MEG-Mamba: A Scalable State-Space Foundation Model for Magnetoencephalography
Abstract
Magnetoencephalography (MEG) is an imaging technique that offers a non-invasive, millisecond-resolution view of human brain activity. The increasing availability of MEG data presents an opportunity to take advantage of a recent advance in artificial intelligence, namely self-supervised foundation models. Existing foundation models for MEG (and electroencephalography) have been built on a transformer architecture. Here, we introduce MEG-Mamba: a generative foundation model for neural activity (source reconstructed, parcellated MEG) built on a Mamba architecture. MEG-Mamba is trained to predict the next token of a discretised MEG signal, conditioned on a brain region and recording session. MEG-Mamba surpasses the generative fidelity of a transformer-based alternative (MEG-GPT) while pre-training in less time (22 vs 400 GPU-hours) and modelling a longer context (4 vs 0.32 s). We evaluate MEG-Mamba by examining the spatio-spectral characteristics of the data it generates and by interpreting its learned embeddings. Furthermore, we demonstrate that lightweight stimulus conditioning (LoRA) can be used to generate realistic task-evoked responses that are not included in pre-training. Our results suggest that Mamba is a promising backbone for scaling neural foundation models.
Related
- Bi-Mamba-Load: A bidirectional structured state space model with sequence relationship adaptive perception for electricity load forecasting
- Phys-Mamba: Physics-informed selective state-space fusion network for high-fidelity underwater image restoration
- Does a map-level insulation constraint improve sequence-based prediction of 3D genome folding? A Stage 1 Registered Report with a reverse-complement-equivariant state-space model (HiFold-Mamba)
- Neural oscillations and network specialisation in early childhood captured with next-generation magnetoencephalography
- Benchmarking single-cell foundation models reveals an advantage in the low-data regime for cell-type and disease-state prediction
BibTeX
@misc{gohil2026megmamba,
title = {{MEG-Mamba: A Scalable State-Space Foundation Model for Magnetoencephalography}},
author = {Chetan Gohil and SungJun Cho and Oiwi Parker Jones and Mark Woolrich},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.00746},
eprint = {2610.00746},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.00746},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - MEG-Mamba: A Scalable State-Space Foundation Model for Magnetoencephalography
AU - Chetan Gohil
AU - SungJun Cho
AU - Oiwi Parker Jones
AU - Mark Woolrich
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.00746
UR - https://arxiv.org/abs/2610.00746
AB - Magnetoencephalography (MEG) is an imaging technique that offers a non-invasive, millisecond-resolution view of human brain activity. The increasing availability of MEG data presents an opportunity to take advantage of a recent advance in artificial intelligence, namely self-supervised foundation models. Existing foundation models for MEG (and electroencephalography) have been built on a transformer architecture. Here, we introduce MEG-Mamba: a generative foundation model for neural activity (source reconstructed, parcellated MEG) built on a Mamba architecture. MEG-Mamba is trained to predict the next token of a discretised MEG signal, conditioned on a brain region and recording session. MEG-Mamba surpasses the generative fidelity of a transformer-based alternative (MEG-GPT) while pre-training in less time (22 vs 400 GPU-hours) and modelling a longer context (4 vs 0.32 s). We evaluate MEG-Mamba by examining the spatio-spectral characteristics of the data it generates and by interpreting its learned embeddings. Furthermore, we demonstrate that lightweight stimulus conditioning (LoRA) can be used to generate realistic task-evoked responses that are not included in pre-training. Our results suggest that Mamba is a promising backbone for scaling neural foundation models.
ER -