Little is known about the nature and development of fingerprint expertise and, therefore, the best way to turn novices into experts. Little is known about the factors that affect matching accuracy and, therefore, what experts can legitimately testify to in court. This thesis explores the factors that affect matching accuracy and the development of expertise in fingerprint identification, in order to inform training, and to provide an empirical basis for expert testimony in the courtroom. The investigation is grounded in exemplar, signal detection, and dual-process theories, and draws from literature on expertise and skill acquisition, and familiar and unfamiliar face recognition.
The thesis comprises four parts. In Part 1—Establishing Expertise—I attempt to find evidence for expert-novice differences in fingerprint matching, and explore where performance differences might lie. In Part 2—Depicting Expertise—I explore alternate methods for presenting signal detection results by depicting the data in a contingency space. In Part 3—Nature of Expertise—I explore the cognitive processes that might account for the superior performance of expert fingerprint examiners, and I explore the limits of rapid expert decision making. In Part 4—Expression of Expertise—I propose a framework for the expression of expert opinion in the courtroom, in order to integrate extra-legal recommendations and emerging research.
Taken together, I find that qualified, court-practicing fingerprint examiners are more accurate and more conservative than novices, and that errors are more likely to occur on prints from large databases, which are highly similar but nonmatching. I find that performance, both in terms of accuracy and response bias, changes as people move from novice, to trainee, to expert. I find that experts can discriminate matching and nonmatching prints that are artificially noisy, spaced by a short time, briefly flashed on screen, and even when presented in the blink of an eye. These findings indicate that experts make use of non-analytic processing when identifying prints, and they perform accurately when information is sparse—experts can do a lot with a little. Further programs of research, like this one, on the factors that affect fingerprint matching accuracy and performance, will create a foundation for evidence-based training, and serve to increase confidence in the legitimacy of claims made by expert witnesses in the courtroom.