Human Brain Mirrors LLMs Processing Sequence

Researchers from Hebrew University, Princeton and Google Research reported in Nature Communications that the human brain processes spoken language through a rapid, stepwise neural sequence mirroring layer-wise computations in large language models. The study shows early neural responses map to shallow model layers while later activity, especially in Broca's area, aligns with deeper contextual representations. The team released their full brain-recording dataset and language features for further research.