I'm very happy to share this work that came out of my visit to the GLySN lab! Let's use artificial speech systems to learn new things about the human brain 🧠
How is uncertainty in LLMs output reflected in internal representations?
In our new work (to appear at ICML 2026), we show that the shape of internal token trajectories provides a direct geometric link to behavioral uncertainty (output entropy). 🧵(1/n)
How would you categorise these images? What features would you use? Preprint just went live looking at how we make sense of things we’ve never seen before (i.e., drawings just like these!). If that sounds interesting, read on or check out the preprint: www.biorxiv.org/content/10.6...
1/
www.nrc.nl/nieuws/2026/...
On May 8, Stichting Open Spraaktechnologie organizes a workshop on evaluation and interpretability of spoken language systems. Featuring @wzuidema.bsky.social, @hmohebbi.bsky.social, @gaofeishen.com, @yuanyuanzhang.bsky.social, @mdhk.net & @cpouw.bsky.social
openspraaktechnologie.org/nieuws/works...
Richard @futrell.bsky.social and I have posted our response to the commentaries on our BBS target article "How Linguistics Learned to Stop Worrying and Love the Language Models." The response is: "You Can't Fight in Here! This is BBS!" arxiv.org/abs/2604.09501
Presenting this at #ICML with @rjantonello.bsky.social and Aditya Vaidya✨
Why do 𝙢𝙞𝙙𝙙𝙡𝙚 layers in LLMs and speech-audio models best predict brain responses to language?
We show a peak in the dimensionality of 🤖 activations (left) to track high 🧠 predictivity (right)
🧵(cross-posted from X)
Excited to share our new publication, “Measuring Naturalistic Speech Comprehension in Real Time”!
➡️ rdcu.be/fa3hk #psynomBRM
w/ @kriesjill.bsky.social, Shiven Gupta, Maria Papworth Burrel, & @lauragwilliams.bsky.social
🧵1/11
New paper out today! When baby zebra finches are tutored with manipulated songs with shuffled syllables, they transform them to be consistent with linguistic laws 🐥 Reanalysis of data from James and Sakata (2017) link.springer.com/article/10.1...
Norm, the formal theoretical linguist, and Claudette, the computational language scientist, have a lovely time discussing whether modern language models can inform important questions in the language ...
