Temporal Context and Architecture: A Benchmark for Naturalistic EEG Decoding

TL;DR

This paper benchmarks various neural network architectures for naturalistic EEG decoding, analyzing how model design and temporal context affect accuracy, robustness, and efficiency across different tasks and conditions.

Contribution

It provides a comprehensive comparison of five architectures on a naturalistic EEG dataset, highlighting trade-offs between accuracy, robustness, and parameter efficiency.

Findings

Longer temporal context improves decoding accuracy.

S5 achieves high accuracy with fewer parameters.

EEGXF offers more robustness to frequency shifts.

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.