Various psychiatric symptoms are often accompanied by impairments in decision-making. Given the high comorbidity of psychiatric disorders, symptoms that specifically couple with the impairment remain unidentified. The computations involved in decision-making that mediate the coupling are also elusive. Here, we conducted an online experiment with healthy individuals (n=939), participating in a decision-making task and completing questionnaires about psychiatric symptoms.The questionnaire data identified two dimensions underlying various symptoms: the first is mainly associated with obsessive-compulsive traits whereas the second is associated with depression and anxiety. Furthermore, by model-based analysis, we found only the first dimension was negatively correlated with the overall performance of the decision-making task, and the negative correlation was mediated by choice stochasticity (random exploration) and propensity to choose options previously unchosen. Given that the preference for previously unchosen options can reflect uncertainty-driven exploration, our findings highlight a key role of exploration-related strategies in psychiatric deficits of decision-making.Decision-making to attain reward is crucial for our survival, such as where to hunt (e.g., forest or lake) and where to invest (e.g., risky equities or safe bonds) 1 . Recent studies in psychology, neuroscience, and psychiatry have suggested that impairments in decisionmaking are often accompanied by mental disorders such as obsessive-compulsive disorder 2,3 , depression 4 , anxiety 5 , and schizophrenia 6 . However, little is known about the mechanism that attenuates decision performance. Here in this study, we provide a mechanistic account of how certain psychiatric symptoms are associated with deficits in decision-making.An increasing number of studies has well-described the decision-making process using a machine learning algorithm known as reinforcement learning 7 . The reinforcement learning framework posits that appropriate decisions require individuals to learn the expected value of each option from the reward experience. Furthermore, in order to maximise long-term reward, they need to choose a less familiar (more uncertain) option on some occasions in order to collect more information about its value (i.e., exploration), while choosing the sure rewarding option on most of the occasions (i.e., exploitation) [8][9][10] . A simple strategy of exploration is to increase choice stochasticity and this is known as random exploration 11 . A more sophisticated strategy would be driven by uncertainty about the value of each option 12 , that is, uncertain options are more likely to be chosen than promising options. The exploit-explore balance is of particular importance in changing environments.The reinforcement learning account of the decision-making process has been of considerable concern in many research fields such as psychology, economics, neuroscience, and psychiatry, as it not only captures the organisms' behaviours but also their neural acti...