Finding Language in the Brain
thereader.mitpress.mit.edu
Finding Language in the Brain
1–10 of 20 posts
Re: Finding Language in the Brain
#2For anyone interested in actual neurocomputational theories I can highly recommend the work of Friedemann Pulvermueller.
https://www.sciencedirect.com/science/article/pii/S030100821...
Re: Finding Language in the Brain
#3I have an idea that's admittedly a bit left of field but not-quite-crackpot.
Purpose and causation (and reasoning therein) is structured like a tree. You can always take a statement and ask for pointers up and down the tree by asking "how" and "why". Ignoring the specific semantics, everything can be modeled as "things" and force in support/opposition from those things to other things. Thus you arrive at a model of whatever semantics which is just networks of lines of support and attack.
So here's my strange idea: what if language maps literally to connections of neurons. The parse tree of a sentence literally encodes the connections of the electrical circuit which models things applying force in support or opposition to other things.
This is probably wrong, but its not obvious why it is wrong, and the process of showing that its wrong probably has some insights into a model that isn't wrong. So I throw my wild conjecture to the wind anyway.
Re: Finding Language in the Brain
#4I have an idea that's admittedly a bit left of field but not-quite-crackpot. Purpose and causation (and reasoning therein) is structured like a tree. You can always take a statement and ask for pointers up and down the tree by asking "how" and "why". Ignoring the specific semantics, everything can be modeled as "things" and force in support/opposition from those things to other things. Thus you arrive at a model of w…
Not too far off base from my understanding. But the first few layers are like firmware they provide the neural interface between hardware (reflexes, processing sensations, etc) and the rest of thinking. So it’s a system that it’s very deeply tied to how we perceive and interact with our bodies and through them the world.
And in fact the first level or two of the hierarchy happen outside the brain in the nervous system or spine (unless hearing or sight related obvs)
There’s one piece missing though and that’s goals/control variable. My thoughts about this is how we encode things and act of things. When we recognize a tree there’s a set of nodes that activate then some higher layer recognizes trust as a tree and a higher layer still decides how to use that information or how to interact with those nodes (it sets a desired state. Ie. Walk to the tree, write the word, recognize the picture, etc).
Take a look into Hierarchical Perceptual Control Theory.
Re: Finding Language in the Brain
#5I have an idea that's admittedly a bit left of field but not-quite-crackpot. Purpose and causation (and reasoning therein) is structured like a tree. You can always take a statement and ask for pointers up and down the tree by asking "how" and "why". Ignoring the specific semantics, everything can be modeled as "things" and force in support/opposition from those things to other things. Thus you arrive at a model of w…
Not too far off base from my understanding. But the first few layers are like firmware they provide the neural interface between hardware (reflexes, processing sensations, etc) and the rest of thinking. So it’s a system that it’s very deeply tied to how we perceive and interact with our bodies and through them the world. And in fact the first level or two of the hierarchy happen outside the brain in the nervous syste…
One interesting implication of this is that language is not something separate in the brain.
It is a higher level abstraction over the existing processes and reuses all of the existing machinery. So our conception of the word tree would translate to other animals brains directly, they just lack the higher level abstractions of language, consciousness, and understanding that others think and communicate.
So the additional nodes collected under the “tree” node (or nodes) would be cleared and repurposed because it’s useless information to the other animal.
Re: Finding Language in the Brain
#6I have an idea that's admittedly a bit left of field but not-quite-crackpot. Purpose and causation (and reasoning therein) is structured like a tree. You can always take a statement and ask for pointers up and down the tree by asking "how" and "why". Ignoring the specific semantics, everything can be modeled as "things" and force in support/opposition from those things to other things. Thus you arrive at a model of w…
It would be interesting to try to map this tree with a language model like GPT.
Re: Finding Language in the Brain
#7I have an idea that's admittedly a bit left of field but not-quite-crackpot. Purpose and causation (and reasoning therein) is structured like a tree. You can always take a statement and ask for pointers up and down the tree by asking "how" and "why". Ignoring the specific semantics, everything can be modeled as "things" and force in support/opposition from those things to other things. Thus you arrive at a model of w…
My suggestion, try the book "realism and reason." It's a collection of papers by Hilary Putnam. It gets into this topic in several different ways, surveys some of the other thinking on it, and so on.
Re: Finding Language in the Brain
#8Protip/shameless plug: the MIT Press website now includes a reading enhancement feature called BeeLine Reader, available under the 'Aa' icon. Disclosure: I'm the founder.
Re: Finding Language in the Brain
#9Protip/shameless plug: the MIT Press website now includes a reading enhancement feature called BeeLine Reader, available under the 'Aa' icon. Disclosure: I'm the founder.
Do the changes in color actually reflect the importance of the content? Because if not, I find myself focusing on the wrong things.
Giving that you are a founder, how much money can be made with such a thing?
Re: Finding Language in the Brain
#10I have an idea that's admittedly a bit left of field but not-quite-crackpot. Purpose and causation (and reasoning therein) is structured like a tree. You can always take a statement and ask for pointers up and down the tree by asking "how" and "why". Ignoring the specific semantics, everything can be modeled as "things" and force in support/opposition from those things to other things. Thus you arrive at a model of w…
There was such a theory, developed in the 1960s and 1970s, I think, although I can remember neither the name of the linguist (I don't think it caught on broadly, so it really was just one researcher) nor the name of the theory.
I think it foundered for a number of reasons, one of which was that it had--so far as I know--no account of the creativity of humans "doing" language. That is, we can understand and speak any number of sentences we've never heard before, and it's hard (or impossible) to imagine that all those novel sentences were ever in the brain, or that new nerve connections grew so rapidly.
For example, I can tell you that I've invented a fugelizer, and that it is useful for turning bryles into fugels. Now I've just used three words that you've never heard before, but you can sort of make sense of what I've said. But clearly there's no 'fugelizer' neuron in your brain, etc.