ABSTRACT
The law of effect has been largely concerned with the effects of reinforcer contingencies on behavior under highly discriminable conditions. There has been little regard paid to the role of controlling stimuli in matching research (but see Logue, 1983; Miller, Saunders, & Bourland, 1980, Section 8.7; White, Pipe, & McLean, 1984; Section 9.6). Yet, failures to obtain the expected responsereinforcer relation have often been ascribed to failures of the stimulus-response relation (e.g., Baum, 1974a). In an attempt to measure stimulus effects in schedule-control experiments, one common research practice has been to apply, cookbook fashion, detection-theory statistics to data obtained from operantconditioning experiments (Wright, 1982). Such approaches (e.g., Logue, 1983; Stubbs, 1980) are not of concern here. Rather, our interest lies solely with attempts to integrate quantitatively the matching law with the theory of signal detection (see Davison & McCarthy, 1980; Davison & Tustin, 1978; Nevin, 1981; Nevin, Jenkins, Whittaker, & Yarensky, 1977, 1982). It is the goal of behavioral detection theory to include the effects of discriminative stimuli in the law of effect and thus to provide a unitary model of stimulus and reinforcer control in a variety of discrete-trials detection paradigms and free-operant conditioning procedures. (Usually, in a trials procedure, a subject is allowed to emit only one response at some point in a series of environmental events-this series constituting the trial. In a free-operant procedure, many responses may be emitted in the presence of one or more environmental events.)
11.2. THE STANDARD YES-NO DETECTION EXPERIMENT
With minor variations, the signal-detection task is a discrete-trials procedure in which stimuli are presented to a subject trained to report which of two or more stimuli were presented. With human subjects, typically two stimulus classes (signal superimposed on noise, and noise alone), and two response classes (“ Yes, there was a signal” and “ No, there was no signal” ) are arranged. Thus, there are four stimulus-response events: Hits (correctly reporting the signal as present), correct rejections (correctly reporting the stimulus as absent), false alarms (reporting the signal as present when it was absent), and misses (reporting the signal as absent when it was present). Hits and correct rejections typically lead to reinforcers, and false alarms and misses lead to penalties. It is the explicit use of these reinforcement and punishment contingencies that provides direct contact with analyses of choice behavior (Nevin, 1969).
