Temporal Feature Extractors in EEG Foundation Models: A Controlled Comparison Including a Pretrained Time-Series Model

2026-06-29

AI Summary

This study investigates whether general-purpose pretrained time series foundation models, such as MOMENT, can be effectively transferred to electroencephalography (EEG) representation learning and examines the impact of different temporal feature extractors on downstream tasks. Within a unified EEG foundation model framework, the authors compare a linear baseline, a convolutional encoder, and a frozen pretrained MOMENT model, evaluating their performance on motor imagery and emotion recognition tasks. Notably, this work is the first to employ a non-EEG-specific pretrained time series model as a frozen feature extractor in EEG modeling, thereby assessing its cross-domain transferability. The results demonstrate that motor imagery benefits from relatively simple temporal modeling, whereas emotion recognition requires more complex temporal structures; despite not being specifically optimized for EEG, MOMENT still exhibits strong feature extraction capabilities.

Abstract

Electroencephalography (EEG) foundation models aim to learn generalizable representations from large-scale brain recordings. However, the role of temporal feature extractors and whether pretrained time-series foundation models (TSFMs) can be effectively transferred to this setting remains underexplored. We conduct a controlled comparison of three temporal feature extraction strategies, including a linear baseline, a convolutional encoder, and a frozen pretrained TSFM (MOMENT), within a unified EEG foundation model. We evaluate their impact on representation quality using two downstream tasks: motor imagery and emotion recognition. Results reveal different trends across the evaluated benchmarks. On the motor imagery dataset, simple temporal representations perform competitively, whereas the emotion dataset benefits from richer temporal modeling. Although not specifically adapted to EEG, the pretrained TSFM serves as an effective temporal feature extractor, suggesting that general-purpose time-series representations can be transferred as frozen temporal feature extractors within EEG foundation models.

Problem

• EEG foundation models
• temporal feature extractors
• pretrained time-series models
• representation transfer
• time-series representation

Innovation

• EEG foundation models
• temporal feature extraction
• pretrained time-series models
• transfer learning
• MOMENT