Learning What to Remember: Selective Memory for Long-Context EEG Representation Learning

Long EEG recordings contain brief informative events within extended background activity. Learning from longer inputs therefore requires a representation that preserves useful evidence as context accumulates. We introduce SPARK, an encoder combining recurrent memory across temporal chunks, selective compression, and utility-guided controllers. Self-supervised pretraining learns to propagate and summarize context; subsequent utility adaptation refines the representation using curated diagnostic labels and physiological objectives. Under matched 120-second pretraining, SPARK reaches $87.67\pm 0.56\%$ recording-level balanced accuracy on CHB-MIT, compared with $76.19\pm 1.76\%$ for a near-parameter-matched dense-attention control. Utility adaptation raises performance to $92.14\pm 1.00\%$, exceeding a cross-entropy control with shared auxiliary objectives and matched adaptation data and updates ($90.57\pm 0.31\%$). Evaluation on three recording-level datasets and ten short-window benchmarks shows the strongest adaptation gain on CHB-MIT, alongside task-dependent context preferences and transfer behavior. These results identify recurrent memory and utility adaptation as complementary contributors to long-context EEG representation learning.