Our ability to categorise and identify objects in our environment is impressive. We can classify objects, that vary dramatically, into distinct categories and we can recognise our friends' faces effortlessly – tasks that are notoriously difficult for computers to perform. This ability, however, can sometimes lead to overconfidence.
Indeed, fingerprint examiners claim they can match a fingerprint found at a crime-scene to a fingerprint from a suspect with 100% accuracy and to the exclusion of all other human beings that ever were or ever will be. Although the current identification process primarily involves human perception and judgement, very little psychological research has been conducted on these processes.
In my thesis, I explore the cognitive and perceptual mechanisms that underlie people's identification of complex visual patterns and aim to develop a statistical model of this task. Specifically, I will examine the function of expertise in forensic identification and apply this knowledge to improve the accuracy of fingerprint examiners.