Accuracy, confidence and consistency in repeated recall of events

Abstract

In legal practice, both confidence and consistency of the testimony of eyewitnesses are often used as indicators for accuracy, but their usefulness has been questioned. The present study was designed to determine the relationship between accuracy, confidence and consistency in episodic memory. After viewing a video of a complex series of events, one group of participants was given an initial cued recall test after one week, and repeated recall tests after three and five weeks. A second group of participants was tested after three and five weeks, and a third group was tested only once after five weeks. Accuracy and confidence (at least for incorrect answers) decreased with longer initial retention intervals, but there was no decrease in either accuracy or confidence when recall was repeated. Repeated testing also did not lead to confidence inflation. Correlations between accuracy, confidence and consistency varied from medium to large. Inconsistencies were mainly caused by forgetting and reminiscence. These inconsistencies were recalled almost as accurately as consistently recalled information.