ERP-XTTN: Interpretable Prototype-Guided Cross-Attention for Cross-Subject ERP Classification

2026-06-01

AI Summary

Achieving calibration-free, cross-subject, and interpretable event-related potential (ERP) classification remains a key challenge in brain–computer interfaces. This work proposes the ERP-XTTN architecture, which introduces— for the first time—a prototype-guided, value-free projected cross-attention mechanism for ERP classification: EEG segments are routed via query–key cross-attention to fixed prototypes automatically constructed from extremal points of difference waves in the training set, rendering classification entirely dependent on attention-based routing and thereby endowing the model with intrinsic interpretability. Evaluated across three public datasets and eight ERP components, the model achieves near-optimal performance (AUROC gap ≤ 0.034) under both 3-channel and full-montage settings, demonstrates strong generalization under causal filtering and leave-one-subject-out cross-validation, establishes the first epoch-level leave-one-subject benchmark on ERP CORE, and yields misclassifications that are neurophysiologically interpretable.

Abstract

Interpretable brain-computer interface classifiers that generalize across subjects without calibration remain an open challenge. We test whether prototype-based cross-attention can provide competitive, interpretable event-related potential (ERP) classification under deployment-compatible conditions. We propose ERP-XTTN, a cross-attention architecture that routes input EEG patches to fixed difference-wave prototypes via query-key-only cross-attention with no value projection, so classification depends entirely on attention routing and attention faithfulness is structural rather than post-hoc. Prototypes are derived automatically from extrema in the training-fold difference wave. We evaluate across three public sources (BNCI Horizon 2020, HRI Cursor, and ERP CORE) spanning eight ERP components (ERN, LRP, ErrP, N170, P300, N2pc, MMN, N400), using leave-one-subject-out (LOSO) evaluation with causal filtering at two channel counts (3-channel and full montage), against EEGNet and xDAWN with Riemannian geometry (xDAWN+RG). The mean gap between the best baseline and ERP-XTTN was .018 AUROC at 3 channels and .034 at full montage, arising from two largely distinct sources: a temporal-flexibility cost relative to EEGNet and a spatial-exploitation cost relative to xDAWN+RG, the latter driven by signal-to-noise ratio at full montage. Beyond accuracy, the transparent routing reveals cross-subject signal structure that black-box models cannot: false positives resembled true positives more than true negatives did, indicating that classification errors are neurophysiologically explicable. ERP-XTTN generalizes across diverse ERPs under causal, calibration-free conditions with a small interpretability cost at minimal montages. To our knowledge, this is the first epoch-level LOSO benchmark on ERP CORE.

Problem

• cross-subject
• ERP classification
• interpretable
• calibration-free
• brain-computer interface

Innovation

• prototype-guided cross-attention
• calibration-free ERP classification
• interpretable BCI
• difference-wave prototypes
• cross-subject generalization