In my last post, I asked why, if psychology was so interesting, scientific psychology PAPERS tend to be so boring?
My conclusion: expecting a typical scientific paper to be interesting is like randomly selecting a sentence from a great novel and expecting it to be profound. That's because scientific progress is made up of many small steps - many of which are represented in single scientific papers. To understand why a paper is exciting, you need to understand the sweep of the literature and the puzzle being addressed. That's why good journalists, good writers, and great textbooks make science fascinating. They tell us the story.
I had written that piece as I started to think about another blog called Bad Science and the post You're Ooonly Cheating Yourself. The blog - like a lot of writing about science - is pretty snarky - pointing out just how dumb scientists are and how dumb all of us are to accept as scientific things that really aren't. Things like, according to the author, sniffer dogs. My last post talks about all the ways in which we, as scientists, are rewarded for tearing down each other's work. I might also have said (but didn't) that it is much harder to do science - or anything else - than it is to critique it.
What started me on this journey had nothing, however, to do with the philosophy of science Rather, it was the simple little study that prompted the Bad Science blog: Handler Beliefs Affect Scent Detection Dog Outcomes published in Animal Cognition.
One of the classic anecdotes told by Intro Psych professors throughout the last century is the story of Clever Hans. Clever Hans was a horse made famous by the fact that he could do math, count, keep track of days of the month, read music, etc. Unfortunately for those interested in using this as evidence for animal intelligence, careful investigation showed that Hans could only correctly answer questions that his owner also knew the answers to. Not because Hans was telepathic. Rather, his owner inadvertantly telegraphed his tension as the horse got closer and closer to the correct response, then relaxed when Hans got to the right answer. This signaled the answer to the horse.
The Clever Hans Effect is known to most psychology students as a warning of just how easy it is to inadvertantly influence research participants and the importance of using double-blind procedures in the laboratory. It makes a good story.
Sniffer dogs are used by many police departments and particularly in airports to search for hidden contraband. The idea is that, because of their outstanding sense of smell, dogs can detect the subtle odor of drugs or explosives that might otherwise be undetected. They alert their handler to the presence of contraband by sitting down next to a suspect. Then the handler takes the next steps.
Four researchers at UCDavis replicated the Clever Hans effect in a study of how sniffer dogs are influenced by handler's beliefs. In particular, the authors took 18 dog-handler pairs and took them through a series of exercises in which the dogs searched for what were purportedly explosives and drugs. As is typical in both dog training sessions and in real life, the dogs were confronted by many distractions and falso leads: empty boxes and packages of yummy sausages.
What was interesting about this study was that there were, in fact, no drugs or explosives to be found. But the handlers thought there were. In fact, there were boxes labeled as containing contraband or being decoys. The handlers could read the labels. Presumably the dogs could not.
When the handlers thought there was something to be found, the dogs were much more likely to falsely signal that there was something to be found.This was true in all conditions, but particularly true when the target was falsely labeled as having something in it the dog was supposed to find.
In other words, when trainers thought there was something to be found, the dogs found it.
This is an interesting study, not just because it nicely replicates 100 year old work on Clever Hans. From a policy perspective it's certainly more important. It strongly implies that when a handler believes that someone looks suspicious - like a potential smuggler or terrorist - the dog may give an alert based on the HANDLER'S SUSPICION rather than based on INDEPENDENT CONFIRMATION.
In other words, handlers and dogs work as a team. Either the dog or the handler can trigger an alert.
That's important for interpretting dog behavior, because it means that the dog is not acting independently and can't be seen as an 'objective' and 'unprejudiced' reporter.
This is also important for training. It means that for handlers to take best advantage of their dogs' abilities, they need to train themselves to not react on their own suspicions. How to do that is actually an empirical challenge, because it is doubtful that the trainers were trying to influence their dogs - any more than was the owner of Clever Hans or any researcher working with research participants.
Just because the dogs can be influenced, doesn't mean they can't do their jobs
What this research does not do is say that sniffer dogs can't find contraband.
That is, however, the implication of the Bad Science piece, which suggests sniffer dogs are another example of pseudoscience. The authors of the original study made no such conclusion. In their conclusions, they write:
"It may be more parsimonious to suggest that dogs respond not only to scent, but to additional cues issued by handlers as well. This is especially plausible since, in training, alerts are originally elicited through overt handler cueing. Cueing in initial training may include overt cues, verbal commands and physical prompting. Cues may also include more subtle unintentional cues given by handlers such as differences in handler proximity to the dog according to scent location, gaze and gesture cues, and postural cues."
They go on to say:
"It is important to emphasize that this study did not evaluate performance of dogs when presented with scent. Handler-dog teams undergo substantial training and rigorous certification prior to deployment; all teams included in this study confirmed prior successful finds during active deployment. This study only considered number of alerts under the artificially manipulated condition of handler belief of scent when in fact no scent was present.
In conclusion, these findings confirm that handler beliefs affect working dog outcomes, and human indication of scent location affects distribution of alerts more than dog interest in a particular location. These findings emphasize the importance of understanding both human and human-dog social cognitive factors in applied situations."
So why, then, did the author of Bad Science write sarcastically of his enjoyment of 'high tech sniffer dogs', urging smugglers to get a haircut and buy a suit to avoid detection?
Because it makes a good story. And like many scientists, the author of Bad Science is trained and rewarded to tear down - critique - scientific research. And, like Clever Hans, Pavlov's dogs, and all the rest of us, he tends to enact behaviors more frequently that have been rewarded in the past.
Physics has gravity. Psychology has reinforcement. Both laws tend to work very reliably.
© 2011 Nancy Darling. All Rights Reserved