The emergence of expertise with novel objects

Abstract

How do humans learn to discriminate novel objects? The puzzle of novel object recognition is particularly pertinent in forensics and security settings where human judgement is the primary means of identifying individuals based on unfamiliar photographs, fingerprints, footprints, bite marks, or CCTV footage. In this thesis, I use fingerprints as a testbed to examine the development of perceptual expertise in novel object recognition.There are four main empirical chapters. In Chapter 2, I probe the effect of context and familiarity on fingerprint matching decisions. In Chapter 3, I test fingerprint experts and novices on a family resemblance matching task that shifts the level of specificity from discriminating fingers to discriminating people more broadly. In Chapter 4, I contrast people's reliance on a general or an instance-based visual skill by looking for expert-novice differences across distinct changes in the structure of the task and stimulus class. In Chapter 5, I follow a group of fingerprint trainees as their perceptual expertise emerges over twelve months. I examine whether individual differences across four measures of expertise remain stable as trainees gain experience with matching prints. I find that fingerprint discrimination judgments are influenced by the similarity of the context and image-level information to specific prior cases. I provide evidence for a flexible memory retrieval process for recognising fingerprints across levels of specificity, and show that expertise in print discrimination is robust to changes in the structure of the task, but not to changes in stimulus class-fingerprint experts are not inverted face experts. I also find that fingerprint expertise emerges with domain-specific experience, and that it is possible to predict some aspects of performance ahead of time. When the images are briefly presented or when matching family resemblances, however, early individual differences are diluted with experience. I conclude that novel object recognition is facilitated by domainspecific experience and that experts rely on their memories for information that is distributed across their repository of prior instances. A greater understanding of individual differences, and the sorts of experiences that facilitate perceptual expertise will help to inform training and recruitment programs in applied visual domains.iii iv