Login / Signup

Evaluating the Reliability and Validity of the Famous Faces Doppelgangers Test, a Novel Measure of Familiar Face Recognition.

Emilia PozoLaura T GermineLuke ScheuerRoger W Strong
Published in: Assessment (2022)
Face recognition assessments that use images of celebrities require not only face recognition ability but also pop-culture knowledge and successful recall of identifying information. Here, we introduce a task designed to measure face recognition more specifically: the Famous Faces Doppelgangers Test (FFDT). Participants ( N = 57,407) identified 40 celebrities paired with lookalike doppelgangers, allowing face recognition ability to be assessed without requiring information recall. In addition, participants reported whether they were familiar with each celebrity, allowing poor face recognition ability to be differentiated from low pop-culture knowledge. FFDT performance was reliable ( r xx = .80), similar across participants of different racial and ethnic groups, and more highly correlated with memory for faces ( r = .50) and self-reported face recognition ability ( r = .48) than processing speed ability ( r = .10). Thus, the FFDT is a reliable, valid, and specific measure of the ability to identify familiar faces, making it a promising new tool for assessing face recognition ability.
Keyphrases
  • deep learning
  • machine learning
  • working memory