Mapping the neuronal building blocks of human language with language models

Researchers identified neurons in the human brain that encode fundamental language components, including grammatical relationships, parts of speech, and syntactic structures of phrases and sentences. These findings, published in Nature on June 17, 2026, were achieved through wide-scale neuronal recordings. The study utilized advanced language models to map these neural responses, demonstrating a significant overlap between the representations learned by artificial intelligence and the biological processing of language in humans. This research provides a deeper understanding of how the brain constructs and comprehends complex linguistic information, potentially paving the way for new insights into language disorders and the development of more sophisticated AI language systems. The work highlights the power of integrating computational neuroscience with artificial intelligence to unravel the complexities of the human mind.