The limits of human perception make it difficult for airport screeners to spot threats. Researchers aim to test those limits and find ways to improve performance.
Talk of security screening challenges typically focuses on concealment methods that terrorists might employ and on how technological advances might help screeners find whatever threat is being concealed. But the most critical link in the screening process is also often the weakest: the human screener.
It’s not a matter of sloth or distraction, but of the limitations of human perception and cognition. Research funded in part by the U.S. Transportation Security Administration (TSA) has shown that in a long series of images containing target objects, people are likelier to spot more common threat objects than rare ones and that overall detection drops precipitously when test subjects are required— as airport screeners are—to search for various threats at once. The fact that screeners must also focus on liquids of more than three ounces exacerbates the problem because they are constantly distracted from rarer threats by more common ones.
Another concern is called the “satisfaction of search” error. It was a problem when the same technology was first applied for medical purposes. In that case, a radiologist might spot a harmless anomaly on a patient’s chest x-ray, and terminate the search, failing to look further for another anomaly that could be a tumor. In security operations, the risk is clear: a perpetrator might line a backpack with explosives, and place a small pocket knife in the bag. When the person passes through the security checkpoint and sets off an alarm, the secondary hand search reveals the knife, which satisfies the screener, and the would-be attacker is sent on his way with the bag containing explosives.
Existing TSA screening procedures dictate that once an inert threat object—like a pocket knife—is found and confiscated, a physical search is conducted and then the bag is sent through the x-ray machine again. But with the constant focus on liquids, the screener’s likelihood of spotting objects like boxcutters or well-concealed explosives, even on a secondary scan, is still a concern.
With that in mind, the TSA funded research on satisfaction of search through the Institute for Homeland Security Solutions, a North Carolina research consortium that includes Duke University and the University of North Carolina at Chapel Hill.
According to results published in the Journal of Experimental Psychology: Applied, volunteers were asked to look at a standard computer monitor to spot and mouse-click on small Ts at different orientations and of varying darkness, from gray to black, against a clouded background shaded white to light gray. Compounding the challenge, among the Ts were “distractor” Ls, some with their longer, upright line offset slightly toward center. Subjects were told that in each frame there would be zero, one, or two targets.
Three experiments examined the impact of target frequency (number) and salience (prominence) on performance. The finding: salience does not initially affect detection, but—reinforcing prior research—as the number of high-salience (darker) targets increases, subjects are less likely to spot low-salience targets. Experiments four through six doubled the time allotted for each scan to 30 seconds, which, not surprisingly, reduced errors.
Two more experiments looked at the effects of pressure and reward. In the first of these experiments, the volunteers scanned the monitor, luggage icons appeared in succession at the top of the screen, implying a backlog, which grew faster in the second experiment. Further, the most successful volunteer among groups of three won $10, doubling their basic compensation. Satisfaction-of-search errors increased in both experiments, in particular the latter.
The final two experiments examined the impact of volunteers’ expectations about targets’ salience and frequency by including multiple targets with equal prominence in each screen. In one such experiment, screens were equally likely to contain one or two targets, which did not cause a change in performance. In a second experiment, the likelihood was randomized, which did increase errors. The conclusion: multiple categories of targets do not affect error rates, but screeners are “indeed sensitive to expectations about the number of targets.”
The goal of the research, still in its early stages, is to develop training and procedure options to prevent satisfaction-of- search errors, lead researcher Stephen Mitroff, an associate professor at Duke and founder of the school’s Visual Cognition Lab, tells Security Management.
This finding about workload and pressure should “be an important consideration…in constructing the environment in which searches are conducted,” the researchers wrote, suggesting queues should be out of screeners’ view.
Volunteers in Mitroff’s experiments on search ability and threat sensitivity complete standard clinical self-assessment questionnaires on attention-deficit hyperactivity (ADHD) disorders. TSA spokesman Greg Soule explains that the agency’s screeners are subject to similar evaluation prior to assignment. “If there is a concern about a current employee’s ability to be attentive, a fitness-for-duty evaluation would be conducted which would rely on the same standards that are used during the application process,” he says.