Role of relativistic knowledge in intelligent tutoring

摘要

Understanding complex cognitive behavior necessarily involves interaction between two (or more) individuals. The teaching-learning process is just a special case. From the standpoint of an intelligent tutor (or human teacher) all student/learner knowledge is either assumed to be available or is judged relative to the teacher's model(s) of the content and/or skills to be acquired. In this regard, learner models play a central role in most contemporary approaches to intelligent tutoring. These models typically assign to the learner certain capabilities which are assumed by the intelligent tutor in administering instruction. If too much is assumed, however, it is impossible to determine where the student is going wrong, or how to correct the problem—unless the tutor model is enhanced to incorporate the assumed capabilities and these capabilities correspondingly are eliminated from the learner model. Research associated with the Structural Learning Theory shows that a common “goal switching” control mechanism is all that may safely be assumed with confidence. Maximum flexibility is achieved by assuming only this mechanism, along with basic encoding and decoding capabilities.