Laboratory evaluations of reinforcement contingencies.

Abstract

In a previous issue of this journal, we discussed the logic of conducting basic research on problems related to application (Borrero, Vollmer, Samaha, Sloman, & Francisco, 2007). For example, it is common in some fields of basic science (e.g., genetics) to take a real world problem (e.g., cancer) and conduct laboratory research to better understand the causes of real world problems (e.g., genetic predispositions) and potential solutions to those problems. Although the field of behavior analysis has a long history of translating basic research into application, it is less common in our field to take questions about the human situation and make efforts to address them via laboratory research and then, in turn, use the findings to promote better application of behavior analysis. Such an approach has been the focus of a decade-long effort to conduct research in an operant (rat) laboratory for the purposes of addressing questions pertinent to the behavioral assessment and treatment of behavior disorders. We have empirical papers in various stages of preparation and thought it might be useful to provide an update on some findings from two specific lines of research. The first line of research is a series of studies constituting the dissertation experiments by the second author, which involved evaluations of reinforcement contingencies as a comparison of two conditional probabilities (a probability of a reinforcer given the occurrence of behavior and a probability of a reinforcer given the nonoccurrence of behavior). The second line of research is a series of studies constituting the dissertation experiments by the third author, which involved testing a variation of differential reinforcement of other behavior (DRO) known as momentary DRO (mDRO). On the notion of contingency Reynolds (1975) distinguished between contingency and dependency. Dependencies are relationships that describe essentially if-then statements. For example, flipping a light switch causes a light to turn on because of the circuit and the light will come on if and only if the switch is flipped. If a computer is programmed to deliver reinforcers on a fixed ratio schedule, pressing a lever in an operant chamber causes a pellet to be dispensed if and only if the lever is pressed. Of course, a dependency does not require that the consequent event occur every time the response occurs; for example, lean ratio schedules and intermittent interval schedules are still examples of dependency schedules because the reinforcer is delivered if and only if behavior occurs. Although dependencies are certainly a type of contingency, a contingency can more broadly include relations that are obtained from (a) dependencies, (b) accidental relations (Reynolds, 1975), or (c) blends of events including when there is some probability of an event given behavior and some probability of the event given no behavior (Catania, 1998). As an example of the first type (dependencies) relating to behavior disorders, consider the possibility that a severely disabled student in a wheel chair is unable to speak. The only time he receives attention from his teacher is when he begins banging his head against the back of his wheelchair. In short, he obtains teacher attention (positive reinforcement) if and only if the self-injurious behavior (SIB) occurs. This type of contingency is commonly used to test sensitivity to reinforcement during a functional analysis (e.g., Iwata, Dorsey, Slifer, Bauman, & Richman, 1982/1994), where the test conditions involve response-stimulus dependencies insofar as the putative reinforcing event occurs if and only if the target behavior occurs. As an example of the second type of contingency (accidental relations), suppose a teacher is told to give a student some magazines (or other preferred items) once every 5 min during her break. Further suppose that the student becomes aggressive toward the teacher immediately prior to the delivery of the magazines. …