Proceedings of the Annual Meeting of the Cognitive Science Society

In this study, we investigate how high- and low-performance learners (N=12) act differently while using a cognitive tutoring system. We examine three research questions: (1) Can we predict learners’ performance using only their visual attention (eye movement data)? (2) Can we predict learners’ performance from visual attention data and initial performance? (3) Are age, gender, first language, where they look, and the sequence of Areas of Interests (AOIs) significant factors in the learners’ performance? Learners more correctly answer questions taken from larger rather than smaller AOIs. Our results show that high-performance learners pay more attention to the content that contains answers to later questions. Surprisingly, the tutor did not change the learners’ visual search to a goal-oriented search. Our analyses can help instructional designers create a more productive learning experience because visual search behavior as part of a learner model with acceptable accuracy in early stages can be used in adaptive tutors. Additionally, we trained a classifier on the eye movement data to predict learners’ performance for each question. Its results provide a list of suggestions for designing more productive learning experiences, such as enticing user attention by increasing the size of the content that contains answers and changing the order of contents.