Would it be nessesary to model systems memory consolidation as well? The classic view on why we have systems memory consolidation is that hippocampus is optimized for fast event-based learning, while cortical areas are optimized for storing stable semantic knowledge. So, information is first stored in the hippocampus and then with time cortex extracts stable semantic representations from it (especially in the absence of new sensory input, like during sleep). If evolution came up with this design, then perhaps to make information processing more efficient (like making semantic representations more stable), one would have to also model the hippocampus, as cortex stores information through it (as demonstrationed by inability to learn new things when the hippocampus is damaged)?
I believe they were referencing "Why Red Doesn't Sound Like a Bell: Understanding the feel of consciousness". Amazon Link: www.amazon.com/Doesnt-Sound-Like-Bell-Understanding/dp/0199775222
What if the "algorithm" changes at around 4 years of age, and some of the strange connections are left-overs from that birth-4 years period, which we see manifested as childhood amnesia?
@@snippletrap agreed. just clarifying the difference between consensus and voting...many scientists use the same simplification, but it gives lay people the wrong idea of the mechanisms...
Of all the lectures I think this guy is getting somewhere. If you want to fly you need to understand birds. If you want to create a mind study the brain. 😏
Would it be nessesary to model systems memory consolidation as well? The classic view on why we have systems memory consolidation is that hippocampus is optimized for fast event-based learning, while cortical areas are optimized for storing stable semantic knowledge. So, information is first stored in the hippocampus and then with time cortex extracts stable semantic representations from it (especially in the absence of new sensory input, like during sleep). If evolution came up with this design, then perhaps to make information processing more efficient (like making semantic representations more stable), one would have to also model the hippocampus, as cortex stores information through it (as demonstrationed by inability to learn new things when the hippocampus is damaged)?
I think modeling the hippocampus properly will be crucial to getting a working model of the neocortex.
Can the paper fleetingly mentioned by BIll ("What Ideal Brain Would Exist on the Other Side of the Universe?") be cited? It sounds very interesting.
Numenta should raise more money and build larger team.
BTW: what happens with Ray Kurzweil ? He went very quiet once he got employed by Google ..
He got lots of money and lost his drive
Maybe they paid him to shut up!
34:35
That’s very Piagetian!
What was the colour theory book referenced in the Q&A? I couldn't hear it properly.
I believe they were referencing "Why Red Doesn't Sound Like a Bell: Understanding the feel of consciousness". Amazon Link: www.amazon.com/Doesnt-Sound-Like-Bell-Understanding/dp/0199775222
Thank you
It was quite funny when Kevin O'Regan, the author of the book was sitting there.
What if the "algorithm" changes at around 4 years of age, and some of the strange connections are left-overs from that birth-4 years period, which we see manifested as childhood amnesia?
I agree.
Awesome thx !
Cortical columns do not vote, they reach consensus.
They reach consensus by aggregating their predictions, which is what Hawkins means by "voting".
@@snippletrap agreed. just clarifying the difference between consensus and voting...many scientists use the same simplification, but it gives lay people the wrong idea of the mechanisms...
37:10 Sounds like Buddha : ]
Of all the lectures I think this guy is getting somewhere. If you want to fly you need to understand birds. If you want to create a mind study the brain. 😏
Скучно без перевода😂
*If you are forgetting about the heart, it does not make any sense.*