Metacognition refers to the ability to monitor and regulate one’s own cognitive processes. A key aspect of metacognition is metacognitive calibration, which involves the accuracy with which individuals assess their own knowledge for errors. Despite the recognized importance of metacognition in STEM education, longitudinal studies exploring the stability of metacognitive monitoring and error detection are conspicuously scarce. While undergoing fMRI, life science students evaluated a series of biology models for errors, rating their confidence after each trial. A year later, most of these students returned to complete another biology error detection task and an interview. Findings suggest that students with higher self-monitoring skills may more effectively deploy lateral prefrontal brain regions when encountering errors or conflict, ultimately supporting their longer-term concept retention. This longitudinal study enabled us to draw accurate inferences about students’ neural and behavioral performances, shedding light on the metacognitive skills needed to evaluate biology models and the stability of students’ metacognitive monitoring over time. These results underscore the potential effectiveness of instructional strategies that focus on improving metacognitive monitoring among STEM learners.