John A. Nevin, Michael C. Davison, and Michael L. Commons
This volume is based on the 10th annual Harvard Symposium for the Quantitative Analyses of Behavior. The first Harvard Symposium, in 1977, was devoted to signal-detection analyses of reinforcement and choice behavior. Contributions based on the symposium presentations were published in 1981 as the first volume of the Quantitative Analyses of Behavior series. Subsequent symposia and volumes have addressed topics ranging from theories of operant choice behavior to models of neural mechanisms. In this volume we return to the signal-detection theme, with added insights based on 10 years of experimental and theoretical analyses.
The fundamental signal-detection problem requires the subject to detect a signal embedded in a background of noise by responding Yes if the signal is present, and No if it is absent. The probability of presenting a signal may be varied, and explicit consequences -- payoffs, costs, or feedback for humans; food, water, or shock for nonhuman animals -- may be arranged for correct or incorrect choices. In signal-detection experiments of this sort, signal intensity can be varied independently of stimulus presentation probability or the consequences of choice to separate the discriminability of the stimuli from the biasing effects of nonstimulus factors. Related problems require the subject to differentiate between two stimuli that differ in location along some physical continuum such as duration. Again, stimulus values may be varied independently of feedback or reinforcement.
When signal-detection theory was introduced into psychology by Tanner and Swets ( 1954), its principal impact was in sensory psychology, especially on theories of the threshold. Unlike traditional theories that emphasized a discontinuity between observable (supraliminal) and unobservable (sublim-