Abstract
Identity comparisons of photographs of unfamiliar faces are prone to error but imperative for security settings, such as the verification of face identities at passport control. Therefore, finding techniques to improve face-matching accuracy is an important contemporary research topic. This study investigates whether matching accuracy can be enhanced by verbal instructions that address feature comparisons or holistic processing. Findings demonstrate that feature-by-feature comparison strategy had no effect on face matching. In contrast, verbal instructions focused on holistic processing made face matching faster, but they impaired accuracy. Given the recent evidence for the heredity of face perception and the previously reported small or no improvements of face-matching ability, it seems reasonable to suggest that improving unfamiliar face matching is not an easy task, but it is presumably worthwhile to explore new methods for improvement nonetheless.
Highlights
Matching a face to a photo ID is a very common procedure in security settings a large number of experimental studies showed that face matching is rather error-prone
High error rates were observed using a range of person-to-photo matching tasks, which resemble to a great extent the widely used verification procedures in security settings (Davis & Valentine, 2009; Kemp, Towell & Pike, 1997; Megreya & Burton, 2008; White et al, How to cite this article Megreya (2018), Feature-by-feature comparison and holistic processing in unfamiliar face matching
Participants were quicker to make the face-matching decisions after receiving holistic instructions using correct identification, t (19) = 4.63, p ≤ 0.001, Cohen’ d = 1.60, and correct rejections, t (19) = 4.44, p < 0.001, Cohen’ d = 1.24. This experiment examined the effects of verbal instructions focused on feature-by-feature comparisons and holistic processing on unfamiliar face matching
Summary
Matching a face to a photo ID is a very common procedure in security settings a large number of experimental studies showed that face matching is rather error-prone (e.g., for a review see Robertson, Middleton & Burton, 2015). All images were taken on the same day, under good lighting conditions, and were presented in a full-face view In spite of these optimal conditions, which could be never met in any real-life scenario, participants’ performance was rather low, with an error rate of 30% for target-present and target-absent trials. This low level of performance has been widely replicated (e.g., Bindemann et al, 2012; Megreya & Burton, 2008; Megreya, White & Burton, 2011; Megreya & Bindemann, 2013; Megreya & Bindemann, 2015; Megreya & Bindemann, 2017) even when the heavy demands of this 1-in-10 array methodology were remarkably reduced to a 1-in-1 face-matching task using a range of face-matching databases (e.g., Bruce et al, 2001; Burton, White & McNeill, 2010; Henderson, Bruce & Burton, 2001; Megreya et al, 2012; Megreya & Burton, 2008).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
