The hypothesis was tested that highly anxious students invited to employ reflective learning strategies would spontaneously use a more reflective, less impulsive approach in solving response-oriented problems. A total of 53 fourth-grade students were administered the Child Anxiety Scale: 12 students were assigned to a high-anxiety treatment group, while 16 served as low-anxiety controls. Anxiety was used as an indicator of the independent variable of impulsivity. The dependent variable, involving both response behavior and learning behavior measures, was determined on a pretest/treatment/posttest basis. Treatment basically consisted of (1) obtaining a baseline measure of the number of stimulus items subjects could recognize, extract from a random array, and order to match a previously seen set; (2) suggesting to subjects a labeling procedure that might improve performance; and (3) continuing the first procedure from baseline until progress could no longer be made. Importantly, no explicit strategy was outlined; subjects chose their own labeling strategy as well as the degree to which they utilized strategy. A measure of learning was taken during treatment administration. Main findings indicated that the higher order strategy adopted by both control and treatment groups resulted in more reflective behavior in learning tasks and also transferred to subjects' response behavior. (RH)