The diagnostic evaluation of a learner's state in adaptive and intelligent tutoring sessions poses a number of difficult issues. We discuss several reasons for focussing more on process-oriented metrics of student learning. In particular we are developing methods of using mistake and hint trends to track the efficiency of intelligent tutoring. We argue that the goal of efficient learning is strongly linked to the goal of maintaining learning in the zone of proximal development. We illustrate our use an operational definition of the zone of proximal development in the design of student modelling and in the formative evaluation on the Animalwatch mathematics ITS.
展开▼