Police use of dogs reveal police biases

I linked to this article by Jack Hitt earlier, but in case you didn’t click the link, check out his conclusion:

. . . A 2011 study published in the journal Animal Cognition found that even expertly trained dogs and the most professional handlers cannot evade what is called the Clever Hans effect. In tests, dogs trained to detect explosives and drugs were sent, with their handlers, into a series of rooms to find non-existent contraband. In one room, there was a decoy that had been scented with sausage; in another, there was an unscented decoy accompanied by a sign telling the handler, falsely, that it smelled of contraband; a control room had no decoys. The investigators found, overall, that “human more than dog influences affected alert locations”: the meat decoy attracted more false alarms than anything in the control room, but the decoy with the sign prompted nearly twice as many false alerts as the one with the tempting scent. In other words, the dogs found their handlers’ unconscious cues significantly more compelling than the sausage. Trained animals, it turns out, are arguably better at reading our cues than we are at suppressing them.

A friend of mine recently saw a detection dog at a Metro-North station in Connecticut go into full alert when a fifty-something white businessman in a suit walked by. The trainer pulled the dog back and nodded for the man to pass. My friend noted, “Training a dog is not something that happens once and then it’s over. They are always being trained.” And so that Metro-North dog has begun a kind of continuing education. Repeat that action and the dog will learn: give middle-aged white guys in brogues a pass.

In 2011, the Chicago Tribune analyzed three years’ worth of data related to police dogs working in suburban Chicago. The analysis revealed how easy it is for certain cultural stereotypes—in this case, of drug-dealing Hispanics­—to travel straight down the leash. When dogs alerted their handlers to the presence of drugs during traffic stops, the officers found drugs forty-four per cent of the time overall. In cases involving Latino drivers, that number fell to twenty-seven per cent. In other words, the dogs were erroneously implicating people of Hispanic descent far more frequently than other people. When this disparity was made public, the question naturally arose of whether handlers’ unconscious cues might be the cause? Paul Waggoner, a scientist at Auburn University’s Canine Detection Research Institute, who was interviewed by the Tribune, provided the answer. A “big, resounding yes,” he said.

There is a paradox in police-dog work, post-Ferguson. Depending on its training, a dog may or may not be good at alerting us to drugs or bombs. But there is one social ill that all detection dogs, even the poorly trained ones, reveal with searing accuracy: the hidden racial prejudices of the police officers who deploy them.