Transforming a Large-Enrollment Neuroscience Course With Active Learning: Using AI-Based Text Analysis to Capture...

Large-enrollment undergraduate neuroscience courses often rely on traditional lecture formats that limit student engagement with complex material. This study describes the redesign of a large-enrollment neuroscience course into a hybrid active learning format combining asynchronous online lectures with weekly small-group in-person sessions. To evaluate the impact of this redesign, we analyzed 547 open-ended student comments from 232 students across two semesters (one traditional lecture, one active learning) using zero-shot classification (an approach that uses a pre-trained language model to categorize text into researcher-defined themes), via an open-source large language model (BART), and compared these findings with Likert-scale questions. The active learning format produced significant improvements in students' perceived learning experience and engagement, gains that were not visible in the Likert-scale data. This discrepancy highlights the limitations of Likert-scale evaluation tools for capturing the nuanced benefits of pedagogical innovations in large classrooms. We provide a practical framework neuroscience educators can adopt to redesign large courses and assess their impact using accessible AI tools, without requiring specialized technical expertise.