Quantifying the Experience of Difficulty: Behavioral and Trajectory Applications for Adaptive Systems

Many digital systems do not track behavioral metrics that are associated with perceptions of difficulty; however, this perception is an important driver of engagement and learning. The goal of this project was to identify metrics of problem-solving and determine their relationship with subjective difficulty perception. We modeled move-by-move progression through logic puzzles and found that revisiting prior moves and general problem-solving efficiency predicted both subjective difficulty and objective performance. Interestingly, path volatility, caused by a variety of deviations from the optimal path to the solution, uniquely predicted subjective difficulty perception. Using these markers may enable the design of more responsive systems that detect user struggle and prevent attrition.