Aftereffects of surprising nonreward on operant performance of rats and pigeons

Abstract

The frustration effect refers to greater response strength immediately after nonreinforcement (N) than after reinforcement (R). The frustration effect has been traditionally interpreted either as the result of transient invigoration of behavior after N induced by an emotional state, or transient suppression of behavior after R induced by post-consummatory factors. Rats received discrete fixed interval cycles that could end in R or N. In Experiments 1¿2, with a 2-s intertrial interval, performance after N was higher than after R in groups receiving partial reinforcement (surprising N) or discrimination training (expected N), thus providing no evidence of response invigoration after surprising N. Experiment 3, with pairs of fixed interval cycles separated by 2-s mid-trial intervals and 30-s intertrial intervals, yielded evidence of response invigoration after N. The frustration effect was larger in the partial reinforcement group than in the discrimination-training group. The frustration effect was also observed to emerge immediately after a transition from continuous reinforcement to partial reinforcement in Experiment 4, in which the effects of R and N were measured on a common, continuously reinforced variable interval cycle. The length of the mid-trial interval was manipulated in Experiments 5¿6. The frustration effect was observed with 2 and 4 s intervals, but it disappeared at 20 s because behavior both decreased after N (invigoration by surprising N) and increased after R (post-consummatory suppression). Additionally, the after-N process decayed at a higher rate than the after-R process. The results resolve a long-standing controversy about the factors determining the frustration effect in rats. Two additional experiments with pigeons found the frustration effect to be entirely attributable to after-R processes.