Despite facial recognition playing a crucial role in security and ID verification contexts, humans are highly error-prone when identifying unfamiliar faces. To address this, research has explored ways to improve accuracy, including the use of multiple-image arrays to increase exposure to within-person variability. While some studies suggest that arrays improve accuracy when compared to traditional single-image formats, findings across the literature are largely inconsistent.
The current study aimed to determine whether image variability (high or low) within image arrays can be manipulated to enhance performance. We hypothesised that while high variability arrays would enhance face matching performance in match trials, excessive variability may impair mismatch trial accuracy. We sought to identify a sweet spot of variability that maximises overall accuracy, by systematically manipulating variability across 12 experimental conditions. 52 participants were presented with a target image alongside either a single image, or a four-image array of a single individual, and judges had to compare if they were the same person (match) or different people (mismatch).
Results highlighted that high-variability arrays (4H, 3H1L, 2H2L) significantly improved accuracy on match trials compared to single-image conditions. In contrast to previous findings, mismatch trial accuracy was unaffected by variability. These results underscore the utility of high-variability image arrays in enhancing performance in unfamiliar face matching tasks. Moreover, individual differences analysis revealed that higher levels of neuroticism were significantly associated with lower overall accuracy, suggesting the predictive role of personality in face matching performance. These findings have important implications for optimising ID verification formats and suggest that tailored interventions, such as incorporating high variability image arrays, and addressing neuroticism-related impairments, could enhance identification accuracy in applied settings.
PLEASE NOTE: You must be a member of the University of Lincoln to be able to view this dissertation. Please log in here.