NeFut Logo NeFut
中 Admin Login

[CS.AI] Multilingual Brain Decoding Reveals Shared Cortical Semantic Geometry

Published at: 2026-10-22 22:00 Last updated: 2026-10-25 13:04
#AI #Machine Learning #Neural

We built a multilingual brain decoding framework and examined intracranial recordings from 227 participants speaking Chinese, English, and French. Cortical semantic representations displayed consistent geometric relationships across languages and individuals, revealing a shared cortical semantic geometry. This cross‑language consistency grew from sensorimotor to higher‑order association cortex, forming a semantic‑geometry gradient that aligns closely with the principal cortical functional gradient. The study demonstrates that multilingual brain decoding generalizes across datasets and that a stable cortical semantic geometry underlies semantic commonality across languages.

Review: This study provides the first large‑scale evidence of a unified semantic geometry in the human cortex, offering a compelling neural account of cross‑linguistic semantic similarity.

Original Source: https://arxiv.org/abs/2610.26927

[h] Back to Home