Detection Dogs: Are They Reliable? (Part 1)

Dogs can detect incredibly faint odors. But what if their human handlers give them the wrong cues? The Dog Trainer explores dangerous territory.

Jolanta Benal, CPDT-KA, CBCC-KA
5-minute read
Episode #229

The sensory world dogs live in seems unimaginable to me. We hear, more or less, frequencies between 20 and 20,000 Hz. Dogs hear between 40 and 60,000 Hz – again, more or less. Dogs are red-green colorblind, but they see better in the dark than we do; the placement of their eyes, on the sides of their heads, means they have a wider field of vision than we do but a narrower binocular field, which is where depth perception lives.

And then there are dogs’ noses.

Dogs have 44 times as many olfactory receptors as we do – 220 million to our puny 5 million.  Hence not only their ability to sniff out the particles of cheese-like substance left on the fast-food wrapper some jerk dropped in the gutter outside your house, but also such useful skills as finding kids lost in the woods and people buried under rubble in collapsed buildings. And, of course, there are drug-detecting and bomb-sniffing dogs, whose work leads to many an arrest and conviction.

But what if those detection dogs aren’t as reliable as their scenting abilities would make it seem? This week and next, I’ll talk about a couple of studies that explore this question, and offer a sketch of some possible implications..

In 2011, Lisa Lit of the University of California at Davis, and her colleagues Julie Schweitzer and Anita Oberbauer, published a controversial study in the journal Animal Cognition. The title says it all: “Handler Beliefs Affect Scent Detection Dog Outcomes.”  Uh-oh.

Clever Hans, the Very Clever Horse

Here’s the background: In the opening years of the 20th century, a horse named Clever Hans became famous for apparently being able to count, multiply, divide, and tell time – among other intellectual tasks you’d think would be beyond a horse’s capacity. He’d tap his hoof till he arrived at the number of taps that represented the correct answer, and then he’d stop. Clever Hans’s owner, Wilhelm von Osten, was not a fraud; he himself didn’t realize that although Hans couldn’t really add and subtract, he had learned to pick up on von Osten’s subconscious cues. Hans was a remarkably clever horse, but not in the way everybody thought he was.

Now, experiments with dogs – as well as life with dogs! – have shown that dogs are exquisitely sensitive to human social cues. Occasionally they even seem to rely on human cues when those cues contradict their own senses. For example, some of the dogs in one small study would approach an empty bowl that their guardians were pointing to, even if there was another bowl around in which the dog had seen and smelled food.

Clever Hans and Detection Dogs

Given that dogs are so responsive to human social cues, Professor Lit and her colleagues wondered whether the Clever Hans effect might occur even with highly trained detection dogs. In other words, if a human handler falsely believed that drugs or explosives were present, might the human unconsciously cue the dog to give an alert? And if so, would the dog give an erroneous alert?

The researchers recruited 18 dog/handler teams, all of them certified by law enforcement and actively working, with a record of success in real-world situations. The experiments took place in a church that had never been used for detection dog training. Also, the researchers believed it was relatively unlikely that explosives or illegal drugs would have been stored there.

So how did the experiment work?


About the Author

Jolanta Benal, CPDT-KA, CBCC-KA

Jolanta holds professional certifications in both training and behavior counseling and belongs to the Association of Professional Dog Trainers and the International Association of Animal Behavior Consultants. She also volunteered with Pet Help Partners, a program of the Humane Society of the United States that works to prevent pet relinquishment. Her approach is generally behaviorist (Pavlovian, Skinnerian and post-Skinnerian learning theory) with a big helping of ethology (animal behavior as observed in non-experimental settings).