Code for Joint processing of linguistic properties in brains and language models

ARENA Deliverable [P5_WP2_01] Approach to estimate shared and unique info between an AI model and brains w.r.t. linguistic property

Code for estimating the contribution of linguistic properties to brain–language model alignment. It can be accessed in this link: https://github.com/subbareddy248/lingprop-brain-alignment

The repository includes scripts for extracting language model representations, probing and removing linguistic property information, training brain encoding models on fMRI data, and quantifying the shared and unique contributions of surface, syntactic, and semantic properties to the alignment between language models and human neural responses.