Human striatal population state dynamics

Animal models reveal that striatal projection neurons (SPNs) fluctuate between discrete electrophysiological states with distinct levels of cortical interaction. These dynamics and their behavioral relevance remain uncharacterized in humans. Leveraging neurobiologically informed modeling of over 3 billion voxel-frame-wise striatal coactivation profiles with cortex in functional magnetic resonance imaging (fMRI), we identified population-level striatal states in humans resembling canonical SPN states that reorganized systematically with task demands, arousal and behavior. A background of low- and moderate-coactivation "down-like" and "up-like" striatal rest states with high transition reciprocity modulated task reaction times and reward reactivity. Meanwhile, sparse, disproportionately high-magnitude 'bursts' of striatal coactivation with cortex, which emerged preferentially from the up-like rest state and whose cortical input composition varied with task context, tracked task engagement and arousal level. Findings bring a critical feature of corticostriatal neurobiology into systems-level view in humans and reveal a subthreshold state architecture whose balance encodes behaviorally relevant information.