ABSTRACT
A common intuition is that instruction is most effective if it takes into account (a) that learners are different, and (b) that they change as they learn. But learners differ in a great many ways-for example, in their knowledge state, interest, goals, affective state, strategic behaviors, and learning styles. To what learner differences should instruction adapt? Also, how can adaptive instruction take into account that learners change continuously? Which ways of adapting are most effective? An alternative view is that learning technologies should not only adapt to learner differences, but also to learner similarities. How can a deep understanding of the demands a task domain makes on all learners be used to adapt system design to common rough spots?
