Integrating Non-Invasive Neuroimaging and Educational Data Mining to Improve Understanding of Robust Learning Processes
The overall goal of our project is to improve understanding of student cognitive states by drawing a parallel between the cognitive neuroscience constructs of rule learning and goal maintenance, and the learning sciences constructs of induction and refinement, fluency, and mind-wandering. Traditional approaches using interaction logs to model learner knowledge and behaviors may be insufficient in scenarios such as when learners pause during their interactions with the technology and there are no actions to be logged. Many important cognitive states that can be associated with learning still continue to occur during these pauses and detecting them is crucial for a continuous estimation of learner's needs and adapting the technology accordingly. Moreover, many of the cognitive states that are interest of learner modeling and educational data mining research have been studied at a very-high level compared to how their underlying constructs are studied within cognitive science research and the connection between these high-level cognitive states and their lower-level underlying mechanisms are not very well understood. There are three major aims: (1) Integrate multiple data streams, including behavioral and neural data from controlled cognitive tasks and neural data from complex learning activities, to create an interdisciplinary corpus, (2) Detect real-time changes in cognitive states during pauses in log data and connect this information to brain data, and (3) predict learning outcomes from brain-based and log-based inference of cognitive states.
This NSF funded project is a collaboration with Dr. Erin Solovey at Worcester Polytechnic Institute and Dr. Kate Arrington at Lehigh University. Within the lab, Deniz Sonmez-Unal works on the project.
Publications
Howell-Munson, A., Unal, D. S., Walker, E., Arrington, C., & Solovey, E. (2021, June). Preliminary steps towards detection of proactive and reactive control states during learning with fNIRS brain signals. In Proceedings of the First International Workshop on Multimodal Artificial Intelligence in Education (MAIED 2021) (Vol. 2902).
Unal, D. S., Arrington, C. M., Solovey, E., & Walker, E. (2020). Using thinkalouds to understand rule learning and cognitive control mechanisms within an intelligent tutoring system. In Artificial Intelligence in Education: 21st International Conference, AIED 2020, Ifrane, Morocco, July 6–10, 2020, Proceedings, Part I 21 (pp. 500-511). Springer International Publishing.