WASHINGTON - Tracking eye movement can reveal when a person recognizes another, even when they try to hide it, according to new research.
Attempts to conceal recognition made it easier to spot in the new study, which could be used in criminal investigations to gain information from uncooperative witnesses or suspects.
Lie-detector tests are used in such cases as criminal investigations and U.S. Federal Bureau of Investigation interviews, but the results are notoriously unreliable. When asked questions critical to a criminal case, a racing heartbeat or sweaty palms might incriminate a nervous truth-teller, while a practiced liar may be able to control those signs and avoid suspicion.
A team led by Ailsa Millen, a psychology researcher at Britain's University of Stirling, tested a different method to extract information: tracking the movement of people's eyes as they look at photographs of faces. Instead of detecting the physical response to lying, which can be misleading, the researchers looked for the hidden information itself: the knowledge of a familiar face.
"Humans are experts at familiar face recognition. Recognition of a familiar face is fast and reflexive," said Millen.
Your eyes trace a familiar face differently than they do an unfamiliar one. When people look at unfamiliar faces, their eyes tend to dance from feature to feature, pausing frequently but briefly as they try to identify the unknown person. When gazing at familiar faces, people tend to linger on just a few features.
Seeking hidden recognition
The researchers wanted to know if people could control their eye movements when attempting to hide the fact that they recognized a familiar face — or if their eyes would give them away.
They showed 48 students pictures of strangers and familiar professors. They asked participants to try to appear honest while lying about recognizing familiar faces.
The researchers gave half the participants a method they thought might help them hide their recognition: pausing in the same places as they looked at each face, starting from the forehead, then stopping at each eye as they move from one ear to the other, then down to the nose, mouth and chin.
Millen was surprised by how quickly the subjects reacted to both familiar and unfamiliar faces — glancing at just a few features before responding that they did or did not know the person — but the speed at which they reacted didn't prevent the researchers from detecting their recognition of the familiar faces.
"Concealing markers for facial recognition in eye movements is difficult, especially if you know that person well," said Millen. "The harder our participants tried to conceal recognition, the more apparent it was."
In most cases, the participants who were given a hint as to how to hide their knowledge of the familiar person were unable to do so. Like those who were instructed to do their best to seem honest without other instructions, lingering too long on familiar facial features was a giveaway.
A better way to collect information
Millen hopes that her findings can someday be used in a law enforcement setting — not to assign guilt, but to collect information and filter out who is connected to whom in a criminal case.
"I think the work is pretty novel," said Deborah Hannula, a psychology professor at the University of Wisconsin-Milwaukee who was not involved in the research. "It's incredibly important to detect whether someone knows something and isn't willing to reveal that in cases that have high importance, like terrorist investigations."
Hannula and Millen agree that while the technique is promising, it needs a stronger experimental base before it can be used in a law enforcement setting.
Millen noted that she and her collaborator explored only one method of hiding recognition, but many other methods could exist, some of which might be more effective. Additionally, it's not yet clear what role the degree of familiarity plays. In order for the method to be applied to law enforcement scenarios, it needs to be effective for faces that are very familiar as well as only slightly familiar.
The research was published in the scientific journal Cognitive Research: Principles and Implications.