Feature and Conjunction Errors in Recognition Memory: Evidence for Dual-Process Theory

Abstract

Feature and conjunction errors in recognition memory were investigated using a dual-process framework. In Experiment 1, dividing attention at study or test decreased old word recognition but did not influence feature and conjunction recognition errors after correcting for false alarms to new words (baseline). In Experiment 2, a response deadline manipulation influenced old word recognition but not feature and conjunction effects (i.e., feature or conjunction error rate minus baseline). Across Experiments 3 and 4, study repetitions increased the probabilities of feature and conjunction errors for participants under strong pressure to respond quickly. However, no such increases were observed for participants who were given more time to respond, providing evidence that the familiarity underlying feature and conjunction errors can be countered with recollection. Thus study repetition increased both familiarity and recollection. Feature and conjunction errors are based on familiarity in the absence of recollection. An approach that combines an item-associative distinction with a dual-process framework (e.g., Yonelinas, 1997) also can account for these errors. However, an approach that uses a feature-configuration distinction must be modified to account for these results.