Analysis Of Learning Confidence Using Multimodal Neurophysiological Data

The advent of neurotechnology has opened new avenues for enhancing educational experiences by providing personalized feedback based on neural and behavioral data. Electroencephalography (EEG) is particularly promising for real-time monitoring of cognitive states, such as learning confidence, which can be pivotal for personalizing educational interventions. This paper aims to help learners recognize their cognitive states by providing real-time personalized feedback. We conducted an experiment involving 27 subjects, aged 18 to 40, who learned a foreign language online. Throughout the learning sessions, data were collected using EEG, eye-tracking, and mouse-tracking devices. EEG data were preprocessed to extract power spectral densities, mainly focusing on the alpha and beta frequency bands. Eye-tracker and mouse tracker data were analyzed to extract features such as fixation duration, saccade length, and interaction patterns. These multimodal data were used to train an ML model to predict learning confidence. Analysis revealed that the Pz location of the EEG sensor showed a significant difference (p-value $\lt 0.05$) in EEG power within the alpha and beta bands approximately 800 milliseconds before learners submitted their answers, correlating with their correct or incorrect answers. Eye-tracker and mouse tracker data provided additional insights into the learners’ engagement and decision-making processes. Our findings demonstrate the feasibility of using multimodal physiological data to model the accuracy of answers in real time. Upon success of this research, we would be able to provide learners with immediate feedback to ensure that all learning materials are comprehensively understood, leading to enhanced learning quality.