Undergraduates in two experiments learned ten mathematical rules under different strategies of adapting instruction to individual needs. An adaptive model developed by Hansen was structured to use multiple predictors for two purposes: (1) to classify students into four independent learning style groups, thus allowing for group-based adaptation; and (2) to predict individual performance, thus allowing for instructional adaptation within groups. Results indicated that when the incentives attached to rules were adapted to individual subjects' predicted rule scores (high prediction = low incentives, and vice versa), retention was facilitated relative to treatments which varied incentives in a non-adaptive manner or held them constant across all rules. A more finely-graded incentive distribution (five different levels) was more advantageous than a less finely-graded one (three different levels). Additional findings showed that the benefits of differential incentives increased when the strategy was used in combination with individualized adaptation of the quantity of support material prescribed. These strategies are valuable because they allow students the freedom to self-manage, while arranging incentives and practice opportunities so as to promote retention. (Author/CP)