Extinction of within-event learning is contextually controlled and subject to renewal.

Five experiments examined within-event learning in rats by inducing an appetite for one of the elements (salt) of a compound stimulus and assessing preference for the other element (almond). Almond preference was conditional upon (1) the almond flavour having been presented in compound with the salt, and (2) the assessment being conducted when the rats were out of sodium balance (Experiment 1). Presentations of the compound in one environment (A) and of the salt and almond elements in a second environment (B) resulted in greater almond preference when rats were tested in A than in B (Experiment 2). Almond preference was reduced when separate presentations of the compound and almond (Experiment 3) or of the compound and salt (Experiment 4) occurred in the same environments but not when these presentations occurred in different environments. Rats exposed to the compound in A and then extinguished to the elements in either A or B showed a reduced almond preference when tested in the extinction environment, but not when tested in the other environment (Experiment 5). Thus, extinction of within-event learning is context-specific and subject to renewal. The results were interpreted in terms of an associative model whereby separate presentations of the elements result in a symmetrical inhibitory link which is contextually gated (Bouton, 1993).