ABSTRACT
Recent years have seen a proliferation of different views, explanations, and quantitative accounts of choice (e.g., Baum, 1981; Herrnstein, 1970; Herrnstein & Vaughan, 1980; Heyman & Luce, 1979; Lea, 1981; Rachlin, Battalio, Kagel, & Green, 1981; Shimp, 1979; Staddon & Motheral, 1978). Within this context, a controversy of sorts has arisen between molar and molecular accounts of choice. The molar-molecular issue consists of several related sets of issues (see, for example, the chapters of Volume 2 of the present series); but a basic problem, which relates to these issues, concerns the appropriate way to characterize the data in choice experiments. The molar position considers the relatively long-term interaction between behavior and environment as basic. The well-known matching law provides an excellent example of the molar position. It states that the relative output of behavior equals or matches the relative frequency of reinforcement for that behavior. The data used for the matching relation are the numbers of responses and reinforcers that are averaged together over a relatively long period, typically the entire session. The advantages of the molar matching law are that it provides a simple but elegant relation between behavior and its consequences and that it applies to a wide variety of experiments that encompass different operants, consequences, and procedural modifications (e.g., de Villiers, 1977; Herrnstein, 1970). The disadvantage of the molar analysis is that the long-term averaging of numbers of responses and of reinforcers necessarily obscures local changes in behavior and the local effects of different patterns of reinforcers. Molecular positions place a greater emphasis on these local changes with the result that molecular analyses deemphasize average response rates and, instead, emphasize sequences of choices. One group of molecular positions emphasize the structure of behavior, taking the position that response patterns may be formed and act as functional units of behavior; responding is viewed as being related to the prior sequence of choices. The appeal of the molecular analysis is that the data analysis is more detailed than the molar, and that the molecular results may provide an explanation for matching or show that matching is the outcome of a more basic, molecular process (e.g., Shimp, 1969).
