International Journal of
Information and Education Technology

Abstract

There is increasing evidence that learners’ affective and cognitive states play a key role in the learning process. This suggests that systems which are able to detect these states can dynamically use adapted strategies to increase the pace of the learners’ skill acquisition and improve their learning experience. In this work, we present a novel approach for automatically adapting the learning strategy in real-time according to the learner’s detected mental state. The main goal of the approach is to maintain the learner in a positive state during a lesson by adaptively selecting the best interaction strategy between either using problem solving or worked examples. Two mental indexes, namely, cognitive load and mental engagement were extracted from electroencephalogram (EEG) signals, and used to adapt the system’s interaction. The cognitive load index was developped by training and validating a prediction model on various types of memory and logical tasks. The engagement index was directly computed from the EEG signal frequency bands. An experiment with 14 learners was performed in order to evaluate this approach. The obtained results showed that using the learner’s mental state to adapt the system’s interaction has a positive impact on the learning outcomes, the learning experience and the learners’ reported emotional states.