Unmasking a shady mirror effect: Recognition of normal versus obscured faces

Abstract

Hockley, Hemsworth, and Consoli (1999) found that following the study of normal faces, a recognition test of normal faces versus faces wearing sunglasses produced a mirror effect: The sunglasses manipulation decreased hit rates and increased false-alarm rates. The stimuli used by Hockley et al. (1999) consisted of separate poses of models wearing or not wearing sunglasses. In the current experiments, we separately manipulated same versus different depictions of individual faces and whether or not the faces were partially obscured. The results of a simulation and four experiments suggest that the test-based, mirror effect observed by Hockley et al. (1999) is actually two separable effects.

Keywords:

• Mirror effect
• Signal detection
• Principal Components Analysis neural networks
• Dual-process theories
• Face recognition

Acknowledgments

This work was supported in part by Natural Sciences and Engineering Research Council of Canada (NSERC) discovery grants to each of the authors. We thank Megan De Jong for her able assistance in collecting the data. We also thank Mike Burton and Hervé Abdi for their insightful comments on an earlier draft of this paper.