The effects of repeated testing on face recognition: Some new twists on a classic paradigm

Repeatedly testing memory for faces can ultimately impair identification accuracy. Ironically, eyewitness identification often involves multiple identification procedures. Eyewitness identification is often significantly influential in legal cases despite the fact that the most common error leading to the arrest and conviction of innocent persons in the United States is mistaken eyewitness identification. Fifteen experiments (n&barbelow;=493) investigated the effects of repeated testing on final recognition for original study faces ( OSFs) and altered study faces (ASFs) differing only by a group of facial features. An intervening recognition test presented between study and final test paired some OSFs (OSF-Shown) and ASFs (ASF-Shown) with new distractors, and omitted some study faces (Not-Shown). Instructions implied that all intervening face pairs contained an OSF. On a final recognition test, all OSFs along with an ASF counterpart corresponding to each OSF were presented. Participants attempted to correctly identify which faces were OSFs. Final recognition was generally enhanced for OSF-Shown relative to Not-Shown faces, and was always greater for Not-Shown relative to ASF-Shown faces. Response bias was liberal for faces shown on the intervening test and was conservative for Not-Shown faces. These general results were found when OSFs were studied once and multiple times, final test format was two-alternative forced-choice (2AFC) and yes/no, face visibility was reduced, study and test face format was congruent and incongruent, and internal and external ASF features were altered. Limiting study face exposures, reducing face visibility, and incongruent study/test face format impaired recognition and shifted response bias in a conservative direction. When face visibility was reduced, recognition was greater for faces with altered external versus internal features, suggesting greater reliance on memory for external features in poor viewing conditions. The likelihood of incorrectly recognizing a final test ASF-Shown face was greater when the intervening ASF was chosen versus seen (but not chosen). A response strategy based on familiarity and source monitoring difficulties likely contributed to these results. Consistent intervening impairment effects across numerous experimental manipulations emphasize the susceptibility of memory for faces to subtle changes, the potential contaminating effects of repeated testing, and the care that must be taken when collecting eyewitness information.