Separating logic and language
news.mit.edu
news.mit.edu
I trialed for a month it as an off-label migraine preventative on recommendation from my neurologist.
Topamax can dampen neuronal activity in the language area of the brain for some people.
It was a terrifying experience, but it was also a valuable one as it changed the way I view intelligence.
About 10 days in, my writing and speech skills had devolved to that of a primary school aged child. I tried to type a text message and struggled to come up with simple words like “and”. My speech slowed down considerably.
The terrifying thing was that my internal world was still as complex and meaningful as it was before. All of the emotions I felt were real and legitimate. My cognition outside of language was intact, I could do math just fine and conceptualize and abstract problems.
In spite of this, I was unable to convey what I was feeling and thinking to the outside world. It felt like I was trapped inside my own body.
After recovering, my intuitive understanding of the link between language and intelligence was changed forever.
See the following for more info on Topamax/Topiramate language impairment: https://link.springer.com/article/10.1007/s10072-008-0906-5
This seems to corroborate my feeling that aptitude with language is a very poor proxy for intelligence, although it's often used as one. My own experience is that I'm less struggling to find the right words and sounds more clever at the times when my brain is actually empty and I'm "not there", and at the contrary the more "self-aware" I am and the more I struggle to articulate my thoughts, as if my brain (or ego?) was stepping in the way and slowing me down by trying to supervise what I'm saying too closely, making me sounds dumb.
The phenomena you’re describing sounds similar to how I experience ADHD in conversation.
To clarify "cognition outside of language" and "unable to convey"... did it purely affect external communication? Were you able to think "in language"? Do you tend to rely on thinking "in language" (whether inner monologue, describing problems like drafting prose, or otherwise)?
To get retrospectively nerdy, it felt like switching from a raw text based lookup system to a knowledge graph / vector search.
Instead of accessing memories / concepts at a raw text level, I was pulling up lots of metadata and making links to these fusiform concepts that it was invoking. I would then try to find the words to describe what the links were conveying.
I do find myself thinking in one language or the other at times, but none is exclusive, often they get mixed together (the "next-token prediction" in my mind allows picking words in other languages that may more accurately describe an idea) and most frequently I do not have any internal dialogue at all.
I am pretty sure this is true for everybody, that the internal dialogue is not something natural but rather something people are so used to that we identify with it. Meditation and flow-state inducing activities (physical excercise, playing music, etc) make me think that this is the default for people, as these allow to access that lower layer of mental activity where internal dialogue disappears, even if briefly.
Babbling toddlers can do logic. I sometimes can also do logic, even though I don’t have an inner voice and don’t formulate thoughts in words of any kind. As I’m relatively sure I’m not an ET alien, there are probably plenty of people like me.
Quantity is not necessarily a problem: it feels a lot like the same tradeoff as if you need to picture a lot of apples in an image of fixed resolution, you can make them smaller/less detailed just by, loosely, "allocating fewer pixels per apple", rather than needing the same detail for each that you'd need if you're trying to zoom in depth into one apple.
However, if I have a situation where I have to picture a lot of very detailed things of irreducible complexity (the details matter) that need to be placed in a specific configuration (the ordering matters), then it gets harder. Statistics was so intuitive for me that I would read the equations as visual images in my head because it was throwing out the individual details and dealing with population mechanics as individual objects of study. However, combinatorics was initially very hard for me to learn because it was dealing with very large numbers of hierarchies of orders as well as the specifics of their interrelations, so building intuitive understandings of why the theorems were correct was very difficult.
I have learned to lean on a very internally barebones representation of language/symbolic structures when needing to reason about certain complex topics that require it, but, even then, often when I'm conscious of it, it's almost like I'm visualizing the letters or forming the words as sound in my head. When I'm forming sentences to actually speak or write, it feels like I've got a concept that I've turned into a "rendering" that I'm trying to communicate via language substructures. To do so, I'll break it down into component parts and just start describing the substructures as if I'm perceiving it and describing it live in small parts. It feels like the learned skill required for speaking a language for me is learning how to decompose and reorder these objects into chunks that are expressible within the vocabulary and grammar of the target language and finding the correct words&patterns that represent and order these chunks most efficiently for effective communication. Almost feels like a SAT problem or a language diffusive process (and likely is, to some degree, related to those processes or similar under the hood), but, it actually qualitatively feels that way to me, consciously/experientially. Language feels almost bolted onto my experience rather than the primary mode of thought, even if I can think in it if needed.
I think the visual "instant counting" also goes up to that number, so in some way we are all "bottlenecked" and require accessing "the heap".
LLMs might be a step towards AGI (assuming that's possible) but the prevailing belief that LLMs and human brains (the latter being the most complex structures in the universe that we know of) are already functionally equivalent in every way that matters seems more like an article of faith to me than fact.
Also, regarding LLMs, 2 years ago another famous expert in the field was adamant that LLMs can't plan, can't do math and they can't work at long contexts. Obviously they can do all of that now. So, really, no one knows.
We use harnesses. Practically useful AI is a system of pieces already. It hasn't been pure LLM since ChatGPT 3.5 release.
The AI implication is fascinating though. We tend to build systems where the language model is simultaneously the language interface, memory, planner, reasoner and controller. Human cognition may be pointing in the opposite architectural direction: separate persistent state, reasoning/control mechanisms and language, then let language be one tool among several.
That distinction seems more interesting to me than simply asking whether LLMs “reason in language.”
Language as an internal conceptualization is still (obviously) essential to logical reasoning. The question of how language and logic relate can, therefore, concern either the internal language of concepts or the expressed language and its grammar, the latter of which is related to logic in that it reflects to some degree the internal language, but it is not identical with it.
can you give me example
Are they looking at blood flow in areas to map "language network" and other stuff? I remember a few recent papers that found that a) blood flow doesn't necessarily mean "active area" and b) two brands of fmri found completely different "areas" activating for the same patient + same puzzle.
fMRI is one of the best tools we have, but that doesn't mean it's good. fMRI shit should always be taken with spoon of salt. source: a bunch of grad school friends in neuroscience and psychology a few years ago.
"on two patients whose language networks were destroyed by stroke, plus fMRI on healthy volunteers solving the same puzzles silently. Both lines of evidence point the same way: the language network sits out during reasoning." that doesn't mean they didn't learn the meta-ability of how to solve such things without language. You can learn something with language, move it to a separate "database" in some abstracted form then pull from that "database" from a non-language component. In fact I would think the language-isn't-thinking people would favor an explanation where the thinking ability is stored independently of language. I guess my point is that no "language" component being lit up during a specific reasoning task doesn't exclude the possibility we learned that specific reasoning algorithm (for lack of a better word) via language + other things.
TL;DR maybe this specific "logic" ability is a learned-via-language+ component of the brain. Language not being active during that puzzle could just mean language+ generated that component and stored it in a more abstract form than language.
Because the system modifies itself it's going to be difficult to untangle cause and effect here.