Visual Working Memory in Action: Investigating Identity Judgments across Fingerprints, Faces, and Paintings

Abstract

Fingerprint experts are skilled at matching fingerprints in a variety of contexts, including when prints are briefly shown, are inverted or visually noisy, and can discriminate between prints from the same individual but from different fingers (i.e., Smith's thumb and index finger). This ability has been attributed to superior working memory; however, theoretical models of visual working memory rely only on basic visual stimuli. What happens when these theoretical models are applied to real-world stimuli? This experiment tested whether increasing the amount of visual information available to people, from two to four to six spotlight samples of an image, improved their identification judgements across three stimulus types: fingerprints, faces, and paintings. As informational load increased from two to four to six spotlight samples, people were more accurate at determining whether the two images depicted the same identity. However, their confidence ratings were not aligned with their accuracy, suggesting a complex relationship that may be explored in future research through the application of signal detection models. Our findings supported summary-based encoding models of visual working memory, suggesting that when it comes to complex stimuli, we rely on summary statistics in encoding visual landscapes rather than encoding items independently. We also found that increasing visual informational load in the process of making identification decisions resulted in improvements in identification judgements across a broad range of naturalistic stimuli (namely, fingerprints, faces, and paintings). This experiment specifically provided a potential explanation for how fingerprint experts achieve such remarkable feats utilising their visual working memory. Keywords: visual working memory, informational load, identification judgements, real- world stimuli, fingerprints.