This study demonstrates that frozen language models can serve as effective neural predictors for brain activity during natural speech and text comprehension, while distinguishing predictive utility from claims about shared neural organization. The analysis of MEG and ECoG data revealed widespread positive prediction gains over low-level baselines, though participant-level advantages were localized rather than uniform.
- Analyzed locked derived data from Brain Treebank, MEG-MASC, and Podcast EcoG using eight frozen language models and blocked encoding models.
- Found that 67 of 432 evaliable rows met a controlled predictive-only criterion across Brain Treebank and Podcast ECoG datasets.
- Confirmed component-level sensitivity through brain-derived, timing-linked, acoustic, and implanted-signal controls.
- Demonstrated that model-side feature ablations changed prediction scores in most evaluable source rows.
These findings identify language-model features as useful neural predictors and clarify that predictive usefulness is distinct from claims regarding shared neural organization or specific language-processing computations.