Same Brain, Different Prediction: How Preprocessing Choices Undermine EEG Decoding Reliability

2026-05-07

AI Summary

This study addresses the high sensitivity of EEG decoding performance to preprocessing pipelines, which substantially undermines prediction reliability. The authors model preprocessing choices as a space of counterfactual interventions and propose an efficient method based on Walsh–Hadamard decomposition to quantify and disentangle the effects of different intervention combinations on deep learning predictions. They introduce a novel metric, termed “preprocessing uncertainty” (PU), to diagnose instability arising from such choices, and design a graph-structured regularizer, NA-PGI, to enhance model robustness. Experiments across six datasets reveal that single-trial predictions can flip in up to 42% of cases due to preprocessing variations, while the proposed approach effectively measures and significantly mitigates this instability.

Abstract

Electroencephalography (EEG) is a cornerstone of brain-computer interfaces and clinical neuroscience, yet deep learning models are typically trained and evaluated under a single, unreported preprocessing pipeline. We formalize preprocessing choices as a counterfactual intervention space and show that EEG predictions are surprisingly unstable under this space: across six datasets spanning four paradigms, up to 42% of trial-level predictions flip when only the preprocessing changes, a variability that standard uncertainty methods do not explicitly quantify because they condition on a fixed preprocessing pipeline. We provide three tools to make this instability measurable, decomposable, and reducible. First, a Walsh-Hadamard decomposition of the 2^7 pipeline space reveals that sensitivity is near-additive in practice under the binary intervention design, enabling efficient step-by-step optimization. Second, we introduce Preprocessing Uncertainty (PU), a per-trial diagnostic that captures a dimension of instability complementary to model-based confidence. Third, we study Normalized Adaptive PGI (NA-PGI), a graph-structured regularizer that exploits the compositional structure of preprocessing interventions as one mitigation strategy with clear scope conditions.

Problem

• EEG decoding
• preprocessing variability
• prediction instability
• uncertainty quantification
• counterfactual intervention

Innovation

• preprocessing uncertainty
• counterfactual intervention
• Walsh-Hadamard decomposition
• graph-structured regularization
• EEG decoding reliability