At 38:38, they discuss a claim that LLMs are merely 'statistical parrots.' But when I consider how I'm forming this sentence, I'm not sure I'm doing anything more sophisticated than an LLM's autocomplete based on learned patterns. What key differences between human cognition and current AI capabilities might I be overlooking?
I left graduate school in behavioral neuroscience because I knew that the paradigm was wrong but they kept rejecting my ideas. I left and wrote a book instead of a dissertation in an institutionalized context. 😂 So I can relate to this guest Jeff Hawkin’s experience. 😊 very enlightening, I absolutely love the thousand brain idea, but it still says nothing about the binding problem. The binding problem is quite difficult to see, let alone conceptualize. I used the experience of meaning and the logistics of that to get a handle on it-such as in reading and analogy and metaphor. Integration of information is the best or closest theory so far, but it still has problems. I used the experience of synchronicity to get a handle on it. It’s much more related to physics than people imagine, even Penrose and Hameroff. I think the issue is that we do not have evolved senses for exactly ONE area of the local universe-brains. So when we open up a brain and look at it, our brains fail in one regard. We do not see an inherent property of brains that is there only symbolically to us. So we can look at how the universe does things when it thinks and organizes itself and solves its own problems. I write about that in my books and papers. (“Eye of God: Language of Universal Mind” is my latest book on the symbolic universe-focusing on the human eye as an inherently symbolic structure) Anyway, this was awesome, Thank you 🙏🏻 ❤ both very much ❤
this is one of the greatest conversations in a podcast-completely fascinating while being easily followable. could watch a series of these between you two… haha. left me wondering things, like: why do all the (non-differentiated) spikes gather in a general/specific area? is it soley/mostly based on the input mechanism (?) like you mentioned for the "problem of qualia". how does the roaming use of the cortex happen; i.e. no sight so now auditory model building-does it have something to do with its inability to "vote" so it gets taken over? what creates the nature of dreaming? is it created by the melding of known input(s) (handled normally) and lack of known input(s) so it creates a hallucination in absence of the known/regular/woke input? same for "day-dreaming"? what about lucid dreaming? (do remember your episode about dreaming; probably should revisit it). thanks for sharing!
These topics are always so interesting - especially from the viewpoint of biology on cellular levels, for different forms of "intelligence" and problem solving, for how cells have capacities to solve entirely novel problems in different spaces. What sort of problems are they trying to solve giving us our gestalt experience of self? Why those, and not others? Hacking together access to some different elements of the "extra-biological" electromagnetic spectrum, and other entirely more platonic forms-style abstract information has brought me to a similar theory on the structure of data forming representation in qualia, along with how attention and expectations play into it. (2d spatial information comprehension transference between different sensory abstract signal types was a fun surprise!) These brains are so hungry to form meaning around patterns and signal, thankfully! Reality is wild, and conversations like this video has are deeply fascinating - thank you!
Are there any other theories about the neocortex that are as interesting? Also what papers are their showing grid cells in the cortical column, I’d like to read up on it
I sometimes use the word hear to describe thoughts. I don’t actually hear them, nor do I think they come from anything or anyone other than me. Language is funny.
But we're still trying to describe the ultimate nature of the brain and its processes, aren't we? Yet, we speak figuratively, symbolically, allegorically, and draw analogies. We never really get to the brain's real nature. I'm curious if we are stuck in the same world as quantum physics. We may be able to predict brain activity, but its ontology is beyond our current knowledge. Richard Feynman counseled us to not worry about ultimate reality. "Just shut up and calculate!"
Simple language question: So are we able to replace some skill points of boring topics with subjects we love? Lets say, geography and language is stuff you don't wanna waste your time on, because you play chess.
Great content btw. I am following it all. Peace. PS. Unless the background music has a real important purpose, could you please consider making it optional (turning it off)? :)
I don't mean to be mean, but he doesn't really have a grasp on this like he thinks. It not sensory motor, it's sensory reward activation reinforcing the strongest concurrent activation patterns. He is describing a learned algorithm, rather than the algorithm that physically derives the algorithms you learn.
At 38:38, they discuss a claim that LLMs are merely 'statistical parrots.' But when I consider how I'm forming this sentence, I'm not sure I'm doing anything more sophisticated than an LLM's autocomplete based on learned patterns. What key differences between human cognition and current AI capabilities might I be overlooking?
I left graduate school in behavioral neuroscience because I knew that the paradigm was wrong but they kept rejecting my ideas. I left and wrote a book instead of a dissertation in an institutionalized context. 😂 So I can relate to this guest Jeff Hawkin’s experience. 😊 very enlightening, I absolutely love the thousand brain idea, but it still says nothing about the binding problem. The binding problem is quite difficult to see, let alone conceptualize. I used the experience of meaning and the logistics of that to get a handle on it-such as in reading and analogy and metaphor. Integration of information is the best or closest theory so far, but it still has problems. I used the experience of synchronicity to get a handle on it. It’s much more related to physics than people imagine, even Penrose and Hameroff. I think the issue is that we do not have evolved senses for exactly ONE area of the local universe-brains. So when we open up a brain and look at it, our brains fail in one regard. We do not see an inherent property of brains that is there only symbolically to us. So we can look at how the universe does things when it thinks and organizes itself and solves its own problems. I write about that in my books and papers. (“Eye of God: Language of Universal Mind” is my latest book on the symbolic universe-focusing on the human eye as an inherently symbolic structure) Anyway, this was awesome, Thank you 🙏🏻 ❤ both very much ❤
Awesome! Love the graphics too
Thanks for doing this series
hawkinz is amazing , i have read his books , his hypothesis a thousand brains is fabulous
this is one of the greatest conversations in a podcast-completely fascinating while being easily followable. could watch a series of these between you two… haha.
left me wondering things, like: why do all the (non-differentiated) spikes gather in a general/specific area? is it soley/mostly based on the input mechanism (?) like you mentioned for the "problem of qualia". how does the roaming use of the cortex happen; i.e. no sight so now auditory model building-does it have something to do with its inability to "vote" so it gets taken over? what creates the nature of dreaming? is it created by the melding of known input(s) (handled normally) and lack of known input(s) so it creates a hallucination in absence of the known/regular/woke input? same for "day-dreaming"? what about lucid dreaming? (do remember your episode about dreaming; probably should revisit it). thanks for sharing!
Absolutely fascinating! This is why I love this channel so much.
How is it only 50 minutes long?!
Please do one more interview with Jeff 🙏
So glad to find you here.
These topics are always so interesting - especially from the viewpoint of biology on cellular levels, for different forms of "intelligence" and problem solving, for how cells have capacities to solve entirely novel problems in different spaces. What sort of problems are they trying to solve giving us our gestalt experience of self? Why those, and not others?
Hacking together access to some different elements of the "extra-biological" electromagnetic spectrum, and other entirely more platonic forms-style abstract information has brought me to a similar theory on the structure of data forming representation in qualia, along with how attention and expectations play into it. (2d spatial information comprehension transference between different sensory abstract signal types was a fun surprise!)
These brains are so hungry to form meaning around patterns and signal, thankfully!
Reality is wild, and conversations like this video has are deeply fascinating - thank you!
Love Jeff!!
love this, but the background music in the interview is cheesy
Several of my cortical columns voted: The interview is great -- but no music needed.
Once I noticed it I couldn't ignore it anymore 😢
Are there any other theories about the neocortex that are as interesting? Also what papers are their showing grid cells in the cortical column, I’d like to read up on it
Where does thinking about ideas come in? How do we decide on a political candidate?
Helps explain why some people “hear” colors
I sometimes use the word hear to describe thoughts. I don’t actually hear them, nor do I think they come from anything or anyone other than me. Language is funny.
Interview starts at 9:00
But we're still trying to describe the ultimate nature of the brain and its processes, aren't we? Yet, we speak figuratively, symbolically, allegorically, and draw analogies. We never really get to the brain's real nature. I'm curious if we are stuck in the same world as quantum physics. We may be able to predict brain activity, but its ontology is beyond our current knowledge. Richard Feynman counseled us to not worry about ultimate reality. "Just shut up and calculate!"
Simple language question: So are we able to replace some skill points of boring topics with subjects we love? Lets say, geography and language is stuff you don't wanna waste your time on, because you play chess.
Great content btw. I am following it all. Peace. PS. Unless the background music has a real important purpose, could you please consider making it optional (turning it off)? :)
I don't mean to be mean, but he doesn't really have a grasp on this like he thinks. It not sensory motor, it's sensory reward activation reinforcing the strongest concurrent activation patterns. He is describing a learned algorithm, rather than the algorithm that physically derives the algorithms you learn.
MR. Eagleman can I send you my book about emotions?