Use this URL to cite or link to this record in EThOS:
Title: Face perception and hyper-realistic masks
Author: Sanders, Jet G.
ISNI:       0000 0004 7657 561X
Awarding Body: University of York
Current Institution: University of York
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Previous research has shown that deliberate disguise deteriorates human and automatic face recognition, with consequences for person identification in criminal situations. Common forms of deliberate disguise (e.g. balaclavas or hoodies) are easy to detect. When such disguises are used, viewer can distinguish between an unmasked individual - whose identity they knowingly can observe from facial appearance - and a masked individual - whose identity they knowingly cannot. Hyper-realistic silicone masks change this. Their recent use in criminal settings suggests that they effectively disguise identity and are difficult to detect. In this thesis, I first show that viewers are strikingly poor at distinguishing hyper-realistic masks from real faces under live and photographic test conditions, and are worse in other-race conditions. I also show large individual differences in discriminating realistic masks from real faces (5%-100% accuracy), and use an image analysis to isolate information that high performers use for effective categorisation. The analysis reveals an informative region directly below the eyes, which is used by high performers but not low performers. These findings point to selection and training as routes to improved mask detection. Second, I examine the reliability of estimates made of the person beneath the mask. Demographic profiling and social character estimates are poor, and results show that recognition rates were only just above chance, even for familiar viewers. This analysis highlights a systematic bias in these estimates: demographics, traits and social characteristics of the mask were attributed to those of the wearer. This bias has theoretical and applied consequences. First, it supports the automaticity with which viewers use a face to judge a person, even when they know the face is not that of the person. Second, it suggests that predictions of the person underneath the mask, by familiar and unfamiliar viewers alike, should be treated with great caution.
Supervisor: Jenkins, Rob Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available