Acute Anxiety Selectively Enhances Value-Free Random Exploration through Frontoparietal Engagement

Motivational decision-making is most adaptive when exploitation of known rewards is effectively balanced against exploration of uncertain alternatives. Trait anxiety influences this balance across individuals, yet less is known about the dynamic effects of acute anxiety on behavior. This study combined a decision-making task, the multi-armed bandit Maggie's Farm task, with fMRI and computational modeling to examine how experimentally induced anxiety modulates exploration strategies (random, value-free exploration vs directed, novelty-seeking exploration) and their neural correlates. Forty-seven healthy participants (24 female; 23 male) completed the task under alternating conditions of safety and anxiety induced by unpredictable electric shocks. Exploratory behavior was promoted through different task decision horizons and modeled using parameters capturing random value-free exploration (ε) and novelty seeking (η). Acute anxiety impaired performance and decision consistency, selectively increasing random, value-free exploration without affecting novelty-driven exploration. Reaction times decreased, indicating faster, less deliberative responding. Neuroimaging revealed that horizon manipulations recruited circumscribed frontoparietal regions. Value and policy signals were dissociable, with ventromedial prefrontal cortex encoding expected value and superior parietal lobule mediating exploratory control. Critically, acute anxiety increased activation in right superior parietal lobule and frontopolar cortex specifically during random, value-free exploration, independently of value processing. Together with the behavioral findings, this pattern identifies a neural signature of anxiety-related decision noise. In conclusion, acute anxiety increases random, value-independent exploration, promoting less goal-directed, more stochastic behavior. These findings link anxiety-induced decision noise to frontoparietal activation and identify random exploration as a computational marker of dysregulated anxiety-related decision-making.