Poster E2, Monday, March 26, 2:30-4:30 pm, Exhibit Hall C
A neural score for engineering concepts: predicting STEM learning with multivariate pattern analysis of functional neuroimaging data
Joshua S. Cetron1, Andrew C. Connolly2, Solomon G. Diamond3, Vicki V. May3, James V. Haxby1, David J. M. Kraemer1; 1Dartmouth College, 2Geisel School of Medicine at Dartmouth, 3Thayer School of Engineering at Dartmouth
Learning in any conceptual domain can be described along a progression from naïve understanding to expertise. Behavioral tasks and tests of concept knowledge generate scores that can be used to assess the degree to which an individual learner has acquired expert-level understanding of a concept. In previous research on mechanical engineering concepts, we have shown that expert-level information is also represented in patterns of neural activity exhibited by participants over the course of a concept learning task. In the present study, we investigated whether those patterns of neural activity can be used to compute a “neural score” for an individual learner that would complement existing, behaviorally-derived scores of that individual’s engineering concept understanding. Using representational similarity analysis and an expert model of mechanical engineering information, we successfully derived a neural score from patterns of activity across the brain reflective of expert-level knowledge. This score correlates with two independent behavioral scores of concept learning. Further, neural scores differentiated individuals across two groups of participants with different levels of prior experience with the learned concepts. More broadly, this neural scoring method could be applied to predict learning outcomes in additional content domains and over longitudinal neuroimaging studies of learning, most notably those involving classroom-based research on education.
Topic Area: THINKING: Reasoning