Mental Workload Analysis Associated with Emotional Design in E-learning Contexts: Combining EEG and Eye-Tracking Data
In this study, we designed a mental workload induction experiment in the context of online learning, in which EEG and eye-tracking data of participants were synchronously recorded with the aim of investigating the association between different design principles and multimodal physiological features and then applying machine-learning technology to classify mental workload states induced by those principles. This paper systematically reviews three kinds of EEG and eye-tracking features used for mental workload classification, compares the accuracy of mental workload classification between single-modal and multimodal features, modifies the mental workload index proposed by Pope et al. to monitor the variation of mental workload in E-learning contexts, and reduces the dimensions of features for more convenient use in daily life. The results of the experiment demonstrate that (1) The classification ability of wavelet power features and eye-tracking features are better than that of entropy features in E-learning contexts; (2) Multimodal physiological data can significantly improve the accuracy of mental workload classification in E-learning contexts; and (3) Correlation-based feature selection (CFS) was employed to rank all features in descending order, and when the feature dimension is reduced to 30, the optimal average classification accuracy obtained by linear-SVM is 80.2%. Furthermore, the EEG frequency bands that are highly correlated with mental workload were analyzed, and the correlation between different brain areas and mental workload discussed. All these results lay the foundation for continuous monitoring of participants’ mental workload, making it possible to endow computers with the ability to understand mental workload in E-learning contexts, which will in turn remarkably enhance participants’ learning efficiency and performance during the pandemic, and in other circumstances necessitating online learning.