Computational models of human skill can impart the necessary sense of realism to the actions and behaviors of virtual humans in the virtual world. Accurate models of human skill can contribute to improved expert training and human-computer interfacing. Models of human skill find application in far-ranging fields, from autonomous robot control and teleoperation to human-robot coordination and human-robot system simulation. Interest in modeling human control goes all the way back to World War II, when engineers and psychologists attempted to improve the performance of pilots, gunners, and bombardiers. Robot learning from human experts has also been applied to a deburring robot. Interest and research in neural network-based learning for control has exploded in recent years. Research into feedforward neural networks began in earnest with the publication of the backpropagation algorithm in 1986. Neural networks have received great attention for nonlinear learning and control applications.