False Memories of Familiar Faces.

Prior familiarity has been shown to increase memory for faces, but different effects emerge depending on whether the face is experimentally or pre-experimentally familiar to the observer. Across two experiments, we compared the effect of experimental and pre-experimental familiarity on recognition and source memory. Pre-experimentally familiar faces were nameable US celebrities, and unfamiliar faces were unnamable European celebrities. Within both sets, faces could be made experimentally familiar via repetition during the learning phase (studied once or thrice). At test, all studied identities were represented by novel (i.e., not studied) photos, allowing us to test memory for the identity rather than the picture. In Experiment 1, repeated presentations of both face types increased recognition rates, but accuracy was generally higher for pre-experimentally familiar faces. Experiment 2 expanded on these findings by pairing the faces with background locations and manipulating associative strength of the face-location pairs. Although pre-experimentally familiar faces were again recognized more often, they were also more likely to be falsely labeled as "old" when paired with new background locations. These results have implications for basic and applied studies examining familiar versus unfamiliar face recognition.