Interesting report from Ben Goldacre today about drug sniffer dogs used by police. Which might not leap out at you as a fascinating source of potential gossip, unless you have a better sense of these things than I did.
But it’s turning out that the Clever Hans effect might play a bigger role in sniffer dogs’ drug detection than was previously thought. That is, the human handlers’ expectations of where the drugs are might have at least as much impact on the dogs’ behaviour as the actual scent of drugs.
So, if you look like the sort of person who might have drugs on you, it’s possible that you’re just as likely to be barked at than someone who looks innocent but is smuggling dope through customs.
It’s not the scariest affront to civil liberties I’ve heard this week – I doubt anyone’s going to be arrested just because a dog growled at you if you’re not actually carrying anything illegal (though more horrible things have happened). But as Ben points out, it highlights the “theatre” aspect of much of what passes for important national security these days.
One thing I’m left wondering, though, is how sniffer dogs’ efficiency could be improved. I mean, I have no idea how they’re normally trained, or exactly what they’re picking up on in their handlers that makes them think they should take a particular interest in the guy with the beard – but are they perhaps learning these cues in the training itself?
If, when the dogs are learning what scents to look out for and how to respond when they track one to the source, they’re always in the presence of handlers who know exactly where the drugs are that they’re supposed to be finding, then maybe it only makes sense for them to also associate certain human behaviours with the items they should respond. In which case, there may exist a way to re-design their training, so that the only thing they’re cued to react to is the smell of drugs.
Anyone who knows anything about how this kind of thing actually works should feel free to educate me, as ever.
