Perceptual expertise in forensic facial image comparison

Area of Science:

• Forensic science
• Psychology
• Computer vision

Background:

• Facial identification is crucial in criminal investigations, often involving images with significant variations.
• Existing research on face matching accuracy primarily uses untrained individuals, limiting understanding of expert performance.
• There's a need to investigate human error sources in forensic facial examination.

Purpose of the Study:

• To assess the face matching accuracy of experienced forensic facial identification examiners.
• To compare the performance of forensic examiners with untrained participants and computer algorithms.
• To identify qualitative differences in face matching strategies between experts and novices.

Main Methods:

• Administered three challenging face matching tests to experienced forensic examiners.
• Compared examiner performance against untrained participants and automated facial recognition algorithms.
• Analyzed performance variations based on factors like exposure duration and image inversion.

Main Results:

• Forensic examiners significantly outperformed untrained participants and computer algorithms.
• Computationally fusing the judgments of multiple experts led to near-perfect accuracy.
• Examiners showed greater accuracy with longer exposure durations and were less affected by image inversion compared to students.

Conclusions:

• Experienced forensic facial identification examiners possess expertise in matching unfamiliar faces.
• Expertise in forensic face matching relies on distinct cognitive processes compared to general face memory.
• Collaborative decision-making among experts enhances face identification accuracy substantially.