Ears Are Most Reliable Feature for Identifying Faces, Study Finds

Forget the eyes or mouth. New research suggests your ears might be the real secret to identification. Scientists at the National Institute of Standards and Technology in Maryland put forensic experts to the test with a surprising twist. They showed specialists photos of two faces plus a close-up of just one feature from either person. The single clue was often enough. Ears proved to be the most reliable marker for getting it right.

Lead author Carina Hahn expressed genuine shock at what she saw. What surprised me most, she stated, is that specialists could accurately judge identity from a single facial feature most of the time. It is remarkable how much identity information they can extract from so little. Can you do better than an expert? Try taking the test below to find out where you stand.

The team wanted to understand exactly how forensic face specialists work during investigations and court trials. Forensic face specialists are tasked with comparing facial images to determine identity, a process often guided by protocols emphasizing a comprehensive analysis of the full face. However, the strategies specialists employ regarding feature prioritization and accuracy based on specific features remain under-explored. They published these findings in PLOS One.

A total of 34 experts took part in the experiment. A total of 34 experts were given photos of two similar faces side-by-side, alongside a photo of a hidden face. For each pair, they were tasked with identifying which of the two faces matched the hidden face, with as little information as possible. The results revealed that the experts got 90 per cent of the answers correct on average. They could identify the target after seeing just one or two features.

Ears and forehead proved to be the most useful indicators in this specific study. In contrast, necks and chins were the least helpful features for making a match. Specialists have a striking ability to adapt their strategy depending on the face in front of them, rather than applying one fixed rule, Ms Hahn said. The combination of relying on features known to be useful, like ears, with adjusting when needed, hasn't been documented before.

These findings carry significant implications for training both human specialists and AI systems. It is feasible that facial specialists could significantly contribute to forensic contexts by assisting in training face recognition algorithms. Existing literature indicates that combining human and algorithmic decision-making yields more accurate outcomes than either approach alone. Studies have already begun to examine how skilled individuals can improve face matching algorithms, demonstrating that visual fixations from experts can enhance algorithmic accuracy. Visual information sampled from viewing patterns of face super-recognizers improved identity matching accuracy in a set of deep neural networks. Future research should explore how trained face expert-defined diagnostic features can further enhance the reliability of face recognition algorithms.