AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

The Mind's Transformer: Computational Neuroanatomy of LLM-Brain Alignment

Cheng-Yeh Chen, Raghupathy Sivakumar

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Linguistic encoding in LLM representations

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use best-per-voxel intermediate transformer state instead of input hidden stateencoding model · Whole brainmean encoding correlation: 0.275 -> 0.296Improved
use best-per-voxel intermediate state instead of input hidden stateencoding model · Language networkmean encoding correlation: 0.433 -> 0.450Improved
use best-per-voxel intermediate state instead of input hidden stateencoding model · Auditory cortexmean encoding correlation: 0.407 -> 0.475Improved
use MindTransformer Mode 2 multi-state integration instead of input hidden stateencoding model · Heschl's Gyrusmean encoding correlation: 0.356 -> 0.467Improved
use Mode 1 single selected state instead of context vector baselineencoding model · Heschl's Gyrusmean encoding correlation: 0.363 -> 0.454Improved
use Mode 2 multi-state instead of Mode 1 single stateMindTransformer · Angular Gyrusmean encoding correlation: 0.419 -> 0.409Hurt
use Mode 2 instead of standard baselineencoding model · Language averagemean encoding correlation: 0.392 -> 0.402Mixed
use Mode 2 instead of Mode 1MindTransformer · Auditory averagemean encoding correlation: 0.438 -> 0.454Improved
use Mode 2 instead of standard baseline in per-subject analysisencoding model · Heschl's Gyrus (per-subject)mean encoding correlation: 0.104 -> 0.127Improved
scale LLM from 270M to 123B parametersencoding model · Auditory regionsmean encoding correlation: approximately 0.02-0.04Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy