The effects of response mode and complexity level of instructional materials, and the effects of student ability (as indicated by grade point average) on learning performance were studied. Four programmed instructional units on an unfamiliar subject, physiological psychology, were presented to 159 college students. Three of the units required the learners to construct fill-in-the-blank responses to questions of varying complexity; the programmed materials were categorized according to low, medium, or high error rate. The fourth set of materials required reading only. Learning was assessed with a 30-item multiple choice pretest and post test. High ability subjects performed better than low ability subjects, regardless of treatment. Curve-fitting regression techniques showed that different ability groups have different learning curves when plotted against the error rate of the materials. These trends are consonant with achievement motivation theory predictions. The reading-only mode was found to be superior to the constructed response; this result may be explained by the familiarizing effect of the pretest. Related studies are briefly discussed. (GDC)