Two assumptions in support of learner-controlled computer-assisted instruction (CAI) are that (1) instruction administered under learner control will be less aversive than if administered under program control, and (2) the student is sufficiently aware of his learning state to make, in most instances, his own instructional decisions. Some 130 college student subjects were exposed to a CAI module on 12 edible plants native to Texas. The program for one group included pictures of the plants to help in their identification. A second group was never shown the pictures, and the third group had the option of choosing whether or not they wanted to see the pictures. It was anticipated that those subjects having an option would almost always elect to see the pictures. To provide for investigating individual differences in the use of learner control, a second, less appealing option was introduced--a review of the plants' critical features. It was concluded that learner control over a facilitating treatment did not reduce state of anxiety; the student's ability to use learner control effectively appeared to be a function of personality traits as well as cognitive skills. This suggests that state of anxiety tends to be task specific, and that the relationship of tasks to anxiety should be explored in subsequent research. (Author/DGC)