College of LAS « Illinois


Schooling Baggage Screeners


As you wait shoeless in the airport for your suitcase to be x-rayed, put yourself in the shoes of the baggage screener.  A screener has just three seconds, on average, to scan an x-ray image of a suitcase and distinguish knives from combs, guns from hair dryers, and bomb wiring fromelectronics. "It's a pretty tough task when you think about it," says Arthur Kramer, an LAS professor of psychology. So early in 2001, Kramer proposed a study to the Federal Aviation Administration (FAA) to see if training could help them do it better.

Kramer, along with Jason McCarley, who is a professor at U. of I.'s Institute of Aviation, and three other U. of I. colleagues enlisted 16 novice baggage screeners (aka college students) for five days of intensive practice and testing. From the FAA, the researchers obtained 100 x-ray images of luggage cluttered with ordinary items like clothes, hair dryers, and pill bottles. On 20 percent of those images, they used a computer to superimpose images of knives among the other items. The novice screeners viewed those images while wearing an eye tracker—a head-mounted, computer-linked video camera that records eye movements, thereby providing "a window on the mind," Kramer says.

Four days of practice helped screeners significantly improve their performance: They spotted 71 percent of the knives the first day and 80 percent on their final day. What's more, the screeners' eye movements revealed that training helped them quickly distinguish knives from harmless objects.

But training didn't help much when screeners encountered unfamiliar knives. "If the expectation is that we can train screeners on one set of weapons and expect them to transfer that ability to another kind of weapon, that is not the case," Kramer says. The results, which were published in May 2004 in the journal Psychological Science, meant that screeners need continuing education to recognize new varieties of weapons.

Computerized baggage screeners could also help human screeners improve their accuracy, according to a more recent study by Kramer's group. Such machines automatically scan x-rays of luggage and circle suspicious items, but even the best of them work imperfectly. Kramer's team simulated an imperfect automated screen by circling 80 percent of the knife images. Both college-age and retired screeners performed 10 to 15 percent better when most of the suspicious images were flagged, which was encouraging. But today, he says, the best human screeners can do is to "look with their eyes."

Fall/Winter 2005–06