How Can Pedagogical Agents Detect Learner’s Stress?

Abstract

Learning analytics is aimed to analyze different types of data about the learning process of learners and to provide appropriate pedagogical intervention. However, existing research work on learning analytics mostly focuses on cognitive data (e.g., number of problems solved, number of correct answers, lessons visited) and physical data (e.g., clicks on specific media). The research question to be investigated is how to detect and analyze physiological data of learners in real-time. This paper describes an approach to detecting learners’ stress using a pedagogical agent. For that purpose, a wearable wristband sensor is integrated into an existing pedagogical agent for developing human reasoning ability. The pedagogical agent analyzes the heart rate variability of the learner to determine the individual stress threshold. If the learner’s stress exceeds the critical threshold, the pedagogical agent offers support with stress coping strategies. The evaluation study with the physiology-aware pedagogical agent shows that the heart rate variability in terms of RMSSD (root mean square of successive differences) can be used as a relevant indicator for measuring learners’ stress in real time. The results of the evaluation study suggest deploying RMSSD if stress is taken into account in learning analytics.