Much work in optimal control and inverse control has assumed that the controller has perfect knowledge of plant dynamics.  However, if the controller is a human or animal subject, the subject's internal dynamics model may differ from the true plant dynamics.  We developed a probabilistic framework to estimate the subject's internal model from control demonstration.  We then applied the framework to brain-machine interface control by a non-human primate.  We found that the mismatch between the subject’s internal model and the actual plant dynamics explained the majority of the subject's control errors.  This result suggests that accounting for the subject's internal model is essential for understanding control systems involving humans or animals.