Retrospective revaluation effects in the conditioned suppression procedure.

In four experiments using the conditioned suppression procedure, rats received initial reinforced training with two compound stimuli, AX and BY, each compound consisting of one auditory and one visual element. After a second phase of training consisting of nonreinforced presentations of A, the suppression governed by X and Y was tested. In Experiment 1 X evoked slightly less suppression than Y (a mediated extinction effect). This outcome was obtained when the auditory cues served as X and Y(Experiment 1a), when the visual cues served as X and Y(Experiment 1b), and when the number of nonreinforced presentations of A was increased (Experiment 1c) from 18 to 216. In Experiment 2, however, in which the initial training was given with serial compounds (i.e., A --> X and B --> Y) X evoked more suppression than Y (a recovery-from-overshadowing effect). It was argued that extinction of A engages two learning processes, one increasing the effective associative strength of its associate (X) and one reducing it, and that the balance between these two depends on the specific conditions of training.