As the field of Mind, Brain, and Education seeks new ways to credibly bridge the gap between neuroscience, the cognitive sciences, and education, various connections are being developed and tested. In this article, we present a framework and offers examples of one approach, predictive modeling within a virtual educational system that can include representations from the neural level to the policy level. Researchers could calibrate, test, and question the model, potentially providing quicker, more efficient, and more responsible ways of making advances in the developing educational field. Likewise, virtual investigations using models with this sort of capability can supplement the valuable information derived from carrying out policy and instructional experiments in real educational contexts.