Controlled retrieval processing in recognition memory exclusion tasks.

ERPs were acquired in the test phases of two memory tasks where three classes of word were presented: (i) words encountered in a study phase (studied words), (ii) words presented at test for the first time (new words), and (iii) new words repeated after a lag of 7-9 words (repeated test words). In Experiment 1, participants responded on one key to studied words (targets) and on a second to repeated test words (non-targets) as well as to new words. In Experiment 2, participants responded on one key to repeated test words (targets) and on a second key to new and studied words (non-targets). The likelihood of a correct response to a target was higher in Experiment 2 than in Experiment 1. In both experiments, the focus for the ERP analyses was on parietally distributed ERP old/new effects, which are assumed to index recollection. Reliable parietal old/new effects were obtained for targets as well as non-targets in Experiment 1, but for targets only in Experiment 2. This pattern of data is consistent with previous suggestions that, when the likelihood of recollecting information about targets is high, participants use the success or failure of an attempt to recollect information about targets as the basis for distinguishing between targets and all other classes of test word. The findings in these two experiments are informative because they: (i) generalise those obtained in previous work to a different exclusion paradigm, (ii) add emphasis to claims regarding the potential utility of this particular paradigm in studies where changes in memory control according to age are assessed, and (iii) highlight important considerations when behavioural data obtained in exclusion tasks are employed in order to make estimates of the relative contributions of recollection and familiarity to task performance.