Face recognition is widely held to rely on 'configural processing', recently defined as an analysis of metric distances between features. Given that face recognition concerns those faces of people who we know, it is suggested that our unique representations of familiar faces contain information about these metric distances. The experiments in this thesis examine the hypothesis that face recognition relies on 'configural processing' by comparing performance between familiar and unfamiliar faces in a range of tasks. Experiments in the first half of the thesis investigate the effects of geometric distortions on different face tasks. Experiments in the second half examine familiarity advantages in rescaling distorted facial images. The main findings are that face recognition might not rely on simple measures of metric distances between features, and that observers show a surprising degree of tolerance to configural changes applied to familiar faces. This suggests that an operationalisation of configural processing will need to consider other measures that do not survive the image deformations tested in this thesis. The findings are discussed in relation to existing research on familiar face recognition as distinct from unfamiliar face perception.
|Publisher||University of Aberdeen|
|Source Sets||Ethos UK|
|Type||Electronic Thesis or Dissertation|
Page generated in 0.003 seconds