Concurrent performances: Extinction, noncontingent reinforcement andvariably delayed reinforcement

Abstract

Transitions from variable-interval (VI) reinforcement of a pigeon’s pecks on one key to extinction, noncontingent reinforcement or variably delayed reinforcement of those pecks were arranged with and without concurrent reinforcement of pecks on a second key. Within sessions, the two keys were sometimes available singly and sometimes concurrently. Response rates during extinction and during noncontingent reinforcement decreased comparably. Similar rate decrements occurred early after initial exposure to variably delayed reinforcement; thereafter, delayed reinforcers maintained relatively higher rates of responding that produced little if any loss of scheduled reinforcers. Extinction, noncontingent reinforcement and delayed reinforcement typically maintained lower rates of responding when operating concurrently with VI reinforcement than when operating alone. Responding was maintained by the contingent relation between responses and reinforcers whether that contingency involved immediate or variably delayed reinforcers. Responding was not maintained when that contingency was broken whether reinforcers were completely discontinued or were delivered independently of responses (noncontingently). These results add to the data supporting a distinction between the effects of initiating or terminating response-reinforcer contingencies and those of initiating or terminating reinforcer deliveries. When responding appears to be maintained by noncontingent reinforcement, in so-called superstition or adventitious reinforcement, it is likely either that the maintenance will be transient or that other variables are operating, as when reinforcer deliveries have eliciting or adjunctive effects.