Improving Expert Decision Making Processes on High Impact Low Probability Events

Abstract

Decision-making for system design in the context of high-impact low probability events is challenging because individual cognitive biases tend to neglect extreme probabilities, with potentially catastrophic consequences. In this study, we aim to determine the circumstances under which decision-makers facing such events should rely on categorical, rather than precise, representations of risk. Our novel contribution is to test a strategy suggested by Fuzzy Trace Theory [1], which argues that risk-avoidance associated with categorical risk perception (e.g., “some risk” vs. “no risk”) may actually lead to better outcomes when compared to decisions based on precise (and potentially biased) interval-level representations of risk. We compare the results of computational models in which decision makers use these paradigms to react to high-impact low probability events. Our results elucidate the circumstances under which deviations from classical rationality may lead to better outcomes for design.