Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Language is not essential for the cognitive processes that underlie thought

scientificamerican.com

Post reply on HN
Print view View original

371–380 of 424 posts

Re: Language is not essential for the cognitive processes that underlie thought

#371

Post by ryandv »

psychoslave wrote: post #366
ryandv wrote: post #60

Earlier quoted context omitted.

They do, in the first section of the journal article itself: > Do any forms of thought—our knowledge of the world and ability to reason over these knowledge representations—require language (that is, representations and computations that sup-port our ability to generate and interpret meaningfully structured word sequences)? Emphasis on "word sequences," to the exclusion of, e.g. body language or sign language. They g…

I don't find these excerpts in the linked article. Are you consulting an other document than the one pointed here?

The paper itself: https://www.nature.com/articles/s41586-024-07522-w.epdf?shar...

Re: Language is not essential for the cognitive processes that underlie thought

#372

Post by mountainriver »

Animats wrote: post #48

This is an important result. The actual paper [1] says that functional MRI (which is measuring which parts of the brain are active by sensing blood flow) indicates that different brain hardware is used for non-language and language functions. This has been suspected for years, but now there's an experimental result. What this tells us for AI is that we need something else besides LLMs. It's not clear what that someth…

Transformers are just sequence predictors, it doesn’t need to be language, increasingly it’s not

Re: Language is not essential for the cognitive processes that underlie thought

#373

Post by stevebrown »

Language plays a role similar to that of paper and pen in solving certain math problems. As a tool, it aids deeper thinking. It serves two key functions: facilitating communication and enhancing thought processes. This is why "chain of thought" type of intermediate language prompts improve reasoning in OpenAI's o1 model.

Re: Language is not essential for the cognitive processes that underlie thought

#374

Post by mountainriver »

fhdsgbbcaA wrote: post #116
Animats wrote: post #48

This is an important result. The actual paper [1] says that functional MRI (which is measuring which parts of the brain are active by sensing blood flow) indicates that different brain hardware is used for non-language and language functions. This has been suspected for years, but now there's an experimental result. What this tells us for AI is that we need something else besides LLMs. It's not clear what that someth…

My first thought as well - “AGI via LLM” implies that our grey matter is merely a substrate for executing language tasks: just swap out bio-neurons for a few H100s and viola, super intelligence.

It’s AGI via transformer

Re: Language is not essential for the cognitive processes that underlie thought

#375

Post by codebolt »

Animats wrote: post #48

This is an important result. The actual paper [1] says that functional MRI (which is measuring which parts of the brain are active by sensing blood flow) indicates that different brain hardware is used for non-language and language functions. This has been suspected for years, but now there's an experimental result. What this tells us for AI is that we need something else besides LLMs. It's not clear what that someth…

> What this tells us for AI is that we need something else besides LLMs.

An easy conclusion to jump to but I believe we need to be more careful. Nothing in these findings proves conclusively that non-verbal reasoning mechanism equivalent to humans couldn't evolve in some part of a sufficiently large ANN trained on text and math. Even though verbal and non-verbal reasoning occurs in two distinct parts of the brain, it doesn't mean they're not related.

Re: Language is not essential for the cognitive processes that underlie thought

#376

Post by ninetyninenine »

haswell wrote: post #368
ninetyninenine wrote: post #245

Earlier quoted context omitted.

>What this tells me is there is clearly no “reasoning” happening whatsoever with either model, despite marketing claiming as such. Not true. You yourself have failed at reasoning here. The problem with your logic is that you failed to identify the instances where LLMs have succeeded with reasoning. So if LLMs both fail and succeed it just means that LLMs are capable of reasoning and capable of being utterly wrong. It…

Success doesn’t imply that “reasoning” was involved, and the definition of reasoning is extremely important. Apple’s recent research summarized here [0] is worth a read. In short, they argue that what LLMs are doing is more akin to advanced pattern recognition than reasoning in the way we typically understand reasoning. By way of analogy, memorizing mathematical facts and then correctly recalling these facts does not…

You’re still suffering from the biases of the parent poster. You are picking and choosing papers that illustrate failure instances when there are also an equal amount of papers that verify successful instances.

It’s bloody obvious that when I classify success I mean that the llm is delivering a correct and unique answer for a novel prompt that doesn’t exist in the original training set. No need to go over the same tired analogies that have been regurgitated over and over again that you believe LLMs are reusing memorized answers. It’s a stale point of view. The overall argument has progressed further then that and we now need more complicated analysis of what’s going on with LLMs

Sources: https://typeset.io/papers/llmsense-harnessing-llms-for-high-...

https://typeset.io/papers/call-me-when-necessary-llms-can-ef...

And these two are just from a random google search.

I can find dozens and dozens of papers illustrating failures and successes of LLMs which further nails my original point. LLMs both succeed and fail at reasoning.

The main problem right now is that we don’t really understand how LLMs work internally. Everyone who claims they know LLMs can’t reason are just making huge leaps of irrational conclusions because not only does their conclusion contradict actual evidence but they don’t even know how LLMs work because nobody knows.

We only know how LLMs work at a high level and we only understand these things via the analogy of a best fit curve in a series of data points. Below this abstraction we don’t understand what’s going on.

Re: Language is not essential for the cognitive processes that underlie thought

#377

Post by alserio »

branko_d wrote: post #370
alserio wrote: post #147

Earlier quoted context omitted.

The idea that programming languages and natural languages are processed with the same wetware should be testable with something like the tests described in this submission. I don't expect it to be true, but only expecting something is not science

Other than the word “language”, programming languages and natural languages really have very little in common. Anecdotally, when I write code, I don’t “talk in my head”. The structures that I have in my brain are in fact difficult to put into words, and I can only vaguely describe them as interconnected 3D shapes evolving over time, or even just “feelings” and “instincts” in some cases. The code that comes out of tha…

I do subjectively agree with this. I, too, don't "code by words". However, it's the first time someone has described their personal experience to me as interconnected 3d shapes. Really fascinating and really distant from my own experience. For the second part of your message, code comments are a possible place where you can store the process, via the medium of words, this time.

Re: Language is not essential for the cognitive processes that underlie thought

#378

Post by Dylan16807 »

weard_beard wrote: post #302
Dylan16807 wrote: post #225

Earlier quoted context omitted.

Okay, but does that help us reach any meaningful conclusions? For example, okay some AI system doesn't have the capabilities of an auditory cortex or somatosensory cortex. Is there a reason for me to think it needs that?

Name a creature on earth without one. Imagine trying to limit, control, or explain a being without familiar cognitive structures. Is there a reason to care about such unfamiliar modalities of cognition?

> Name a creature on earth without one.

Anything that doesn't have a spine, I'm pretty sure.

Also if we look at just auditory, tons of creatures are deaf and don't need that.

> Imagine trying to limit, control, or explain a being without familiar cognitive structures.

I don't see why any of that that affects whether it's intelligent.

Re: Language is not essential for the cognitive processes that underlie thought

#379

Post by NemoNobody »

senand wrote: post #243
CSMastermind wrote: post #149

Earlier quoted context omitted.

When you look at how humans play chess they employ several different cognitive strategies. Memorization, calculation, strategic thinking, heuristics, and learned experience. When the first chess engines came out they only employed one of these: calculation. It wasn't until relatively recently that we had computer programs that could perform all of them. But it turns out that if you scale that up with enough compute y…

This seems quite reasonable, but I recently heard a podcast ( https://www.preposterousuniverse.com/podcast/2024/06/24/280-... ) that LLMs are more likely to be very good at navigating what they have been trained on, but very poor at abstract reasoning and discovering new areas outside of their training. As a single human, you don't notice, as the training material is greater than everything we could ever learn. After…

What part of AI today leads you to believe that an AGI would be capable of self directed creativity? Today that is impossible - no AI is truly generating "new" stuff, no poetry is constructed creatively, no images are born from a feeling, inspiration is only part of AI generation is you consider it utilizing it's training data, which isn't actually creativity.

I'm not sure why everyone assumes an AGI would just automatically do creativity considering most people are not very creative, despite them quite literally being capable, most people can't create anything. Why wouldn't an AGI have the same issues with being "awake" that we do? Being capable of knowing stuff - as you pointed out, far more facts than a person ever could, I think an awake AGI may even have more "issues" with the human condition than us.

Also - say an AGI comes into existence that is awake, happy and capable of truly original creativity - why tf does it write us poetry? Why solve world hunger - it doesn't hunger. Why cure cancer - what can cancer do to to it?

AGI as currently envisioned is a mythos of fantasy and science fiction.

Re: Language is not essential for the cognitive processes that underlie thought

#380

Post by lanstin »

mannykannot wrote: post #311
adrianN wrote: post #264

Earlier quoted context omitted.

That philosophers talk about p-zombies seems like evidence to me that at least some of them don't believe that consciousness needs to have observable effects that can't be explained without consciousness. I don't say that I believe that too. I don't believe that there is anything particularly special about brains.

The p-zombie argument is the best-known of a group of conceivability arguments, which ultimately depend on the notion that if a proposition is conceivably true, then there is a metaphysically possible world in which it is true. Skeptics suppose that this is just a complicated way of equivocating over what 'conceivable' means, and even David Chalmers, the philosopher who has done the most to bring the p-zombie argumen…

Like what is magic - it turns out to be the ability to go from interior thoughts to stuff happening in the shared world - physics is just the mechanism of the particular magical system we have.
Post reply on HN