PLoS Computational Biology · 2017 · 96 citations · 55 references
Social SciencesBehavioral SciencesPsychologyCognitive PsychologyDecision TheoryIndividual DifferencesBayesian Experimental DesignBehavioral Decision-makingDecision Making Under RiskDecision-making ModelsInformation ProcessingObsessive-compulsive DisordersDecision Threshold AnalysisDecision Threshold MechanismsDecision ThresholdsDeliberationInformation GatheringSampling-based Decision MakingSubjective Decision MakingUrgency Signals
Patients with obsessive-compulsive disorder (OCD) can be described as cautious and hesitant, manifesting an excessive indecisiveness that hinders efficient decision making. However, excess caution in decision making may also lead to better performance in specific situations where the cost of extended deliberation is small. We compared 16 juvenile OCD patients with 16 matched healthy controls whilst they performed a sequential information gathering task under different external cost conditions. We found that patients with OCD outperformed healthy controls, winning significantly more points. The groups also differed in the number of draws required prior to committing to a decision, but not in decision accuracy. A novel Bayesian computational model revealed that subjective sampling costs arose as a non-linear function of sampling, closely resembling an escalating urgency signal. Group difference in performance was best explained by a later emergence of these subjective costs in the OCD group, also evident in an increased decision threshold. Our findings present a novel computational model and suggest that enhanced information gathering in OCD can be accounted for by a higher decision threshold arising out of an altered perception of costs that, in some specific contexts, may be advantageous.
55
Reinforcement Learning: An Introduction
Richard S. Sutton, Andy Barto · IEEE Transactions on Neural Networks · 1998 · 27.1K citations
Christopher J. Watkins, Peter Dayan · Machine Learning · 1992 · 9K citations · Full text
Christopher J. Watkins, Peter Dayan · Machine Learning · 1992 · 3.7K citations