Distinctiveness Effects in Face Memory Vanish with Well-Controlled Distractors

The “distinctiveness effect” in face memory holds that distinctive faces are recognized better than typical faces, as reflected by higher hit rates and lower false alarm rates during recognition tasks. We propose that this effect is due to a systematic bias in the similarity relationships between target and distractor faces. Specifically, we claim that distinctive faces are recognized more successfully because they are on the whole more distinguishable from a set of distractor faces, but not because of any inherent advantage in the way we represent them in memory. To test this claim, we conduct two memory experiments using a low-dimensional parameterization of face space based on silhouetted face profiles (Davidenko, 2004). In Study 1, we conduct an old/new recognition task with face silhouettes to replicate the classic distinctiveness effect. In Study 2, we construct a set of face stimuli that eliminates the asymmetry of target-distractor distances between typical and distinctive faces. Under these conditions, distinctive faces lose their recognition advantage and actually produce poorer performance than typical faces. We discuss possible causes of this “typicality effect” in terms of perceptual learning.