Virtual reality presents exciting new prospects for the delivery of educational materials to students. By combining this technology with biological sensors, a student in a virtual educational environment can be monitored for physiological markers of engagement or more cognitive states of learning. With this information, the virtual reality environment can be adaptively altered to reflect the student’s state, essentially creating a closed-loop feedback system. This paper explores these concepts, and presents preliminary data on a combined EEG-VR working memory experiment as a first step toward a broader implementation of an intelligent adaptive learning system. This first-pass neural time-series and oscillatory data suggest that while an EEG-based neurofeedback system is feasible, more work on removing artifacts and identifying relevant and important features will lead to higher prediction accuracy.