As an international armada of planes, ships and helicopters continues to comb the Indian Ocean for any sign of Malaysian Airlines flight 370, now missing for more than a week, Interpol confirms that two passengers aboard that flight were traveling on stolen passports.

Aviation experts say the incident highlights a major security gap at many airports: It is simply too easy to board a flight using someone else's photo ID.

A new study looked in to the reliability of facial recognition with photos. Researchers found that the fewer fake IDs people see, the harder it is to spot them when the do come along — and multiple traps can cloud screeners' judgment.

The study, conducted by Megan Papesh of Louisiana State University and Stephen Goldinger of Arizona State University, was published in the journal Attention, Perception, & Psychophysicsin February.

Find The Fake

The study authors showed subjects a set of photos of people they had never met. The set was made up of pairs: One photo was a photo ID taken months or even years earlier, the other was a candid contemporary shot. In some cases, the photos were of two different people — one standing in as a "fake ID." Researchers asked subjects to pick out the pairs that were not of the same people.

When the rate of fakes was high — that is, when half of the photo IDs didn't match their user — those surveyed were wrong 20 percent of the time. But with fewer fakes — more closely resembling a real-world situation like an airport security line — the number of errors skyrocketed. Even when people were given multiple opportunities to detect their errors, they failed to pick out the fake nearly half the time.

Researcher Megan Papesh says one reason we're so bad at picking out fake IDs is that people change the way they look all the time — their hair, weight, whether they wear glasses. "Myriad changes occur, and that makes people willing to accept a lot of changes," she says.

Learning From Bouncers

One lesson from this study may be that security agents, who rarely see fake IDs, can learn something from bouncers at bars, who see many more fakes.

"A lot of my research assistants have told me that bouncers at clubs are really good at spotting fake IDs, despite the motivation to let people in and sell alcohol to them, because they encounter so many of them," says Papesh. "We're interested in providing training to individuals who are tasked with doing this in more security contexts with those bursts of fake IDs, kind of like they would get if they were bouncers at a club."

Another possible way to make the system more reliable, Papesh says, would be to take better pictures in the first place. For example, in passport photos people are asked to put long hair behind their ears and take off glasses. Most states do not require this for driver's license photos. Another possibility would be requiring multiple photos of the same person for a single ID. Papesh says previous research has suggested that having multiple photos, perhaps from multiple angles, can help screeners identify fakes.

Ultimately, though, are photo IDs just a bad way to identify people?

"Unfortunately, it's simultaneously the best and the worst way that we have," says Papesh. Reliable computer face-recognition is still a long way away, and other technologies are more invasive.

"Barring anything much more invasive like retinal scans or thumbprint ID," she says, "face matching is really the best way to go without being too terribly invasive."

Copyright 2021 NPR. To see more, visit https://www.npr.org.