Computational Cognitive Modeling of Adaptive Choice Behavior in a Dynamic Decision Paradigm

摘要

In this project, we braided together three strands of research to study how interface design and decision strategies interact to affect adaptive choice behavior in a dynamic decision paradigm. In the first strand are computational models of dynamic decision-making built within a well-recognized cognitive architecture, ACT-R. Our second strand is composed of the decades- plus research efforts that show that people adopt decision-making strategies that tradeoff effectiveness for efficiency. Of particular importance is work showing that the way in which information is presented and how decision aids are constructed may unintentionally influence the decision strategies adopted. The third strand is work from our own lab showing the influence of hard and soft constraints on routine interactive behavior. This work suggests that the cognitive system works locally to optimize non-deliberate choices but that the sum of such locally optimized choices may be far from an optimized global outcome. These strands are braided together in Argus Prime: a simulated task environment of a radar operator's task that was designed to study the interaction between the sometimes conflicting top-down and bottom-up demands.