| Abstract: According to the standard concept, operants are response classes modifiable by consequences. In this talk, I contend that this concept is incoherent. Modifiability by consequences refers to a kind of behavior change in time. Viewing response classes as modifiable by consequences thus implies viewing them as dynamic. However, interpreting classes as sets and using set theory as a tool for formal analysis raises the following problem: For a response class to be dynamic, we must assume that its identity across time is determined by its defining properties, and that the number of its instances changes. This assumption contradicts the Axiom of Extensionality (AE), one of the foundations of set theory. According to AE, the identity of a set is determined by its instances, not its defining properties, and that the number of its instances changes. Under AE, then, changes in response frequency across time determine different sets, not a single set. Two solutions to this problem are possible. First, AE could be rejected, but this would eliminate set theory and, hence, theorems on which quantitative analyses and modeling in empirical science (behavior analysis included) are founded. A less problematic solution would be to accept AE, abandon the concept of an operant as a response class, and propose a different concept. As an alternative concept, I propose that an operant is a class of functional relations that describe a kind of behavior change. |