For the first few years I had her, I was impressed by my late dog Harper's uncanny ability to assess people's character. She hated every crappy landlord and bad roommate. Barked at them. Snarled at them. Wouldn't go near them. But if I brought home a date I liked, Harper, a Shar Pei/Labrador mix, would curl up right next to the woman and turn on the charm. It took me several years to figure out that my dog wasn't a good judge of character; she was just good at reading me. She liked the people I liked and disliked the people who rubbed me the wrong way. For dogs descended from lines bred for protection and companionship, this talent makes sense. A dog adept at distinguishing friend from foe was likely to be kept around and bred, and one very good way to tell friend from foe is to read your master's body language.
My confusion about what was going on in Harper's head reflects a common misconception that is also apparent in the ways dogs are used in criminal investigations. When we think dogs are using their well-honed noses to sniff out drugs or criminal suspects, they may actually be displaying a more recently evolved trait: an urgent desire to please their masters, coupled with the ability to read their cues.
Several studies and tests have shown that drug-sniffing dogs, scent hounds, and even explosive-detecting dogs are not nearly as accurate as they have been portrayed in court. A recent Chicago Tribune survey of traffic stops by suburban police departments from 2007 to 2009, for example, found that searches turned up contraband in just 44 percent of the cases where police dogs alerted to the presence of narcotics. (An alert is a signal, such as barking or sitting, that dogs are trained to display when they detect the target scent.) In stops involving Hispanic drivers, the dogs' success rate was just 27 percent. The two largest departments the Tribune surveyed—the Chicago Police Department and the Illinois State Police—said they don't even keep track of such information.
But don't blame the dogs; their noses work fine. In fact, the Pentagon's Defense Advanced Research Projects Agency recently conceded, after 12 years and millions of dollars of research, that the canine snout, fine-tuned by millions of years of evolution, is still far more sensitive and reliable than any technology man has been able to muster when it comes to detecting explosives in places such as Iraq and Afghanistan.
The problem is our confusion about when dogs are picking up a scent and when they are responding to cues from their handlers. The Economist's "Babbage" blog summarizes a recent study led by Lisa Lit, a neurologist (and former dog handler) at the University of California-Davis, that demonstrates the startling consequences of that confusion:
[Researchers] asked 18 professional dog handlers and their mutts to complete two sets of four brief searches. Thirteen of those who participated worked in drug detection, three in explosives detection, and two worked in both. The dogs had been trained to use one of two signals to indicate to their handlers that they had detected something. Some would bark, others would sit.
The experimental searches took places in the rooms of a church, and each team of dog and human had five minutes allocated to each of the eight searches. Before the searches, the handlers were informed that some of the search areas might contain up to three target scents, and also that in two cases those scents would be marked by pieces of red paper.
What the handlers were not told was that two of the targets contained decoy scents, in the form of unwrapped, hidden sausages, to encourage the dogs' interest in a false location. Moreover, none of the search areas contained the scents of either drugs or explosives. Any "detections" made by the teams thus had to be false. Recorders, who were blind to the study, noted where handlers indicated that their dogs had raised alerts.
The results? Dog/handler teams correctly completed a search with no alerts in just 21 of the 144 walk-throughs. The other 123 searches produced an astounding 225 alerts, every one of them false. Even more interesting, the search points designed to trick the handlers (marked by the red slips of paper) were about twice as likely to trigger false alerts as the search points designed to trick the dogs (by luring them with sausages). This phenomenon is known as the "Clever Hans effect," after a horse that won fame in the early 1900s by stomping out the answers to simply arithmetic questions with his hoof. Hans was indeed clever, but he couldn't do math. Instead he was reading subtle, unintentional cues from the audience and his trainer, who would tense up as Hans began to click his hoof, then relax once Hans hit the answer.
In her wonderfully written (but strangely titled) book Inside of a Dog, Columbia University psychology professor Alexandra Horowitz further illustrates how humans can unconsciously influence canine behavior. She describes various tests in which researchers measure the problem-solving skills of domestic dogs in comparison to wolves. The tests include activities in which dogs and wolves are tasked with finding a ball or treat hidden in a room, tucked behind a screen, or sealed in a container. When the task involves minimal human contact, domestic dogs perform about as well as their more primitive cousins. But the more the experiments incorporate interaction with humans, the more poorly domesticated dogs fare. They tend to give up, then simply wait for the human researcher to get the prize for them. Horowitz explains:
By standard intelligence tests, the dogs have failed…I believe, by contrast, that they have succeeded magnificently. They have applied a novel tool to the task. We are that tool….We solve the puzzles of closed doors and empty water dishes….We humans are brilliant enough to extract hopelessly tangled leashes from around trees.…Dogs are terrific at using humans to solve problems, but not as good at solving problems when we're not around.
It is hard to overstate the implications of these findings for the use of dogs in police work. The dogs who failed Lit's scent tests did not lose their sense of smell. But in the process of domesticating dogs, we have bred into them a trait that tends to trump most others: a desire to please us—and toward that end, an ability to read us and a tendency to rely on us to help them solve their problems. Any training program that does not take this tendency into account will produce dogs who frequently issue false alerts.
The consequences of those mistakes are profound. As my colleague Jacob Sullum has explained, the U.S. Supreme Court says a dog sniff is not invasive enough to qualify as a "search" under the Fourth Amendment, so police do not need a warrant or probable cause to have a dog smell your luggage or your car. At the same time, however, the courts treat an alert by a drug-sniffing dog as probable cause for an actual, no-question-about-it search, the kind that involves going through your pockets, opening your luggage, looking in your trunk, and perusing your personal belongings. The problem is that a dog barking or sitting may be responding not to a smell but to his handler's hunch about a suspect's guilt. The reason we have a Fourth Amendment is precisely to prevent searches based on hunches.
The consequences of misusing police dogs go well beyond unconstitutional searches. A drug dog's alert can help establish a connection between a suspect's property and drug activity, allowing police to seize the property for possible forfeiture. Even if the owner is never charged with a crime, the burden is on him to go to court to win back what was his, a process that often costs more than the property is worth. In a case I reported last year, for example, college student Anthony Smelley had $17,500 in cash that he'd won in an accident settlement seized when police in Indiana pulled him over and a drug dog alerted to Smelley's car. It took Smelley more than a year to win the cash back in court, even though a subsequent hand search turned up no illegal substances.
Canine testimony can also play a key role in murder cases. Last September the Texas Court of Criminal Appeals overturned the 2004 murder conviction of Richard Winfrey Sr. because the case against him was based on an unreliable, pseudoscientific "scent lineup" in which Fort Bend County Deputy Sheriff Keith Pikett (now retired) claimed his team of bloodhounds alerted to the murder victim's scent on Winfrey's clothing. Pikett and his dogs have assisted in thousands of criminal investigations by police departments all over Texas. As late as last year, prosecutors were trying to use the results from one of Pikett's scent lineups to retry Anthony Graves after a federal appeals court threw out his murder conviction. Graves, who served 18 years on death row, has since been exonerated and freed.
Pikett is now the target of a class-action lawsuit brought by people who say they were wrongly detained or convicted based on his dogs' alerts. Similar questions have been raised about the methods used by the late John Preston, a former Pennsylvania state trooper who found a second career as a freelance dog handler in police investigations, mainly in Florida. Preston's dog helped convict dozens of Floridians in the 1980s. At least three murder convictions secured through Preston's testimony have since been overturned.
In 2006 University of North Carolina law professor Richard Myers conducted a statistical analysis (PDF) of police dog accuracy tests and concluded that the animals were not reliable enough to produce probable cause for a search, let alone serve as the cornerstone of a conviction. At least five states have banned or restricted the use of scent lineups in criminal cases, but they are still frequently used in courtrooms across the country.
Dogs can be valuable investigative tools. They are great, for example, at following a scent in searches for suspects or sniffing out survivors after a disaster. The bomb-detecting dogs in Iraq and Afghanistan are successful because their handlers have no preconceptions about where bombs may lie. Indeed, they are putting their lives in the dogs' paws. With no cues from their masters to cloud their judgment, the dogs are free to go about their task unbiased. But while Canis domesticus retains many of its wilder relative's sensory abilities, it is in many ways a man-made animal. When we don't take that reality into account, a dog can be worse than useless. But that's not the dog's fault. It's ours.
Radley Balko is a senior editor at Reason magazine.