Colton Casto

The cerebellum supports high-level language?? Now out in @NeuroCellPress, we systematically examined language-responsive areas of the cerebellum using precision fMRI and identified a *cerebellar satellite* of the neocortical language network! authors.elsevier.com/a/1mUU83BtfHC-… 1/n🧵👇

The human brain is strikingly modular, with distinct networks for language, formal reasoning, social reasoning, and physical reasoning. Is this a fundamental principle of how intelligent systems are built, or an accident of biological evolution? In our latest preprint, we find that a similar modular organization emerges in Large Language Models, another class of intelligent system. Brains and LLMs are shaped by entirely different kinds of optimization (biological evolution vs. gradient descent). That they arrive at the same modular design anyway suggests modularity may be a fundamental property of intelligent systems. 🌐 Web: pengrui-han.github.io/LLM_Modularity… 📄 Paper: pengrui-han.github.io/LLM_Modularity… 💻 Code & data: github.com/Pengrui-Han/LL… Using circuit analyses across 46 tasks spanning four cognitive domains, we find: 1️⃣ Tasks that draw on the same network in humans recruit overlapping units in LLMs, while tasks drawing on different networks recruit distinct units. 2️⃣ These units are causally linked to model behavior. Ablating the units critical for one domain impairs performance in that domain (−26% accuracy) but barely touches the others (−2.5%). This project has been in the works for a while :) Huge thanks to my advisors @jacobandreas @ev_fedorenko @devarda_a, and to @Nancy_Kanwisher for valuable conceptual input and feedback throughout. #MIT

It is tricky to characterize the features represented by human language cortex. This work is a step toward doing so. Using small, interpretable feature sets, we explain language-network responses and show a shared feature basis across regions with variation across individuals.

🚨 New Preprint! 🧠 We gave an AI model one simple rule: rearrange your neurons so that nearby ones respond alike. We never told it what a face, a voice, or a sentence was. It grew brain-like maps for all three anyway. 🧵👇 🌐 Website: topo-omni.epfl.ch

New preprint: "Monosynaptic connections link functionally similar regions in human cortex." We use electrical stimulation + fMRI in epilepsy patients to map whole-brain monosynaptic connectivity at 42 cortical sites. doi.org/10.64898/2026.… 1/n

We don't start out with a bilateral language system: it's already strongly left-lateralized in young kids! Resilience of language to early LH damage must occur in spite of this early hemispheric bias. Congrats to Ola @olaozpa and Amanda @Amanda_M_OBrien! Out in NatComms now! 🎉

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)

New peer-reviewed paper w/ @m_heilb , @jkbszwczk & Floris de Lange! Pre-onset brain encoding has been taken as evidence that brains–like LLMs–predict upcoming words. We show that the same signatures arise in systems that cannot predict. (elifesciences.org) (1/8)

🎉 Excited to share our new paper in Nature: “Active Dissociation of Intracortical Spiking and High Gamma Activity.” 🧠 Huge thanks to my advisors first: @SlutzkyLab @joshuaiglaser Paper link: nature.com/articles/s4158… Here are some digests that walk you through the results 👇🏼

Our new paper on brain networks engaged during imagining is out now in Neuron! Here is a download link (free for 50 days): authors.elsevier.com/c/1msNE3BtfHGo… Congratulations to Nate Anderson for leading this work @rementurus.bsky.social 🧵

🚨New preprint! LLM teams are being deployed at scale, yet we lack the tools to predict when they’ll succeed, fail, or how to design them. Distributed computing faced the exact same questions and figured out how to answer them. We show those insights apply directly to LLMs 🧵👇

Excited to announce a new paper from our lab, by Ian Griffith @ianmgriffith with help from Preston Hess @phess2, introducing a model of attentional selection. nature.com/articles/s4156… @mitbcs @ScienceMIT @mcgovernmit @SHBTHarvard Here is a summary. (1/n)

Really great work demonstrating the importance of task-based responses in characterizing the functional organization of the brain. Congrats to @carobellum and the rest of the team!

I am totally pumped about this new work. "Task-trained RNNs" are a powerful and influential framework in neuroscience, but have lacked a firm theoretical footing. This work provides one, and makes direct contact with the classical theory of random RNNs. biorxiv.org/content/10.648…

Excited to share new work on how the brain makes social inferences from visual input! 🧠👯‍♂️ (With Leyla Isik, Shari Liu, @tianminshu, & @minjaejk!) biorxiv.org/content/10.648…

Colton Casto, a PhD student working in the lab of Professor Evelina Fedorenko, was able to identify four cerebellar areas that consistently got involved during language use news.mit.edu/2026/satellite…

Are neural nets across modalities really converging to the same representation as they scale, as the Platonic Representation Hypothesis suggests? We show that common representational similarity metrics are confounded by network width & depth. We propose a permutation-based null calibration that fixes this. Result❓ • Global convergence largely disappears. • Local neighborhoods persist. We propose the alternative Aristotelian Representation Hypothesis: Neural networks, trained with different objectives on different data and modalities, are converging to shared local neighborhood relationships Very proud of @FabianGroger and @ShuoWen18 for this work! Paper: arxiv.org/abs/2602.14486 Webpage: brbiclab.epfl.ch/aristotelian Code: github.com/mlbio-epfl/ari…

So great to chat with @_TheTransmitter about our findings :)

How do diverse context structures reshape representations in LLMs? In our new work, we explore this via representational straightening. We found LLMs are like a Swiss Army knife: they select different computational mechanisms reflected in different representational structures. 1/

Thanks to @KempnerInst for highlighting our new study! I'm excited to see where this research will take us in the years to come :)

Thanks for covering our work @ScienceNews!