Benchmarking Positional Encoding Strategies for Transformer-Based EEG Foundation Models

2026-05-28

AI Summary

This study addresses the challenge of positional encoding in Transformer-based EEG signal processing, where permutation invariance necessitates explicit spatial information yet no universally effective encoding scheme exists due to the complex spatial arrangement of electrodes. Within the CBraMod architecture, the authors systematically evaluate five positional encoding strategies—including Spherical Positional Encoding (SPE) and Asymmetric Conditional Positional Encoding (ACPE)—using self-supervised pretraining followed by linear probing and fine-tuning across motor imagery classification and emotion recognition tasks. The results demonstrate that the optimal positional encoding is task-dependent: SPE achieves superior performance in motor imagery tasks, while ACPE exhibits greater stability in multi-task settings. These findings refute the feasibility of a one-size-fits-all positional encoding approach and provide empirical guidance for selecting appropriate encodings in EEG decoding applications.

Abstract

Electroencephalography (EEG) is a widely used non-invasive technique for measuring brain activity in brain-computer interface (BCI) applications. Supervised EEG decoding models often struggle to generalize across tasks, subjects, and datasets, motivating transformer-based EEG foundation models trained with self-supervised learning. Since transformers are permutation-invariant, they require explicit positional information. Unlike textual tokens, EEG electrodes are spatially distributed across the scalp, raising the question of how electrode positions should be encoded in transformer-based EEG models. In this study, we benchmark five positional encoding strategies within the CBraMod backbone and evaluate them under linear probing and fine-tuning protocols on motor imagery classification and emotion recognition. Our results show that no single strategy consistently outperforms across tasks. Spherical Positional Encoding (SPE) yields strong representations for motor imagery but underperforms on emotion recognition, while Asymmetric Conditional Positional Encoding (ACPE) demonstrates more consistent performance across tasks. These findings suggest that the optimal positional encoding strategy is task-dependent, with no universal solution across EEG decoding scenarios.

Problem

• positional encoding
• EEG
• transformer
• foundation model
• electrode spatial distribution

Innovation

• positional encoding
• EEG foundation models
• transformer
• self-supervised learning
• task-dependent representation