SPERA: Spherical Prior EEG Foundation Model with Geometry- and Frequency-Aware Latent Prediction

2026-10-04

AI Summary

This study addresses the challenge of model generalization in electroencephalography (EEG) analysis arising from heterogeneity across subjects, devices, and electrode configurations. To this end, we propose an EEG foundation model based on the Joint Embedding Predictive Architecture (JEPA). Departing from conventional signal reconstruction paradigms, the proposed model performs prediction directly within a latent space. It incorporates Legendre polynomials and spherical harmonics to encode spatial geometric priors, while integrating factorized spatiotemporal attention mechanisms with spectral relational regularization to achieve efficient representation learning. Evaluated across nine downstream tasks, the model attains the highest average balanced accuracy, demonstrating superior parameter efficiency and robust cross-scenario generalizability.

Abstract

Electroencephalography (EEG) provides a non-invasive measure of ongoing neural activity, but building general-purpose EEG models remains challenging due to the heterogeneity of subjects, devices, and electrode montages. Existing EEG foundation models predominantly rely on reconstruction-based objectives defined on the observed signal, which contains both neural and non-neural components. We introduce SPERA (Spherical Prior EEG Representation Architecture), an EEG foundation model that adopts the joint-embedding predictive architecture (JEPA) to predict in latent space. SPERA introduces a Legendre-polynomial spatial prior, incorporated into attention to encode varying scalp electrode geometries. Two further components adapt the model to EEG: factorized temporal and spatial attention interleaved with periodic full-attention blocks, and a relational spectral regularizer aligning latent similarity structure with spectral views. Pretrained on approximately 80,000 hours of EEG from 29,048 subjects across 106 datasets, SPERA achieves the highest average balanced accuracy across nine downstream tasks spanning clinical, cognitive, and BCI applications. SPERA further exhibits strong parameter efficiency under linear probing and robustness across varying recording conditions, suggesting its potential as a general-purpose backbone for diverse EEG analyses.

Problem

• EEG foundation model
• heterogeneity
• reconstruction objective
• non-neural components
• general-purpose

Innovation

• EEG foundation model
• JEPA
• Legendre-polynomial spatial prior
• factorized attention
• relational spectral regularizer