Let EEG Models Learn EEG

High-fidelity EEG generation is critical for alleviating data scarcity and addressing privacy constraints in large-scale neural modeling. Despite recent progress, most existing approaches formulate EEG generation via discrete denoising objectives, which inadequately reflect the inherently continuous temporal dynamics and spectral structure of neural activity. As a result, these methods often struggle to preserve long-range temporal dependencies and exhibit mismatches in the spectral and temporal structure of the generated signals. In this work, we argue that effective EEG generation requires models that operate directly on the continuous evolution of neural signals. We introduce Just EEG Transformer (JET), a generative framework based on conditional flow matching that models EEG as raw sequences evolving along continuous trajectories. By learning a smooth vector field that transports noise to the EEG data distribution, JET captures temporal continuity and transient dynamics without relying on discretized denoising schemes or domain-specific representations. To ensure that the learned dynamics remain consistent with key properties of EEG signals, we introduce principled constraints that preserve spectral structure, temporal stationarity, and signal-level statistics. Across three large-scale benchmarks, JET consistently achieves state-of-the-art performance, reducing TS-FID by over 40\% compared to strong baselines. Extensive analyses show that JET captures key structural properties of neural dynamics, providing a scalable and principled approach to EEG generation.

Brain activity recorded by electroencephalography (EEG) consists of small electrical signals picked up by sensors on the scalp. It helps clinicians diagnose conditions such as epilepsy, and could power a new generation of AI tools for neuroscience. Progress is held back by data scarcity, since high-quality clinical recordings are expensive to collect and tightly protected for privacy. A natural workaround is to generate realistic synthetic EEG, but existing methods borrow ideas from image generation and treat EEG as a noisy picture to be cleaned up step by step. Brain signals do not behave like images. They evolve as smooth, continuous waves with characteristic rhythms and occasional rare bursts that often carry the most important clinical information, and image-style methods tend to wash these features out. We introduce JET, which models EEG generation as a continuous process over time rather than a sequence of discrete denoising steps, guided by principled constraints that preserve the spectral, temporal, and statistical properties of brain signals. Across three large-scale clinical benchmarks, JET produces synthetic recordings substantially closer to real data than prior methods, and using these samples to augment downstream classifiers consistently improves accuracy, providing a scalable and privacy-preserving foundation for clinical neuroscience.