Jean-Rémy KING - 19 mars 2026 à 10h30 au PK-5115 (201, ave President-Kennedy, 5e étage)
TITRE : Language: in search of neural code
RÉSUMÉ
Deep learning algorithms offer new methods to understand and model how language is processed in the human brain. Using both encoding (representation -> brain) and decoding (brain -> representations), we show that comparing modern speech and language models can account for brain responses to natural speech as recorded with EEG, MEG, iEEG and fMRI, including in children between 2 and 12 years old. This provides an operational foundation for modelling language in the adult and developing brain, and a new path to understanding the neural and computational bases of this human-specific ability.
BIOGRAPHIE
Jean-Rémy KING is a CNRS researcher at École Normale Supérieure currently seconded to Meta AI, where he leads the Brain & AI team, which aims to identify the cerebral and computational bases of human intelligence. The focus is on language, developing deep learning algorithms to decode and model brain activity recorded with MEG, EEG, electrophysiology and fMRI.
RÉFÉRENCES:
Evanson, L., Bulteau, C., Chipaux, M., Dorfmüller, G., Ferrand-Sorbets, S., Raffo, E., ... & King, J. R. (2025). Emergence of language in the developing brain. arXiv.
Lévy, J., Zhang, M., Pinet, S., Rapin, J., Banville, H., d'Ascoli, S., & King, J. R. (2025). Brain-to-text decoding: A non-invasive approach via typing. arXiv preprint arXiv:2502.17480.
Banville, H., Benchetrit, Y., d'Ascoli, S., Rapin, J., & King, J. R. (2025). Scaling laws for decoding images from brain activity. arXiv preprint arXiv:2501.15322.