Proceedings of the Annual Meeting of the Cognitive Science Society

In complex tasks, high performers often have better strategies than low performers even with similar practice. Relativelylittle research has examined how people form and modify strategies in tasks that permit a large set of possible strategies.One challenge with such research is determining strategies based on behavior. Three algorithms were developed to trackthe task features people employ in their strategies while performing a complex task. An ACT-R model that performs thetask was created to collect simulated data with a range of known strategies. The performance of the three algorithms iscompared, and a decision tree classification algorithm yielded the best performance across the test cases. Summary datafrom applying the algorithms to human data on the tasks is also presented and highlights potential challenges for futurework. However, this approach to tracking strategy exploration may enable further development of theories about howpeople search for good strategies.