Behavioural contrast in discrete-trial discriminations: effects of non-reinforcement

Six pigeons performed an acquisition and three reversals of a discrete trial simultaneous discrimination in which responses to the positive key were reinforced on a fixed-interval 5-s schedule; trials were terminated by either a response to the negative key or the delivery of a reinforcement. In the initial stage of acquisition and each reversal, where errors were most frequent, response rates rose and latencies fell on positive trials; these effects declined as errors decreased. The birds were also run in two yoked control conditions, and results showed that the critical requirement for the positive trial rate and latency changes was neither the simultaneous presentation of two stimuli nor the formation of a discrimination, but the occurrence of non-reinforced responses. The similarity of these phenomena to conventional behavioural contrast is emphasized, and the results are discussed in terms of frustration theory.