Reinforcement and Punishment in Behavioral Models of Signal Detection

Abstract

Behavioral models of signal detection have focused almost exclusively on the effects of reinforcement for correct choices. In contrast, the effects of punishment for errors have been largely ignored. Two competing models of punishment can be derived from research using simple concurrent-schedule procedures. Subtractive models predict that punishers directly subtract from the effects of reinforcers for the same response alternative, and additive models predict that the effects of punishers add onto the effects of reinforcers obtained for the other response alternative. These two models were incorporated into Davison and Tustin’s (1978) model of signal-detection performance. Some preliminary research using human subjects in a signal-detection procedure provides support for an additive punishment version of the Davison and Tustin model.