Estimating Computational Models of Dynamic Decision Making from Transactional Data

The goal of this work is to estimate and validate computational models of dynamic decision making against data on sequences of actual decisions made in naturalistic settings. While this paradigm has its roots in laboratory studies under controlled conditions, increasing instrumentation of operational environments is enabling parallel investigations in field settings. Here, decision processes associated with the dispatch of debris removal personnel and equipment are investigated using data from a series of tornado storms in the State of Alabama in 2011. A multi-faceted approach to model validation is presented, thereby illustrating how objective, operational data may be used to inform models of complete decision making processes.