To test the hypothesis that a discrepancy between expectation and experience with regard to the controllability of outcomes would produce the greatest effect on later task behavior, 70 undergraduate volunteers were directed individually in a multiphase experiment. The first phase was designed to induce expectations of control or lack of control over outcomes. The second phase was designed to confirm or not to confirm the expectations induced. The third phase tested for the effects on later task behavior. The results indicated that the first phase procedure failed to induce the required expectancies, thus preventing a test of the experimental hypothesis. Possible procedural changes were discussed.