Project R-17038

From Performance to Conceptual Knowledge: Externalizing and Transferring Expert Knowledge for True Skill Learning (Research)

This research addresses the limitation of current digital training systems that prioritize movement imitation over the acquisition of conceptual knowledge, the rationale and strategies behind expert actions. To enable true skill transfer, this research overcomes three critical technical barriers: experts' difficulty in articulating the tacit cues and decision logic underlying their performance, the prevalence of feedback that provides prescriptive guidance without fostering deep conceptual understanding, and adaptation policies driven by surface performance metrics rather than the learner’s internal mental model. The project pursues three research objectives to operationalize a concept-aware training framework. First, it will develop a computational pipeline to localize "meaningful moments" in performance recordings, using these as triggers to scaffold the externalization of expert conceptual knowledge. Second, it will generate reflective, descriptive, and prescriptive feedback grounded in this expert knowledge to shift the focus from movement correction to situational awareness and adaptability. Finally, the project will create an interactive platform that dynamically regulates feedback based on the learner’s inferred conceptual development, aiming to optimize skill transfer and adaptability. By grounding abstract reasoning in objective sensorimotor data, this work aims to cultivate robust expertise that generalizes across diverse contexts.

01 November 2026 - 31 October 2030