2025 Hyper Recent •CC0 1.0 Universal

This work is dedicated to the public domain. No rights reserved.

Access Preprint From Server
June 5th, 2025
Version: 1
University Erlangen-Nurenberg
neuroscience
biorxiv

Prediction, Syntax and Semantic Grounding in the Brain and Large Language Models

Koelbl, N.Open in Google Scholar•Rampp, S.Open in Google Scholar•Kaltenhaeuser, M.Open in Google Scholar•Tziridis, K.Open in Google Scholar•Maier, A.Open in Google Scholar•Kinfe, T.Open in Google Scholar•Chavarriaga, R.Open in Google Scholar•Krauss, P.Open in Google Scholar•Schilling, A.Open in Google Scholar

Language comprehension involves continuous prediction of upcoming words, with syntactic structure and semantic meaning intertwined in the human brain. To date, few studies have used combined magnetoencephalography (MEG) and electroencephalography (EEG) measurements to investigate how syntactic processing, predictive coding, and semantic grounding interact in real time. Here we present the first combined MEG-EEG investigation of syntactic processing and semantic grounding under naturalistic conditions. Twenty-nine healthy participants listened to a German audio book while their neural responses were recorded. Event-related fields and event-related potentials for four word classes - nouns, verbs, adjectives, and proper nouns - showed highly reproducible, characteristic spatio-temporal signatures, including significant pre-onset activity for nouns, suggesting enhanced predictability of this word class. Source-space analyses revealed pronounced activation in the pre- and post-central gyri for nouns, suggesting a deeper semantic grounding of nouns in e.g. sensory experiences than verbs. To further investigate predictive mechanisms, we analyzed the hidden representations of the large language model Llama. By comparing the transformer-based representations to neural responses, we explored the relationship between computational language models and human brain activity, offering new insights into syntactic and semantic prediction. These findings highlight the power of simultaneous MEG-EEG recordings in unraveling the predictive, syntactic, and semantic mechanisms that underlie the comprehension of natural language.

Similar Papers

biorxiv
Fri Jun 06 2025
aNy-way ICA and its application to estimate cortico-thalamo-cerebellar functional links in schizophrenia
Multimodal data collected by international and national biobanking efforts have distinct scales and model orders and provide unique and complementary insights into disease mechanisms. We propose a novel, flexible and efficient data fusion approach, aNy-way independent component analysis (aNy-way ICA). aNy-way ICA fuses N-way multimodal or multidomain data by optimizing the entire loading correlati...
Duan, K.
•
Silva, R. F.
•
Rahaman, M. A.
•
Fu, Z.
...•
Calhoun, V. D.
biorxiv
Fri Jun 06 2025
Estrogen profiling in blood and brain: effects of season and an aggressive interaction in a songbird
Neuroestrogens are synthesized in the brain and regulate social behavior and cognition. In the song sparrow (Melospiza melodia), 17{beta}-estradiol (17{beta}-E2) promotes aggression, even during the non-breeding season, when circulating 17{beta}-E2 levels are low. Measuring estrogens is challenging due to their low concentrations and the limited sensitivity of many existing assays. Moreover, estro...
Jalabert, C.
•
Liu, M. Q.
•
Soma, K.
biorxiv
Fri Jun 06 2025
Working memory modulates auditory perceptual sensitivity during speech planning
Purpose: Working memory plays a critical role in speech production. However, how working memory influences speech systems remains unclear. This study aimed to investigate how maintaining a vowel in working memory alters perceptual sensitivity during speech planning. Method: Thirty healthy adults completed speaking and reading tasks while performing a concurrent working memory task. In speaking blo...
Merrikhi, Y.
•
Daliri, A.
biorxiv
Fri Jun 06 2025
Regulators of Interferon-Responsive Microglia Uncovered by Genome-wide CRISPRi Screening
Microglia dynamically support brain homeostasis through the induction of specialized activation programs or states. One such program is the Interferon-Responsive Microglia state (IRM), which has been identified in developmental windows, aging, and disease. While the functional importance of this state is becoming increasingly clear, our understanding of the regulatory networks that govern IRM indu...
McQuade, A.
•
Castillo, V. C.
•
Hagan, V.
•
Liang, W.
...•
Kampmann, M.
biorxiv
Fri Jun 06 2025
Neurogenesis-dependent transformation of hippocampal memory traces during systems consolidation
Memories for events (i.e., episodic memories) change qualitatively with time. Systems consolidation theories posit that organizational changes accompany qualitative shifts in memory resolution, but differ as to the locus of this reorganization. Whereas some theories favor inter-regional changes in organization (e.g., hippocampus[->]cortex; multiple trace theory), others favor intra-regional reo...
Golbabaei, A.
•
Coelho, C. A. O.
•
de Snoo, M.
•
Josselyn, S.
•
Frankland, P.
biorxiv
Fri Jun 06 2025
Extracellular Vesicles from Multiple Sclerosis White Matter Exhibit Synaptic, Mitochondrial, Complement, and Aging-related Pathway Dysregulation
Extracellular vesicles (EVs) are increasingly recognized as mediators of central nervous system (CNS) function and pathologies, including multiple sclerosis (MS). While plasma-derived EVs have been explored as biomarkers in MS, little is known about EVs in CNS tissue. Here, we characterize EVs from postmortem white matter (WM) of MS and control brains. EVs were separated by differential centrifuga...
Jank, L.
•
Satheesh Kumar, M. K.
•
Ryu, T.
•
Thapa, R.
...•
Bhargava, P.
biorxiv
Fri Jun 06 2025
CAN TRANSFER LEARNING IMPROVE SUPERVISED SEGMENTATIONOF WHITE MATTER BUNDLES IN GLIOMA PATIENTS?
In clinical neuroscience, the segmentation of the main white matter bundles is propaedeutic for many tasks such as pre-operative neurosurgical planning and monitoring of neuro-related diseases. Automating bundle segmentation with data-driven approaches and deep learning models has shown promising accuracy in the context of healthy individuals. The lack of large clinical datasets is preventing the ...
Riccardi, C.
•
Ghezzi, S.
•
Amorosino, G.
•
Zigiotto, L.
...•
Avesani, P.
biorxiv
Fri Jun 06 2025
Phosphoproteome modifications and cortical circuit dysfunction are linked to the early-stage progression of alpha-synuclein aggregation.
Cortical dysfunction is increasingly recognized as a major contributor to the non-motor symptoms associated with Parkinsons disease (PD) and other synucleinopathies. Although functional alterations in cortical circuits have been observed in preclinical PD models, the underlying molecular mechanisms are unclear. To bridge this knowledge gap, we investigated tissue-level changes in the cortices of r...
Dutta, S.
•
Hensel, J.
•
Scott, A.
•
Mohallem, R.
...•
Rochet, C.
biorxiv
Fri Jun 06 2025
Neural coding of spectrotemporal modulations in the auditory cortex supports speech and music categorization
Humans effortlessly distinguish speech from music, but the neural basis of this categorization remains debated. Here, we combined intracranial recordings in humans with behavioral categorization of speech or music taken from a naturalistic movie soundtrack to test whether the encoding of spectrotemporal modulation (STM) acoustic features is sufficient to categorize the two classes. We show that pr...
Ginzburg, J.
•
Debaque, E. C.
•
Borderie, A.
•
Morillon, B.
...•
Zatorre, R.
biorxiv
Fri Jun 06 2025
Widespread cortical representations of innate behaviors in the mouse
Across cortex, a substantial component of observed neuronal activity can be explained by movement. Voluntary movements elicit cortical activity that contain elements related to both motor planning and sensory activity. However, automatic movements that do not require conscious processing, for example grooming or drinking, are to a large extent controlled by subcortical systems such as the brainste...
Michelson, N. J.
•
Gupta, P. K.
•
Murphy, T. H.