Researchers gave LLMs "brain damage" and found they mimic human aphasia symptoms
coryshain · x · 2026-09-26
A NeurIPS-accepted study by nrolling, Jill Kries, Laura Gwilliams, and Cory Shain induced "brain damage"-style lesions in LLMs to study how they process language.
Key finding: damaged models produce symptoms resembling human aphasia, but with distinct distributions — offering a new lens on how model language processing parallels and diverges from human cognition.
More from Research
- Open-source stack plans fuel-optimal A320 routes, could save thousands per transatlantic flight — JeremyCMorgan · 2026-09-26
- Preprint: selecting for 'freedom of function' beats likelihood-based evolution on 5 benchmarks — burny_tech · 2026-09-26
- Baker lab de novo designs cysteine proteases: 13 of 69 designs cleave paired substrates — AllThingsApx · 2026-09-26
- Simulated fruit fly brain is learning to fight its way through Battletoads — Inevitable_Emu2722 · 2026-09-26
- NSF Robotics Meeting Workshop: Four Researchers on Learning, Generalization and Physical Intelligence — YuXiang_IRVL · 2026-09-26
- Xiaomi MiMo open-sources its RL environments and training code on a fork of verl — eliebakouch · 2026-09-26