Gravar-mail: Structural learning in feedforward and feedback control