Temporal Context and Architecture: A Benchmark for Naturalistic EEG Decoding

2026-01-29

AI Summary

This study systematically investigates the interplay between model architecture and temporal context length in decoding naturalistic electroencephalography (EEG) signals, and its impact on performance and robustness. Leveraging the HBN movie-watching dataset, we evaluate five architectures—CNN, LSTM, EEGXF, S4, and S5—across temporal segments ranging from 8 to 128 seconds, and assess their real-world robustness under cross-frequency shifts, out-of-distribution tasks, and leave-one-subject-out settings. We uncover a novel efficiency–robustness trade-off between temporal context and architecture: S5 achieves 98.7% ± 0.6% accuracy at a 64-second context with only 1/20th the parameters of CNN, making it ideal for high-accuracy, low-overhead applications, whereas EEGXF demonstrates superior robustness under frequency shifts, favoring scenarios requiring conservative uncertainty estimation.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Natural Language Processing: Safety and Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

We study how model architecture and temporal context interact in naturalistic EEG decoding. Using the HBN movie-watching dataset, we benchmark five architectures, CNN, LSTM, a stabilized Transformer (EEGXF), S4, and S5, on a 4-class task across segment lengths from 8s to 128s. Accuracy improves with longer context: at 64s, S5 reaches 98.7%+/-0.6 and CNN 98.3%+/-0.3, while S5 uses ~20x fewer parameters than CNN. To probe real-world robustness, we evaluate zero-shot cross-frequency shifts, cross-task OOD inputs, and leave-one-subject-out generalization. S5 achieves stronger cross-subject accuracy but makes over-confident errors on OOD tasks; EEGXF is more conservative and stable under frequency shifts, though less calibrated in-distribution. These results reveal a practical efficiency-robustness trade-off: S5 for parameter-efficient peak accuracy; EEGXF when robustness and conservative uncertainty are critical.

Problem

• naturalistic EEG decoding
• temporal context
• model architecture
• robustness
• cross-subject generalization

Innovation

• EEG decoding
• temporal context
• model architecture
• robustness evaluation
• parameter efficiency