The decision-making field has long attempted to understand the origins of human choice variability. A potential source of variability lies in spontaneous fluctuations of ongoing neural activity, which may influence behavior across several cognitive domains. Here, we developed a closed-loop intracranial brain-computer interface that detects transient fluctuations of broadband gamma activity (BGA, 70–150 Hz) to trigger stimulus presentations contingent on high or low neural states. Using this approach, we examined how spontaneous activity in the anterior insula influences accept/reject decisions involving multi-attribute offers combining pleasant and unpleasant components. Offers preceded by high BGA in the anterior insula were associated with a transient post-offer suppression in anterior insula activity, which biased decisions toward accepting an unpleasant item in exchange for a pleasant one in hypothetical multi-attribute choices. These findings demonstrate that sub-second endogenous neural fluctuations directly modulate choice behavior, challenging current neuro-computational models by emphasizing the critical role of intrinsic brain states in shaping decision variability. Using a closed-loop intracranial brain–computer interface, the authors demonstrate that spontaneous neural fluctuations can causally shape value-based decisions, highlighting the role of ongoing brain activity in human choice variability.