The role of intrasource variability on the accuracy and development of expertise in forensic fingerprint recognition

Fingerprint examination is a fundamental component of forensic practice. Identification of suspects during criminal investigation often rests on the comparison of fingerprints obtained in different situations. Exemplars are high-resolution fingerprints that are obtained in controlled conditions, such those available in police stations. These must be compared with latent prints, which are obtained from surfaces at the crime-scene. In contemporary forensic practice, automated fingerprint identifications systems are often used to select exemplars that are similar to the latent being examined. The classification of exemplar and latent prints as belonging to the same or different individuals, however, rests on the perceptual expertise of experienced human examiners. Identification, therefore, entails judgements of similarity for images that are subject to different sources of variability. Variability in the spatial distribution of ridges and minutiae can be used to distinguish individuals whose prints are characterised by different patterns. Differences in the image obtained from the same finger at different times, however, produce intrasource variability that interferes with the classification process. Changes in the surface and techniques used to acquire exemplar and latent prints, produce perceptually salient changes in the images obtained from the same source. These may include differences in the luminance contrast that defines ridge patterns, as well as perturbations caused by contaminating materials or deformation of the finger during movement. Although intrasource variability is a defining characteristic of the fingerprint examiner’s task, little is known about its affect on the accuracy of fingerprint recognition or the development of expertise in human examiners.

The purpose of the proposed PhD is to investigate the impact of intrasource variation on the accuracy of fingerprint classification in expert and naïve examiners. To do this, the thesis will be organised into three phases.

1. Phase one will parametrically modify low-level image characteristics of a set of fingerprint stimuli to produce the intrasource variability typically confronted by examiners in forensic settings
2. Phase two will assess the impact of intrasource variability on a single and multiple low-level image characteristics on the accuracy of fingerprint recognition in expert and naïve examiners
3. Phase three will establish the potential benefits of manipulating low-level image characteristics to reduce intrasource variability between known and crime-scene (latent) prints during the identification process and training.

The PhD presents a unique opportunity to apply visual science methods in an applied forensic context. The study will primarily employ psychophysical techniques to measure discrimination thresholds as a function of intrasource variability during fingerprint recognition. The supervisory team have an excellent track record in psychophysical methods, as well as the application of Information theoretic and Bayesian models to visual categorisation tasks. Within the context of the thesis, the student will be encouraged to develop an independent approach and incorporate appropriate methods (i.e., eye tracking and electrophysiology) to assess information sampling and recognition processes in expert and naïve examiners.