Loading

Drug dogs false alert over 200 times in UC Davis study

  • by Russ Belville, NORML Outreach Coordinator February 4, 2011

    One of the favorite tools of law enforcement officers looking to bust cannabis consumers is the K-9 unit (or as George Clinton once called ‘em, the “dope dog”). These dogs are highly trained to use their super sense of smell to detect narcotics and explosives. Paired with a handler, they are often called in to search suspect vehicles in traffic stops and signal, or “alert” when contraband is detected.

    Researchers at UC Davis decided to put the K-9s to the test and it didn’t turn out well for the cop’s best friend. These detection dogs, whose alerts are used to justify search warrants and convict cannabis consumers, gave false alerts more than 200 times.

    Where's the ball? Where's the drugs? Where's the food? I'll do anything to make you happy, master!

    (SF Gate) The accuracy of drug- and explosives-sniffing dogs is affected by human handlers’ beliefs, possibly in response to subtle, unintentional cues, UC Davis researchers have found.

    The study, published in the January issue of the journal Animal Cognition, found that detection-dog teams erroneously “alerted,” or identified a scent, when there was no scent present more than 200 times — particularly when the handler believed that there was scent present.

    “It isn’t just about how sensitive a dog’s nose is or how well-trained a dog is,” says Lisa Lit, a postdoctoral fellow in the Department of Neurology and the study’s lead author. “There are cognitive factors affecting the interaction between a dog and a handler that can impact the dog’s performance.”

    Read more: http://www.sfgate.com/cgi-bin/blogs/pets/detail?entry_id=82270#ixzz1D2nQC9Ir

    The researchers took 18 drug dog teams to a church, where it is likely no drugs or explosives had ever been placed in the past.  The cops were told there might be up to three target scents in any one of four rooms.  If they saw a piece of red construction paper in the room, that indicated where a target scent was placed.

    The first room was left untouched.  The second room had a piece of red construction paper on a cabinet.  The third room had two sausages and two tennis balls placed as decoys.  The fourth room had the decoy scents and the red paper.  However, none of the rooms had any drugs or explosives.

    There shouldn’t have been any alerts, but, in fact, handlers indicated their dog had alerted in every room.  There were more alerts in rooms with red paper (which piques the cop’s interest) and no corresponding increase in rooms with sausages and tennis balls (which would pique a dog’s interest).

    In other words, at best, dogs are responding to the subtle non-verbal cues of their masters to find drugs or explosives where the human thinks there should be drugs or explosives.  The cop suspects you have pot so his body language makes the dog alert.  At worst, the cop is purposefully cuing his dog to alert when he wants a handy excuse to violate your 4th Amendment rights.

    Three years ago in Aspen a member of the NORML Legal Committee, Dan Monnat, gave an expert presentation of the faulty use of drug dogs to convict cannabis consumers.  Listen to the presentation below to get a good idea how law enforcement misuses the K-9’s testimony in court.

    Dan Monnat – Aspen Legal Seminar 2008 – Drug Dogs

    60 Responses to “Drug dogs false alert over 200 times in UC Davis study”

    1. [...] substances, not tobacco. However………. Drug dogs false alert over 200 times in UC Davis study Drug dogs false alert over 200 times in UC Davis study | NORML Blog, Marijuana Law Reform Good luck! Granny __________________ "If the truth won't do, then something is [...]

    2. dontdoit says:

      Based on your description of the way this “research” was conducted, it was seriously flawed. Anything that a canine is “trained” to do in law-enforcement ( or sporting ) use is just a focusing of the dog’s natural ability, and his pursuit of the reward (favorite toy) for showing his programmed “indication”. If you’re lying to a handler and telling him that there are narcotics hidden in a room and that the rooms with Red paper have narcotics, he’s for sure going to keep running the dog back and forth until the dog indicates out of frustration to receive his reward- the dog cares not about the drug. Now, if you give the handler 10 rooms and tell him there might be narcotics in any of the 10 rooms, and let him search, you would probably get a more reliable collection of data. Also, keep in mind that these dogs are trained to alert on odor, not the physical drug itself. They are taught to cone in on the point where the odor is the strongest, which might not be where the source is actually at. The dogs nose is thousands of times more sensitive to odor than ours is, which gives them the ability to hunt for the food they need to survive in the wild. This is why they are able to detect what is referred to as residual odor. The dog giving an indication on residual odor (think lingering skunk odor for example) is not wrong.. it is not a “false indication”.. he is only doing what you trained him to do. This is why dogs are able to indicate the presence of narcotics ( or whatever other odor they are trained to indicate on ) despite the coffee grounds, mustard, axle grease, or whatever other masking odors people attempt to use to thwart the dog’s nose. Odor of any type is like smoke on the breeze. It moves with air currents. This is where a handler has to use his common sense and reasoning to determine where the odor might be pulled from when a dog does indicate on a spot it believes the odor is strongest, but nothing is actually found in that one spot. This is why a smart handler will always start his dogs search (whether for narcotics or a person) downwind. It gives the dog the strongest advantage for being able to “tac in” to the source. You might scoff at the term “residual odor”, but think about this: If you take your car keys and throw them into a field of tall grass/weeds (without the dog watching of course) and then direct the dog to find them, he’s able to do it. It’s not because he’s been trained to smell keys as an object. He is hunting the residual human odor that is on the keys. This is also why a canine might often find drugs they are not actually trained to look for, such as LSD, “shrooms” or whatever. The cross-contamination (residual odor of cannabis or cocaine, etc.) catches the dog’s attention and he smells “his toy” is near. You know how you can walk into the house and smell the home-made chicken soup on the stove? You know what it is Mom is cooking. But I would lay odds that if you had the olfactory capabilities of the dog, you’d be able to name every ingredient that’s in that pot of soup, and then tell her what she left out! Just as we constantly are learning (whether you realize it or not) each time we repeat a task, learning how to do it quicker or more efficiently, the dog is always learning. And yes, the dog is relying on his partner (the handler) to teach him constantly, to tell him what it is the handler expects of him, how to respond to his environment. This is why in well structured training, the handler knows in advance where all the narcotics are and guides the dog’s learning, no rewards for incorrect alerts, and strong positive reinforcement for correct ones. Anything short of well structured training is only a test. The “research” as you describe it would be similar to taking a multiple choice test- You trust that the correct answer is among your offered choices. If I did not give you the option of D)None of the above, you’d still (most likely) pick one of the answers figuring that you just didn’t study everything there was to know. Unless of course you trusted yourself 100% because you knew you’d studied 100%. then you might challenge the question. But how may people have the self-confidence to do that? Similarly, the handler must trust his dog 100% based on the fact he knows with certainty he trained his dog to the best of his ability. If you lie, and tell someone the answer is there, they’re bound to give you an answer, even if it’s the wrong one.. just like this “research” did.

    3. Russell says:

      In reply to Dontdoit: I hope you are not nor have ever been a police officer. Your last statement is “If you lie and tell someone the answer is there, they’re bound to give you an answer even if it is wrong.” When I was a police officer, including assignments with a canine and as a narcotics detective, never did I nor any of my partners ever gave an answer if we didn’t know, even if we knew there was one. There is always an answer, but many times a police officer does not know it and should so state: “I do not know.” That goes for the officer handling the drug dog. If drug dogs were so successful, they would agree to participate in double blind studies. Please show me a police department that will allow its drug dogs to participate in one. They won’t because the results will not be acceptable to law enforcement and the public.

    4. Dave says:

      @Russell The police officers would probably state that they won’t participate because they have better things to do.

    5. kitty says:

      @Dave – better things to do than to learn if they are not potentially harming innocent people? There were multiple cases recently when innocent people were harmed, raped (according to legal definition of rape) by police officers or doctors, subjected to potentially harmful medical procedures (Mr Eckert, Jane Doe, Timothy Young) because of dogs.

      I would say that the police should be interested in learning if the dogs are really as effective as they think rather than “having better things to do”. It takes money and resources to train dogs and both false negatives and false positives are harmful.

    6. Dan says:

      An animal with the intelligence and emotional maturity of a well-trained toddler has no business giving a law enforcement officer permission to ruin my day.

      They are not machines, and they are nowhere near as reliable as a machine. Just like a human officer, they have biases, desires, aversions, and emotions that will cause them to adjust their behavior. Unlike an officer, they cannot understand written or verbal instructions, nor do they have any comprehension of things like civil rights, drug control, or the greater good of society. They are completely and totally unfit to fill this particular role of service.

      Trying to locate drugs or explosives that you have already probable cause to search for? Get the dogs. Trying to search a public area for explosives? Get the dogs. Are you actively trying to stop and search random people? That is illegal without probable cause, and a dog is too stupid, too emotional, and too unpredictable to provide it.

      I love my GSD with all my heart, and as dogs go, he is extremely intelligent. That said, I would never put someone’s civil rights in his hands. He has very serious and extremely obvious limitations, because he is a dog. All the training in the world will not change that.

      The fact the dogs are your best or only tool does not make them an acceptable tool. Civil rights are more important than drug control. As someone who has been subjected to several humiliating searches and idiotic lectures, courtesy of incompetent border patrol dogs, handlers, policies, or some combination thereof, I strongly object to handing this sort of judgment call over to a dog.

      And dontdoit, even if you are correct about the reason for the false alerts, it does NOT mean the research was flawed. In such a case, it would have exposed an error in training that caused dogs to associate the act of alerting to a reward: an end to a frustrating search. However, based on my experience, it is much more likely the dogs were responding to the handlers. Every time I was searched, I was wearing old, beat-up clothes, and my car was a mess. Every time I was not searched, my car and I were both very presentable. The dogs could not see inside of my car. Coincidence? I doubt it.

    7. […] they sometimes find drugs? Yes. But far too often they are wrong. And they are wrong enough times that they should not be used to establish probable cause. Frankly, […]

    Leave a Reply