Figure 3.

Memory performance for cued versus uncued low-value associations. Memory change scores were computed as the difference in recall error between the two tests. At retest, low-value associations that were not cued during sleep benefitted to the same extent as those that were cued (A, Experiment II). In contrast, when sounds were applied during wakefulness, cued associations benefited more than uncued associations (B, Experiment IV). Error bars depict SEM after correcting for across-subject variability, such that these measures of variability correspond to the error term for the within-subjects contrast as a function of cuing.