--- begin forwarded text Delivered-To: clips@philodox.com Date: Fri, 30 Dec 2005 19:22:19 -0500 To: Philodox Clips List <clips@philodox.com> From: "R. A. Hettinga" <rah@shipwright.com> Subject: [Clips] Why Airport Screeners Sometimes Don't Spot Guns, Knives, Scissors Reply-To: rah@philodox.com Sender: clips-bounces@philodox.com <http://online.wsj.com/article_print/science_journal.html?mod=djemtct> The Wall Street Journal December 30, 2005 SCIENCE JOURNAL By SHARON BEGLEY Why Airport Screeners Sometimes Don't Spot Guns, Knives, Scissors December 30, 2005 It seemed like good news on the antiterrorism front. Using a system called Threat Image Projection, British authorities digitally inserted one of 250 images of guns or other banned objects into X-ray images of bags passing through screening at a U.K. airport. At first the screeners' performance was mediocre (authorities will not release results, citing security concerns), but over several months it improved markedly. Then the testers changed their image library. Although the new images belonged to the same categories of forbidden items, the screeners' performance dropped off a cliff, to no better than it was when the TIP program began. The screeners had become eagle-eyed at recognizing the 250 images that kept popping up, but were apparently unable to generalize to images of a gun with a different grip or a knife in a different orientation. As experts in "visual search" study explanations for the screeners' blindness, they are discovering that there are more ways to miss targets than they ever imagined ... or feared. Scientists have long known that the ability to pick out a target in a complex scene suffers when there are loads of things you're not looking for. That may be why screeners at Newark Liberty airport missed a butcher knife in a cluttered handbag this year. In all the visual chaos and "distractors," they missed the knife. Another well-known security blind spot is trying to identify an item surrounded by distractors that are similar to it, as when trying to find a particular pair of pumps in a stuffed shoe closet. This "target-distractor similarity" will be exacerbated, experts say, by the Transportation Security Administration's decision to allow small scissors, knives and other tools in carry-ons, as of last week. But it is the problem highlighted by the U.K. test, of being unable to recognize variations on dangerous themes, that has security experts most worried. Called "target-target dissimilarity," it's familiar to anyone who has looked for a beer in the refrigerator. The suds may be front and center, but if the bottle looks different from the one in your mind's eye, it may as well be invisible. In studies funded in part by the TSA, cognitive scientist J. David Smith of the State University of New York, Buffalo, and colleagues trained scores of volunteers to learn a number of origami-like targets, the original shape as well as rotated or slightly distorted versions. The volunteers saw one shape at a time, on a computer screen, and determined whether it belonged to a target category. They made the right call as much as 76% of the time. Since the bird-like shapes were arguably more complex than, say, a knife, that wasn't bad. The scientists then embedded target shapes in cluttered scenes, where items overlapped, touched and took different orientations. The volunteers got better and better at finding the targets, as real screeners do. But as soon as they had to spot a target that was slightly different from the one they had learned, performance plummeted. The volunteers spotted only targets they had seen repeatedly, not variations of them. It was as if someone learned what dogs are by studying dachshunds and poodles, and then didn't recognize a spaniel as a dog. "Screeners have trouble applying their knowledge of what some guns and knives look like to identify other members of that category," says Prof. Smith. "They instead rely on what we call a specific-token strategy, remembering Bowie knives and Beretta guns and other images they've been trained on, but not in a way that lets them generalize to guns and knives they've never seen." Terrorists obviously aren't limited to weapons that screeners are trained and tested on. As Prof. Smith and his colleagues write in the current issue of the Journal of Experimental Psychology: Learning, Memory and Cognition, "It is a serious matter." Just to be sure there wasn't something especially difficult about the origami-like shapes, the scientists had 88 participants try to spot actual knives, scissors and guns in X-ray images of cluttered suitcases. Again, there was a steep learning curve as specific targets repeated, with people eventually spotting 90% of the contraband. But as soon as slightly different guns, knives and scissors were digitally inserted into the image, scores fell. Unable to see family resemblances, people missed three times as many novel items as familiar ones. Oddly, people did hardly better when they had to find one target category rather than three. That suggests that reducing the number of forbidden carry-ons won't make screeners catch more of them. "I was surprised that performance wasn't better with fewer targets," Prof. Smith says. "There's little reason to think that decreasing the number of items screeners have to spot will improve the quality of the search." The TSA has reportedly upped the number of images in its training library to more than 1,000. That may help, but what's really needed is a way to train the brain to broaden its search criteria so that it can spot any knife or gun after being trained on hundreds of specific knives or guns. So far, no one knows how to do that. -- ----------------- R. A. Hettinga <mailto: rah@ibuc.com> The Internet Bearer Underwriting Corporation <http://www.ibuc.com/> 44 Farquhar Street, Boston, MA 02131 USA "... however it may deserve respect for its usefulness and antiquity, [predicting the end of the world] has not been found agreeable to experience." -- Edward Gibbon, 'Decline and Fall of the Roman Empire' _______________________________________________ Clips mailing list Clips@philodox.com http://www.philodox.com/mailman/listinfo/clips --- end forwarded text -- ----------------- R. A. Hettinga <mailto: rah@ibuc.com> The Internet Bearer Underwriting Corporation <http://www.ibuc.com/> 44 Farquhar Street, Boston, MA 02131 USA "... however it may deserve respect for its usefulness and antiquity, [predicting the end of the world] has not been found agreeable to experience." -- Edward Gibbon, 'Decline and Fall of the Roman Empire'