Radley Balko has an interesting column noting the troubling implications of some of the arguments being made in the drug-sniffing dog cases currently before the Supreme Court. He points out that these dogs’ drug alerts are often unreliable because, in many cases, dogs seek to please their handlers rather than search out the truth. Thus, they often alert when they sense that the handler wants them to do so (e.g. – if he has a hunch). This leads to a very high rate of false positives in tests:
[T]he majority of the [Supreme Court] justices assumed that the nose of a dog is infallible — that an alert from a dog indicated the presence of whatever the dog was trained to find, and nothing else. An alert, then, was enough to establish probable cause for a more thorough search by law enforcement personnel.
That assumption was wrong at the time, and it has been repeatedly proven wrong since…
Consider another study conducted by Lisa Lit, a neurologist and former dog handler at the University of California-Davis. Lit brought 18 dog/handler teams currently operating in law enforcement agencies to an empty church. Each team conducted eight searches, each lasting about five minutes. If they were accurate, none of the dog/handler teams should have alerted in any of the searches. There were no drugs or explosives anywhere in the church.
But Lit had set some traps. The handlers were told that each search could have as many as three “target scents” — drugs for the drug dog teams, or explosives for the explosive dog teams. The handlers were told that in some cases hot packages were indicated by a piece of red paper. These red paper packages were designed to trick the handlers. Lit also set a trap for the dogs: Some of the packages contained unwrapped sausages.
The results were striking. The dogs falsely alerted in 123 of the 144 total searches. Because some dogs falsely alerted more than once in the same search, the total number of false alerts was 225. The dogs correctly completed the search without an alert just 21 times, for a success rate of 14.5 percent.
But here’s the more interesting part: The dogs were about twice as likely to falsely alert at the packages designed to trick their handlers than they were at the packages stuffed with sausages.
Why did so many fail? It wasn’t the dogs’ fault. A dog’s nose more than lives up to the hype. It is the finely tuned instrument you’ve always heard it to be. The problem is that for thousands of years, we’ve bred into dogs a more lovable trait: a constant, tail-wagging, cheek-licking desire to please us.
We’ve primarily bred dogs for protection and for companionship. The dogs that exhibited those qualities would get bred again, strengthening the traits from generation to generation. Over time, the dogs that were best at those two tasks were those that could read our body language, and react accordingly. This is why my dog barks when there’s a stranger at the door, but will curl up into a date’s lap within a few minutes of having met her. She’s picking up on my cues.
If a drug dog isn’t trained to account for this, it’s likely only confirming its handler’s biases and suspicions….
Balko also points out that police departments often have perverse incentives to use poorly trained dogs, because doing so makes it easier to get search warrants and seize property for asset forfeitures. I discussed the latter problem in this post. Finally, he notes that the conjunction of several individually plausible potential Supreme Court decisions could have very troubling effects in combination:
It might make sense to rule that a drug dog’s sniff is not a search under the Fourth Amendment, because a sniff is relatively unintrusive. There may be nothing unreasonable about ruling that a drug dog’s alert is enough to establish probable cause. After all, we all know that dogs have a finely honed sense of smell. And finally, it might make sense to rule that it is unreasonable to require prosecutors and police departments to provide a particular dog/handler team’s field history, because doing so would place an undue burden on law enforcement agencies.
Taken in isolation, you could make a good argument that these are all perfectly reasonable rulings. But put them together. By this time next year, we could be facing this terrifying reality: Police could take a dog/handler team into an apartment complex or to a row of townhouses and have them sniff dozens, even hundreds of residences. That team may have a history in which less than half the dog’s alerts lead to any actual recovery of narcotics. No matter. The police could then make note of all the doors at which the dog alerted, and all of those residences could look forward to middle-of-the-night visits from the local SWAT team.