ABSTRACT
In operant psychology there has been a trend toward quantitative analysis. In the first years of this discipline (e.g., Skinner, 1938), research reports were typically organized around a graphic representation of the experimental results. These graphs, called cumulative recorder tracings, showed a moment-to-moment account of the reinforced response: When the subject, such as a rat, pressed a lever, a pen was stepped along a continuously moving roll of paper. Thus, a train of responses would show up as a smooth line, and the faster the rate of responding, the steeper the slope of the line. Recent operant research papers, however, rarely include cumulative recorder tracings (see Skinner, 1976, for a eulogy). Now it is more likely for such papers to be based on a mathematical model of the experimental conditions. The models are derived from theories, for example, the assumption that subjects in operant experiments maximize some dimension of reinforcement (e.g., Rachlin, 1980), and the goal of the research is to fit the model, and thereby test the theory. In effect, vignettes, in which the subject's behavior was played back just as it occurred, (e.g., Ferster & Skinner, 1957), have been replaced by calculations and goodness-of-fit tests.
