Frozen language model features predict held-out neural activity in naturalistic comprehension across multiple brain recording modalities with gains over low-level baselines in many but not all sources, after extensive controls.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Heterogeneous Neural Predictivity from Language Models During Naturalistic Comprehension
Frozen language model features predict held-out neural activity in naturalistic comprehension across multiple brain recording modalities with gains over low-level baselines in many but not all sources, after extensive controls.