Contingencies in Avoidance Learning in Goldfish

Abstract

Two experiments were performed to examine how the “avoidance” response of goldfish in a shuttlebox is controlled by two kinds of contingencies, namely stimulus-reinforcer contingency and response-reinforcer contingency.In Experiment I, two groups were trained under avoidance conditions with or without CS termination by response. The occurrence of CSs and shocks for subjects of two yoked control groups were determined by the responses of subjects of avoidance groups. In the first phase of experiment, response rates of avoidance groups were higher than those of yoked groups (Fig. 1). It seems to show that response-reinforcer contingency is more effective than stimulus-reinforcer contingency. But in the second phase of experiment, where they were tested in classical conditioning situations, responses of all groups were maintained at a high level. It supports the notion that “avoidance” response of goldfish is controlled by stimulus-reinforcer contingency.In Experiment II, two avoidance groups and two yoked control groups were tested in punishment situations. Response rate of an avoidance group (A4) and its yoked group (Y4) differed significantly after the shifts to punishment situations (Fig. 2). It indicates that avoidance response is acquired partly by response-reinforcer contingency. But as response rate of avoidance group (A4) did not go down to zero, responses should also be controlled by stimulus-reinforcer contingency. Response rate of another avoidance group (A3) did not change significantly after the shift. It might be due to the CS termination condition common to both phase 1 and phase 2.Thus, “avoidance” response of goldfish contains both respondent and operant components. It is thought to be comparable to autoshaped key pecking response in pigeons or leg flexion in dogs. Future directions of research to distinguish two kinds of components are discussed.