Nick Barendregt, Department of Applied Mathematics, University of Colorado Boulder
Normative and Dynamic Urgency in Dynamic Decision Environments
In uncertain and dynamic environments, humans must adapt their decision-making strategies to ensure accurate and timely choices. For example, in environments where the reward for a correct decision depends on time, individuals will increase the speed of their deliberation to avoid losing too much reward. This increase, known as "decision urgency", has been studied in behavioral models that can fit subject data, but it is less understood if/how urgency is optimal to decision-making and what the scope of urgency behaviors is.
In this talk, we take a normative, Bayesian approach to derive the optimal decision strategy for urgency-provoking decision tasks. We find that there is a large range of urgency-like behaviors that arise depending on the parameters of the task. We then compare our model to the standard decision urgency model and find that while both models can fit subject data, the normative model is favored using standard model comparison metrics. We conclude by exploring how our findings generalize to different tasks by proposing a modification to the well-established dot motion task.