ABSTRACT

It seems probable that the empirical study of behavior allocation will continue, on the lines discussed here, for some time because many potential areas of research have as yet received only scant attention. The area, however, is due for reorganization. It was the major paradigm shift, formalized by Hermstein (1970), that moved the study of behavior away from a consideration of absolute behavior toward the study of behavior in context. The new paradigm commenced with a series of empirical results (strict matching) that, when set in the context of prior learning theories, appeared simple and elegant. At that time, learning theories had become complex and diverse (see, for example, Volume 2 of Koch, 1959). Two factors ensured the acceptance of the new approach: at one level, the scientific reinforcer of parsimony; and at another level, the more prosaic reinforcers of simply being able to comprehend and teach the new approach. Strict matching ruled. But like any other new, clean, elegant paradigm, it has become complicated and diverse again. We are now faced with the generalized matching law. We are concerned about nonunit sensitivity and the environmental conditions that produce this, and we are troubled by the possibility that we might have to generalize to dual-sensitivity models (Davison, 1982; Prelec, 1984), and we have begun to wonder whether all log independent-variable ratios produce linear changes in log behavior ratios (e.g., delay of reinforcers and amount of reinforcers). The context of our research is now driving us toward another paradigm shift.