes in the brain that cannot be explained deterministically by other brain processes. And that is just not the case. It is not like neuroscientists have not looked.
The only potential way around that involves quantum physics: Since all macroscopic neuronal processes we observe behave like a clockwork, there is no room for outside forces to change its course. EXCEPT for the fact that there is also randomness. This randomness is largely limited to microscopic (synaptic) processes due to quantum effects. And, well, it is random. Or so it seems. But what if it is not? Nobel Laureate John Eccles and philosopher Karl Popper (yes, that Karl Popper) once proposed that there could be "fields" or whatever that ever so slightly manipulate the probability distributions of these random effects - hence giving it some direction. This could lead to something else "steering" the brain, as you suggested. Needless to say this is very fringe. Most people never heard of that theory and the main book is out of print: en.wikipedia.org/wiki/How_the_Self_Controls_Its_Brain
From this very abbreviated presentation, it would seem that a large enough artificial neural network would be conscious according to IIT. Am I getting this right? And can something be said to be conscious without having a subjective experience? If so, it seems that this view of consciousness is shallow.
It depends on how the artificial neural network is implemented. According to this theory, only physically interacting objects can evoke consciousness, and a piece of software cannot. Conventional computers (using von Neumann architecture) only allow for very small values of Phi, if at all. But neuromorphic computers might be a different story.
IIT equates subjective experience with consciousness, so according to this theory there is no such thing as consciousness without experience.
This talk is not helpful to viewers wanting to learn this area of neuroscience. The talk uses presentation tricks & techniques to dazzle viewers new to the topic. Ultimately it does a poor job of explaining IIT, which is a controversial theory. This is why viewers needed to watch this video multiple times, because _he doesn't explain core ideas_ and _he doesn't connect the dots_ (he glosses over important ideas with only a terse explanation, & he jumps across ideas without linking them).
One parlor-trick technique occurs @18:05 when he asks "Do you need your eyes to be able to see?" Just a few seconds before he talked about "information of [from] the eyes", so his question **seems** to be asking if we need _information from our eyes_ to be able to see. In fact he is not asking that, as we learn at the end of his line of questioning. He chose to use a very vague word: "see". He is exploiting the many meanings and vagueness of this word, to perform his intellectual parlor-trick. He first presents "seeing" as "sensation" (this specific meaning of the word) - that our sensory system obtains photonic information from the eyes and sends it to the primary visual cortex (V1). Then he switches the game (he changes the meaning of "see", using it as a synonym for "visualize"), and he says we can dream without information from our eyes. Using different meanings of "see" is a word-game trick, since it fools viewers into making assumption (that he caused). Most viewers know we don't use our eyes to visualize, but do use our eyes to get sensory input. However he tricked viewers to play this game, by using a vague word.
A careful neuroscientist would not use a vague word like "see", to refer to multiple processes: sensation/perception (of sensory information from the eyes), visualization (which occurs during dreaming and involves only recall/processing of information), and imagination (which can occur during visual-creative dreaming). He was only asking this much simpler question: "do you need your eyes to be able to visualize?", which has a simple answer. However he intentionally used a more confusing way to ask the question -- throughout his talk he uses confusion (induced in the viewer) to dazzle us (by showing so many tiny tiny graphs, fMRI scans, equations, tables, etc. -- without explaining most of them). A good presenter puts few key ideas on slides, and references nearly everything they put on a slide.
@@whoknew4722 > Then he switches the game (he changes the meaning of "see", using it as a synonym for "visualize"), and he says we can dream without information from our eyes.
When I visualize Brad Pitt, I definitely don't see him in the sense of sensing photonic information, however when I dream I do see in a way that is indistinguishable from the process of sensing photonic information, at least from what I recall from the dream state. Is my recollection wrong according to neuroscience?
@@whoknew4722 you seem not be following his points very well, it's pretty clear. This is a very basic talk to follow.
You think you can understand IIT in a spoken 45 minute lecture? lol.. guess what the function of the talk is?! to attract people and collaborators! wow! its a pitch, really?! "Oh I get it now!".. seriously what planet are you on not to pick up the plainly obvious. go read the book over several days & think about it over a few months. youll find it compelling.