[Paper Review] Self-Supervised Evolutionary Learning of Neurodynamic Progression and Identity Manifolds from EEG During Safety-Critical Decision Making

Xiaoshan Zhou, Carol C. Menassa+1 more|arXiv (Cornell University)|Mar 7, 2026

EEG and Brain-Computer Interfaces0 citations

TL;DR

The paper presents a self-supervised evolutionary learning framework that discovers individualized neurodynamic progressions and identity manifolds from continuous EEG during safety-critical decisions, enabling authentication and anomaly detection without external labels.

ABSTRACT

Human-vehicle interaction in safety-critical traffic environments increasingly incorporates neural sensing to infer user intent and cognitive state, yet most existing approaches either treat electroencephalography (EEG) as a static biometric credential or train task-specific decoders that ignore long-term neurodynamic trajectories, lacking mechanisms for secure user identity and continual modeling of evolving cognitive states. This work proposes a self-supervised evolutionary learning (SSEL) framework that discovers individualized neurodynamic progressions and intrinsic identity manifolds directly from continuous EEG, without external labels or predefined cognitive stage models. SSEL jointly optimizes within-stage temporal predictability, boundary contrast, cross-trial alignment, and sparse stage-specific feature weights, while a population-based evolutionary search enables direct optimization in the discrete, non-differentiable space of candidate segmentations. We validate the framework on EEG recorded from participants performing a simulated road-crossing decision task, a canonical safety-critical scenario in which perceptual assessment, risk evaluation, and decision commitment unfold over time. The learned segmentations reveal stable, person-specific stage structures and neurodynamic signatures that support authentication and anomaly detection. Compared to inference-based segmentation baselines, SSEL achieves orders-of-magnitude higher boundary contrast, substantial gains in cross-trial generalization of intention boundaries, and more interpretable, sparse stage-wise feature attributions. Beyond performance, the framework advances a progression-aware perspective on cognitive neurodynamics, where security, resilience, and personalization emerge from the intrinsic temporal structure of brain activity, with implications for next-generation smart urban and transportation infrastructures.

Motivation & Objective

• Motivate how EEG can reveal long-term neurodynamic trajectories in safety-critical decision making.
• Propose a self-supervised framework to learn progression structures without external labels or predefined cognitive stages.
• Jointly optimize within-stage temporal predictability, boundary contrast, cross-trial alignment, and sparse, stage-specific feature weights.
• Enable population-based evolutionary search to optimize discrete, non-differentiable segmentations directly from EEG data.

Proposed method

• Introduce Self-Supervised Evolutionary Learning (SSEL) to extract progression structures from continuous EEG.
• jointly optimize within-stage temporal predictability, boundary contrast, cross-trial alignment, and sparse stage-specific feature weights.
• Employ a population-based evolutionary search to optimize in the discrete, non-differentiable space of candidate segmentations.
• Validate on EEG data from a simulated road-crossing decision task to reveal person-specific stage structures.
• Demonstrate that learned segmentations provide stable neurodynamic signatures for authentication and anomaly detection.

Experimental results

Research questions

• RQ1 Can EEG-based neurodynamic progressions be discovered without external labels or predefined cognitive stage models?
• RQ2 Do personalized progression structures improve cross-trial generalization and boundary detection compared to inference-based segmentation?
• RQ3 Can identity manifolds and stage-specific features support authentication and anomaly detection in safety-critical tasks?

Key findings

• SSEL yields orders-of-magnitude higher boundary contrast than inference-based baselines.
• SSEL shows substantial gains in cross-trial generalization of intention boundaries.
• Learned segmentations yield interpretable, sparse stage-wise feature attributions.
• The framework reveals stable, person-specific stage structures and neurodynamic signatures from EEG.
• Progression-aware views on cognitive neurodynamics emerge, with implications for security and personalization in smart infrastructures.