Abstract
There is an ambiguity in formulations of the Law of Effect which stress the importance of the correlation of rate of responding with frequency of reinforcement. The main problem is that such theories have not specified precisely how the correlation of response and reinforcement rates should be determined, with the result that the theories can become irrefutable. Two experiments were carried out which expose some of the problems created by this ambiguity. “Free” food reinforcers were delivered to rats in the absence of responding. Lever responses intermittently provided immediate (contiguous) reinforcement, but cancelled some of the following “free” reinforcements. Responding was established and maintained even when the overall rate of responding was negatively correlated with the overall frequency of reinforcement. Several ways in which correlational theories could attempt to accommodate these results are discussed but rejected as unsatisfactory, either because they severely limit the scope of the theories or because they lose their most important feature: that of treating behaviour at a molar rather than a molecular level.