Abstract
For processing of facial identity, observers utilize visual information from both shape and reflectance. Recent evidence suggests an increasing importance of reflectance, at the expense of shape, with increasing face familiarity (Itz et al., 2014). Moreover, poor compared to good face recognizers were reported to disproportionately rely on shape when attempting to recognize learned faces (Kaufmann et al., 2013). In two experiments, we investigated the relative diagnosticity of shape and reflectance for matching of familiar and unfamiliar faces (Experiment 1) and identification of familiar and newly learned faces (Experiment 2). Within each familiarity condition, faces derived from a 3D camera system were morphed selectively in either shape or reflectance in steps of 20%, while holding the respective other dimension constant. Experiment 1 consisted of an identity matching task with unaltered S1 followed by morphed S2 stimuli. In Experiment 2 familiar and recently learned faces had to be identified from morphs. Individual scores from three tests were used to assess individual differences: the Bielefelder Famous Faces Test (BFFT), the Glasgow Face Matching Test (GFMT), and the Cambridge Face Memory Test (CFMT). Using multi-level model analyses, we examined probabilities of same versus different responses in Experiment 1, and probabilities of original identity vs. other/unknown identity responses in Experiment 2. Overall, our data revealed higher diagnosticity of reflectance compared to shape for both matching and identification, particularly for familiar faces. Moreover, compared to the respective average performance in each test, above-average BFFT (i.e. familiar face) performance was associated with higher utilization of reflectance, whereas both above-average CFMT and GFMT (i.e. unfamiliar face) performance coincided with higher utilization of shape. Our findings highlight the importance of reflectance information for face matching and identification, and indicate different underlying strategies with respect to familiar compared to unfamiliar face processing skills.
Meeting abstract presented at VSS 2015