this is mesmerizing

Check out commentary on the @futrell.bsky.social and @kmahowald.bsky.social BBS Target Article where Jeff Mitchell and myself review the evidence on LLMs learning impossible languages: arxiv.org/abs/2511.11389

Good question! Here are some places to start: jonathancook.substack.com chrishedges.substack.com substack.com/@caitlinjohn... substack.com/@arnaudbertr... substack.com/@mate

Ah yes, such good “theories” of language 🙃🙃🙃

Unfortunate update: Link reached out to BlueSky and got email back now on reason. His appeal is denied. He will remain permanently banned on BlueSky’s side. Here are the screenshots he just sent me as of 8:22 AM EST(my time)

Yes, the Home Secretary really did just say we need to hold off protesting against the British government's continuing complicity in Israel's daily slaughter in Gaza out of respect for Jews grieving over the Manchester synagogue attack.

Check out following article where I challenge some of the conclusions of McCoy and Griffiths (2025) regarding language learning in language models (LMs) doi.org/10.31234/osf...

Q: Why did the LLM cross the road? A: We're not sure, but it achieved 94.7% on CHIKENBench-Large

Critical part of the President's new $100,000 charge for H1-B visas: The Administration can also offer a $100,000 discount to any person, company, or industry that it wants. Replacing rules with arbitrary discretion. Want visas? You know who to call and who to flatter.

WhatsApp wouldn’t let me write the word genocide in a text this week. It kept changing it to ‘Genie I’ (?) I slowed right down and realised the word isn’t in its dictionary. I found that weird.

You assert the data support distributed processing. I show through simulations that the data are consistent with grandmother cells (and that we should not rule out localist models based on current data). I guess we can leave it there.

The following article is now in press at Psychological Review. Interested to hear what people think! "The successes and failures of Artificial Neural Networks (ANNs) highlight the importance of innate linguistic priors for human language acquisition". osf.io/preprints/ps... via

What is it about human brains that allows us to reason symbolically whereas most other animals cannot? And why does AI sometimes struggle to generate images of objects in statistically unusual spatial relationships to each other, when people wouldn’t struggle to imagine such relationships?

this is so wonderfully weird, give her all the golds 🥇

It seems like a lot of systematic-reviews/meta-analyses are analogous to money laundering. You run a load of studies that nobody actually reads, many of them "dirty", through some process. On the other end you get "clean" results that the ignorant or deceitful can easily communicate as "evidence".

Here is a recent PsyArxiv paper where we show that Centaur does not handily outperform domain specific cognitive models because Centaur is not tested in the right way. Its predictions are driven by processes unrelated to human performance. doi.org/10.31234/osf...

This highlights the problems with distributed representations (DR) learned by SGD, and an alternative, that may support the things that DR cannot, such as OOD generalisation, and selective deficits. Stanley is the most important thinker in the field www.youtube.com/watch?v=KKUK...

If you can watch this and then think ‘I won’t share because that might be putting myself on the line’ then you know what you’d have done if you were in Germany when your Jewish neighbours were being loaded into to cattle trucks. Nothing. Please share - it’s our ‘I’m Spartacus moments

The Centauromachy begins

Binz et al. (in press, Nature) developed an LLM called Centaur that better predicts human responses in 159 of 160 behavioural experiments compared to existing cognitive models. See: arxiv.org/abs/2410.20268

Ever wondered what 123 self-citations look like? Scroll no further: Hasan et al. 2021 (DOI: 10.1016/j.chemosphere.2020.128668) in 'Chemosphere' show us how it's done! pubpeer.com/publications... The reference section looks stunning! Senior author Md Rabiul Awual should be proud!

Everyone involved in trying to pass off that ChatGPT-extruded disinformation manifesto as a scientific report should be, at the very least, barred from public office for life. The MAHA report is dangerous criminal misconduct and an offence to the process of science.

What an amazing claim by a very credible source. It is strange I cant seem to find more on this though. youtu.be/QZX1mRrqpPM

And now Labour are actually getting outflanked by Reform ON THE LEFT. (Reform saying they'll scrap child benefit cap and reinstate winter fuel allowance.) Time for Labour to get rid of Starmer and dear god don't replace him with Wes Streeting (of course they will, sigh).

Check out our paper on the limitations of vision reasoning in a range of DNNs. Paper entitled: "Visual Reasoning in Object-Centric Deep Neural Networks: A Comparative Cognition Approach" now out in Neural Networks. authors.elsevier.com/c/1l8P93BBjKvX…

It is worth once in a while remembering what real journalism looks like. www.youtube.com/watch?v=Xqlv...

we have one in the family!

Great that there is so much push back on trump here. But striking to me how little comment about Gaza. Here Israel caught out in a lie where IDF just killed and buried 15 people from Red Cross that happened to be caught on video. Time to push back here too? www.nbcnews.com/news/amp/rcn...

Did he mention anything about genocide? Universal health care? Wealth tax? Reducing military spending? Green new deal?

There are perhaps two dozen larger cities in china. And something like 1.5 billion people in a county the size of USA. But the outbreak was near a Covid research centre. It should at least make you interested enough to watch the video featuring the lead of the lancet Covid inquiry.

still feels like interacting with an idiot to me

Get somebody in this room who knows that the word “transgenic” does not mean “transgendered”

Laughed out loud at this one, You may enjoy it too www.youtube.com/watch?v=1BfJ...

The weird stimuli from psychology are what is largely missing in neuroAI. This exchange seems at odds with your original thread that was highlighting the limitations of research with naturalist images and importance of artificial stimuli. With that I agree and I’ll leave it at that.

But by manipulating models you don’t know why one model predicts better than another. So don’t know whether the better prediction reflects better alignment. I don’t understand why the field is so resistant to manipulating images as standard in psychology. Should be the primary approach in my view

I'm doing all the classics. You can also check out my remaster of McCulloch and Pitts with LIFs and evolutionary algorithms (with @jeffreybowers.bsky.social): neural-reckoning.org/pub_adapting...

Would be great to hear your response to this: bsky.app/profile/jeff...

The ability of ANN to learn languages when trained on a diet of human data is often claimed to undermine the need for innate language priors for human language acquisition. In the following paper I challenge this conclusion. osf.io/preprints/ps...

I disagree. An experiment that tested a hypothesis revealed that many DNNs rely on texture to classify objects. And we have shown that predictivity on brain-score largely reflects the backgrounds rather than the objects themselves. Only experiments can reveal this: 2024.ccneuro.org/poster/?id=156

What about good ol' experiments that manipulate independent variables designed to test which features of inputs drive performance in DNNs and humans? There is a big psychological literature out there that has long ago highlighted problems with current models, and need to add constraints to DNNs.

We then all agree in one respect: There is a lot of hype.  But the problem is with researchers who take the hype seriously?That we should focus on the strengths rather than weaknesses of DNNs? That there is little "confusion that deep neural networks (DNNs) are ‘models of the human visual system’”?🤷‍♂️

This photo is of the inside of a 250 year old Stradivarius violin