Leslie Burkett
University of North Texas

Claremont, CA–Researchers at Claremont McKenna College and the Kennedy Krieger Institute are using “task interspersal” to teach autistic children. With this procedure the instructors mix new learning activities with tasks the children have already learned. Then the teachers provide reinforcers such as food treats and praise for correct answers.

When the teachers attempted to teach the children only new tasks there were few correct responses that could be rewarded. But when the new tasks were interspersed with the old ones, the children began to respond correctly to the old tasks and began to receive some rewards.

The five autistic children in the study had been extremely difficult to teach. But task interspersal resulted in new skills being learned quickly, consistently, and with no negative side effects. And the new technique represented only a slight change in teaching procedures.

When the new learning tasks were mixed with the old, the teachers were able to reward correct responses to the new tasks with what behavior analysts term a “rich reinforcement schedule” for every correct answer. But they gave few if any reinforcers for correct answers on the already-learned tasks, or gave only praise. This established a strong contrast between the rewards for performing the new tasks and the nearly non-existent rewards for maintaining the old tasks.

The result was that the children learned more new things. And the researchers were careful to make sure that what had been learned in the past was not lost. Perhaps best of all there were no bad side effects. Surprisingly, researchers Marjorie H. Charlop, Patricia F. Kurtz, and Janice P. Milstein, discovered that less reinforcement (at least for skills already learned) meant more learning.

Charlop, M. H., Kurtz, P. F. & Milstein, J. P. (1992). Too much reinforcement, too little behavior: Assessing task interspersal procedures in conjunction with different reinforcement schedules with autistic children. Journal of Applied Behavior Analysis, 25, 795-808.

Back to the table of contents for this issue

Return Home