Online Self-Confidence Calibration for Improving Learning Outcomes Via Intelligent Tutoring Systems

Abstract

Intelligent tutoring systems (ITS) are designed to imitate human tutors by closing-the-loop between learners and tutoring agents. It is well-established that the cognitive factors of self-confidence and workload impact learners’ self-awareness of achievements and self-efficacy, which in turn enhances learning outcomes. However, little work has been done to operationalize these concepts in ITSs for psychomotor learning. In this work, the authors consider learners’ skill progression while repeatedly landing a quadrotor in a simulator. The landing simulator is enabled with automation assistance that can turn on or off; when on, the automation assistance augments the learner’s input to mimic an expert’s landing trajectory. The authors design an algorithm to calibrate learners’ self-confidence to their performance and compare it against learners’ who do not receive any assistance. Statistical analyses revealed that participants who received assistance according to the calibration algorithm demonstrated more self-efficacy and less fatigue than those who did not.