Punishment a primary process?

Lerman and Vorndran have undertaken a heroic task. They have reviewed the human and nonhuman basic behavior-analytic research and the applied behavior-analytic research on the effects of punishment and then made recommendations for the use of punishment to reduce problem behavior. They demonstrate that the effects of punishment vary as a function of other prevailing conditions, such as past and concurrent reinforcement contingencies. Their review is extensive, thoughtful, and well balanced. Most of the recommendations that they make for treatment seem sound. The article provides an excellent review of the literature and makes many thoughtful recommendations for treatment. However, as any discussion article should, it raised a number of issues. The first issue relates to definition as presented in their first sentence: ‘‘Punishment is generally defined as an environmental change contingent on behavior that produces a decrease in responding over time (Michael, 1993)’’ (p. 431). That definition is elegant. It is short, direct, and parallels the definition of reinforcement. However, there is also Skinner’s (1953) less elegant and somewhat speculative conception of the effects of punishment as secondary effects. Skinner’s conception states that there are three ways in which a stimulus can reduce the rate of a response. First, the stimulus may elicit responses that are incompatible with the punished response. Second, the stimulus may result in conditioned emotional responses that are incompatible with the response (conditioned suppression). Third, operant responses that are incompatible with the punished response