Flattening of a generalization gradient following a retention interval: Evidence for differential forgetting of stimulus features

•Rats with a salt need show a preference for a flavour previously paired with saline.•Rats can distinguish between this flavour and a second novel flavour.•Placing an interval between training and test flattens the generalization gradient.•It is proposed that stimulus representations change over the passage of time. In two experiments, rats received exposure to a compound consisting of a solution of salt plus a distinctive flavor (A), followed by an injection of furo-doca to induce a salt need. Experiment 1, established that this procedure successfully generated a preference for flavor A in a subsequent choice test between A and water. Experiment 2 used this within-event learning effect to investigate generalization, testing the rats with both A and a novel flavor (B). For different groups the interval between the training phase and the test phase was varied. Subjects tested immediately after training showed a steep generalization gradient (i.e., a strong preference for A, and a weak preference for B). Subjects given a 14-day retention interval showed a flattened gradient, a reduced level of preference for A and an enhanced preference for B. These results are interpreted in terms of changes in stimulus representations over the retention interval that act to reduce the effectiveness of the distinctive features of stimuli (the features that are necessary to ensure discrimination between them).