Brain Networks Underlying Sex and Pedagogy Differences in Physics Learning

Bartley JE, Riedel MC, Salo T, Bottenhorn KL, Bravo E, Odean R, Nazareth A, Laird R, Sutherland MT, Pruden S, Brewe E, Laird AR, HBM (2019).



Physics is a challenging academic pursuit wherein students regularly struggle to achieve success. Discouragingly, female students tend to underperform relative to their male peers on introductory physics conceptual inventories (1). Active engagement teaching is known to improve performance on these tests relative to lecture instruction methods (LI; 2), yet the neurocognitive networks linked to physics learning, pedagogy, and sex differences have not been evaluated. Modeling Instruction (MI) is one such active engagement approach that emphasizes experimentation and visualization, and supports learning in both males and females (3). Here, we used fMRI to delineate physics reasoning and retrieval-related brain networks in introductory physics students and probed for class, sex, and time-related differences resulting from a semester of MI or LI physics instruction.


FMRI data were acquired from 107 right-handed participants (55 MI, 52 LI; 48 female, 59 male; age 18-25) on a 3T GE 750W scanner. Participants were undergraduates at Florida International University and first-time enrollees in college-level LI or MI physics courses. Students completed one beginning-of-semester fMRI session (pre-instruction) and a second identical end-of-semester session (post-instruction) where they completed the Force Concept Inventory (FCI; 4) physics reasoning task, a physics and general knowledge retrieval (RETR) task, and a content-general transitive inference (TI) task. Analyses were performed in FSL and activation maps were generated and thresholded at P<0.05 (clusters z>2.3). Maps were computed by session individually, then class, sex, and longitudinal changes were assessed using a three-way fixed effects ANOVA to identify regions more engaged during task within one group relative to another, at the end relative to the beginning of the semester, and to test for significant interactions between class, sex, and time.


Pre- and post-instruction revealed similar activity in a fronto-parietal network (FPN) for FCI, RETR, and TI tasks (Fig. 1a). Behavioral results indicated students scored better post-instruction on FCI and RETR, but not on TI questions. We found learning-related changes in female and male LI and MI students within a mixed FPN and default mode network (DMN) associated with both physics tasks, but not with the TI task (Fig. 1b). Increased physics-related activity was observed at both time points in a FPN in male students, and in DMN and visual areas in female students (Fig. 2a). Significant interactions (class x sex x time) were observed during RETR in lingual and fusiform gyri, suggesting male and female students may engage brain areas differently during physics thinking, depending on whether they completed an LI or MI class (Fig. 2b).


Pre- and post-instruction fMRI data from physics students revealed similar FPNs underlying physics thinking and transitive inference, consistent with known problem solving-related brain function (5). Increased FPN-DMN activity, accompanied by improved physics scores, following instruction suggest this network’s key role in physics learning, indicating potential reliance on memory-linked mental exploration during answer making (6). Sex differences in the neural systems supporting physics thinking may indicate dissociabilities in how students answer questions (7). However, brain function varied in sex-specific ways by class in visual regions, suggesting sex differences may be mitigated via increased emphasis on visualization and use of diagrams and pictures during reasoning, as well as enhanced incorporation of classroom demonstrations to promote physics understanding. To date, this is the first study to characterize how real-world classroom learning drives functional reorganization of large-scale brain networks among physics students. Our results indicate potential sex and pedagogy differences underlying the neural mechanisms supporting physics learning.

This work was supported by NSF REAL Grant DRL-1420627.