Influences of temporal factors on memory conjunction errors

Subjects viewed a series of faces presented two at a time for 16 seconds. Following either a 15-minute (Experiment 1) or 24-hour (Experiment 2) retention interval they received a recognition test that included old faces as well as faces constructed by recombining features from simultaneously presented study faces (simultaneous-conjunction condition), faces from successive pairs (near-conjunction condition), and faces that were two pairs apart (far-conjunction condition). In Experiment 1. false alarm rates decreased as the temporal distance between the relevant study faces increased. In Experiment 2, the false alarm rate in the simultaneous-conjunction condition was equal to the hit rate for old faces, and the false alarm rates for the other conditions was much lower. There was no effect of serial position during the study phase on the likelihood that parts of a face would later be miscombined to produce a recognition error in either experiment. The results suggest that witnesses to a crime are more likely to miscombine features of a to-be-remembered stimulus with those of another stimulus that was simultaneously present at the crime scene than with those of a stimulus encountered either earlier or later, especially when the test is delayed. Copyright (C) 2000 John Wiley & Sons, Ltd.