A logical analysis of operant learning is presented. In total, the analysis makes a number of predictions that are different from the predictions of any other theory. Individual predictions can be explained by other theories, but the pattern of predictions is unique. Some tests of the predictions of the analysis with human newborns are described. The analysis predicts increased variance in sucking with the introduction of continuous reinforcement. This does occur. The analysis predicts a decreased rate of sucking with a shift from continuous to partial reinforcement. This does occur. The analysis predicts an increased rate of sucking with a shift from continuous reinforcement to continuous plus noncontingent reinforcement. Due to methodological deficiencies, we have been unable to test this prediction. However, it has been confirmed by others. The most exciting prediction of the analysis is a rapid way of producing extinction. That has not been tested with newborns; however, there is confirmatory evidence in the literature.