Neural flow geometry differentiates gambling and cognitive control despite matched sympathetic arousal

Using the NeuralManifoldDynamics pipeline, we show that local neural-flow-geometry metrics (the Meta-Noetic Jacobian) sharply differentiate a gambling task from cognitive control in the OpenNeuro ds004511 EEG dataset, even though global manifold coordinates and peripheral arousal (tonic EDA, corrugator EMG) do not differ between tasks. The effect survives a ten-part robustness audit, including a conventional-EEG baseline and a direct event/response-rate sensitivity check, supporting MNJ as a sensitive marker of task-context reorganization beyond global brain-state position.