Language Impairment as a Lens onto Large Language Models
Muxuan He · Wang Bill Zhu · Yue Chen · Elsi Kaiser · Weizhi Du · Yao Du · Samuel Nastase · Yanjun Gao
Uses classifier-guided lesioning and activation steering derived from human aphasic transcripts to induce and map Broca-like and Wernicke-like breakdowns across 11 language models.
JOURNALScientific Reports · Nature Portfolio19 OCT 2026
Simultaneous multi-task assessment of five neurological language disorders via cross-corpus multimodal learning
Byounghwa Lee · Hwa Jeon Song · Byung Ok Kang
A unified multimodal model classifies aphasia, Alzheimer disease, MCI, TBI and right-hemisphere damage across five TalkBank corpora while estimating severity and progression.
Language impairmentSpeech · text · metadataCross-corpus decoding
Biophysically constrained learning from healthy neural priors improves motor decoding in pathological cortical networks
Yihan Wu · Tao Chang · Siliang Chen · Yuxin Quan · Qiang Sun · et al.
Pathology-Aware Consensus Training transfers healthy neural priors to high-density ECoG from pathological cortex, improving decoding and restoring distorted semantic geometry.
JOURNALCommunications Biology · Nature Portfolio22 OCT 2026
Individualized morphometric-similarity deviations in autism linked to cortical hierarchy and microarchitecture
Makliya Mamat · Yiyong Chen · Lin Li
Normative modeling across 1,344 brains maps individualized ASD cortical deviations and links them to cortical gradients, transcriptomics and neurotransmitter-receptor density.
Efficient and accurate steering of large language models through attention-guided feature learning
Parmida Davarmanesh · Ashia Wilson · Adityanarayanan Radhakrishnan
An attention-guided framework identifies concept-bearing tokens and layers, nearly doubling successful activation steering across 512 concepts and models up to 70B parameters.
Unstable word form representations and greater reliance on right hemisphere during word learning in adults with a history of developmental language disorder
A. Nora · C. Yan · V. Husa · V. Wilkki · M. van Vliet · E. Service · R. Salmelin
MEG reveals unstable phonological word-form representations and increased right-hemisphere recruitment during vocabulary learning in adults with developmental language disorder.
Individual Differences in Sustained Attention Tune Semantic Abstraction in Neural Encoding of Narratives
S. Gao · Y. C. Leong · M. D. Rosenberg
GPT-2 encoding models over audiobook fMRI show that sustained attention shifts neural representations toward compressed gist-level and predictive semantic formats.
Frequency-Resolved Connectivity Graphs for EEG-Based Classification of Alzheimer and Frontotemporal Dementia
M. Radwan · A. Yazidi · P. G. Lind
A compact graph model uses frequency-resolved EEG connectivity to distinguish Alzheimer disease and frontotemporal dementia while exposing electrode- and band-level saliency.
Mapping the brain signature of dyslexia risk onto histamine, acetylcholine and glutamate receptor systems
M. C. Celik · Y. Cicek · M. A. Sakiroglu · D. Aslan · A. T. Altunc · H. A. Velioglu
Links polygenic structural-brain signatures of dyslexia risk to normative neurotransmitter maps and reading skill, identifying a testable biological axis for impairment.
When Words Sound Alike: Decoding Homophone Processing in Chinese Primary Progressive Aphasia
B. L. Tee · T.-F. Chen · L. K.-C. Li-Ying · Y. Chuang · et al.
Across Mandarin- and Cantonese-speaking cohorts, homophone impairment differentiates PPA from controls and maps to left anterior and inferior temporal atrophy.
Beyond Agrammatism: Syntax Processing in Chinese Primary Progressive Aphasia
Y.-W. Cheng · T.-F. Chen · L. K.-C. Li-Ying · R. Y. Lo · et al.
Shows syntax impairment across Chinese PPA variants and associates comprehension and production with distributed frontotemporal and frontostriatal anatomy.
CortexDecoder: fMRI-to-image Reconstruction from Cortical Flatmaps
T. Imoto-Tar · M. Fung · A. Gurha
Maps cortical fMRI flatmaps through transformer brain tokens and separate semantic and low-level bridges to reconstruct viewed images with diffusion models.
EEG-Captioner: Grounding EEG-to-Language Generation in Fundamental Signal Properties
Leyuan Jin · Jiahao Mei · Yixuan Yuan · Shimeng Li · Mengyue Wu · Yue Ding
Introduces 160,781 EEG segments with signal-grounded language annotations and tests whether EEG-to-language models preserve spectral, spatial and temporal facts.
Network geometry shapes the transformation of task representations across the human cortex
Lakshman N. C. Chakravarthula · Takuya Ito · Alexandros Tzalavras · Michael W. Cole
Shows how intrinsic connectivity dimensionality compresses or expands task representations, offering a systems-level principle for flexible biological and artificial computation.
Probing Spatial and Spectral Grounding in EEG Foundation Models
Kleanthis Avramidis · Ivan Tashev · Dimitra Emmanouilidou
Tests whether EEG foundation models actually rely on physiological channel and frequency structure, separating predictive performance from neural grounding.
BrainCodes: A Lightweight EEG Foundation Model over Discrete Neural Codes
Kleanthis Avramidis · Ivan Tashev · Dimitra Emmanouilidou
Learns discrete neural codes from more than 800 hours of EEG while using up to 97% fewer trainable parameters, suggesting compact principles for transferable AI.
Are Linguistic and Arithmetic Circuits Dissociated in LLMs?
Riku Kisako · Tatsuki Kuribayashi · Ryohei Sasano
Uses neurolinguistic dissociation as a hypothesis for mechanistic interpretability and finds largely separate linguistic and arithmetic circuits across five LLMs.
Finds low-dimensional rotational dynamics across spikes, calcium imaging, fMRI and MEG, then uses recurrent-network models to explain how ongoing brain state constrains behavior.