The Role of Familiarity in Face Learning and Face Inversion

Our ability to recognise familiar faces is excellent, yet our perception of unfamiliar faces is surprisingly poor. This large difference between familiar and unfamiliar faces is present across many areas of face perception research, such as face learning and face inversion. In the first experimental chapter, the role of top-down cues on face learning was investigated across 4 experiments, by manipulating the presence or absence of top-down information while participants learnt to recognise faces from either low or high variation images. A small benefit of learning faces from highly variable photographs compared to low variation images was found when faces were learnt with top-down information but not without. The largest effects in the first experimental chapter however were not related to the presence or absence of top-down cues as expected, but concerned whether faces were learnt (familiar) or novel (unfamiliar). Top-down cues were helpful but not necessary for face learning, and whether faces were familiar or unfamiliar showed the biggest effects. In the second experimental chapter, the role of stimuli cropping on the face inversion effect was investigated across 6 experiments. It was found that for famous faces, photographs cropped around the face produced large inversion effects, however, whole uncropped images produced much smaller inversion effects. Conversely, for unfamiliar faces, image cropping did not affect the size of the inversion effect. This thesis as a whole suggests that familiar and unfamiliar faces are processed qualitatively differently, and the differences between them can be explained by the nature of their underlying face representations. A new model of face representations is presented to explain the results, bringing together ideas from Valentine’s face space model (1991) and Bruce and Young’s recognition model (1986).