Tracing the learning path: Revealing skill acquisition phases in a virtual environment through behavioral tracking

Through repeated practice, humans acquire skills and develop a high level of expertise, but this learning process is sometimes long and costly. It may even be impossible to train in situ, although training in virtual environments can offer an alternative. Such supports also have the capacity to automatically collect a large amount of behavioral data, which can be used to personalize training based on learners' requirements. To help identify individual needs, the present study focuses on the three phases characterizing the acquisition of a new skill and distinguishable at behavioral, cognitive, and neuropsychological levels. We present a behavioral approach to phase identification based on the number of times learners consult instructions and the number of incorrect actions they make. Sixty-three participants were asked to repeat an assembly procedure in a virtual environment in two sessions, one week apart. Analyses of activity logs and mental workload (NASA-TLX) throughout the two sessions revealed a progressive acquisition of the procedure, with evidence of retrieval from long-term memory. The identified phases align with the behavioral and cognitive patterns described in the Adaptive Control of Thought-Rational (ACT-R) theory and prior research. Further validation of the proposed phase-identification method could involve physiological measures and a dual-task paradigm to more effectively assess automaticity and phase transitions. Overall, these findings provide a real-time, non-intrusive, and validated method for identifying learning phases, offering prospects for developing adaptive virtual environments tailored to training and rehabilitation.