Inference in Dynamic Environments

Real world decision contexts are continually varying, and good decision makers must continually adjust their behavior to track environmental changes. This research investigated decision-making behavior in dynamically changing decision contexts. New experimental paradigms were developed in which dynamic decision making environments forced observers to change t heir decision-making strategies. Computational models for the decision process in dynamic environments were developed. One model is an ideal observer system in which statistical evidence for a changed environment is weighed in optimal fashion against evidence for a stable environment. The ideal observer analysis resulted in estimates for the optimal number of trials it takes to adjust to new decision environments. The comparison of the ideal observer model against empirical results revealed that some decision-makers adjust too quickly to changes in the environment, and hence perform sub optimally due to excess variability. Other individuals adjusted too slowly and failed to notice short -term temporal trends. A psychologically plausible particle filter model was also developed. This model accounted for all data, describing both optimal and sub-optimal performance. By varying the number of particles, and t he prior belief about the probability of a change occurring in the environment , most of the observed individual differences could be modeled.