The illusion that a horse could do maths may be behind sniffer dogs falsely ‘detecting’ illicit substances according to an intriguing study covered by The Economist.
The horse in question was called Clever Hans and he was rumoured to be able to do complicated maths, work out the date, spell German words – all from questions called out by the audience.
The trainer would run his hand across possible responses on, for example, a piece of paper, and Hans would tap with his hoof to signal when the correct answer was being pointed to.
Psychologist Oskar Pfungst became suspicious and eventually worked out than the horse was doing no more than waiting until his trainer changed his body posture when he hit on the right answer.
His trainer was completely unaware that his expectancies were shaping the horse’s behaviour but this form of unintentional behavioural influence over animal behaviour has become known as the ‘Clever Hans effect‘.
The Economist reports on a new study of sniffer dogs that seems to show a similar effect in action.
Sniffer dogs and their handlers were told to search an area that that might have up to three target scents and that on two occasions the scents would be clearly marked with bits of red paper.
In reality, there were no target scents, so anything the dogs detected was a false alert.
When handlers could see a red piece of paper, allegedly marking a location of interest, they were much more likely to say that their dogs signalled an alert. Indeed, in the two rooms where red paper was present and sausages were not, 32 of a possible 36 alerts were raised. In the two where both red paper and sausages were present that figure was 30–not significantly different. In contrast, in search areas where a sausage was hidden but no red piece of paper was there for handlers to see, it was only 17.
The dogs, in other words, were distracted only about half the time by the stimulus aimed at them. The human handlers were not only distracted on almost every occasion by the stimulus aimed at them, but also transmitted that distraction to their animals–who responded accordingly. To mix metaphors, the dogs were crying “wolf” at the unconscious behest of their handlers.
In other words, when the human handlers become suspicious the dogs are more likely to seem to detect suspicious scents, making the process a lot more subjective than the search teams like to believe.
Link to The Economist article ‘Clever hounds’.
Mind Hacks 
You meant:”a lot less objective than the search teams like to believe.”
Let rats do the sniffing!
Thanks antag9, now tweaked to say what I meant!
Fascinating!
For one thing although disconcerting this doesn’t have to be all bad. The human unconscious has access to far more information than the conscious. Therefore, the handlers may unknowingly add vital information to the scene as well. On the other hand it probably leads to the strengthening of stereotypes and social/racial profiling.
With good reason in psychology, medicine, etc. research has to be done double-blind. Even if the experimenter knows which one is the drug and which the placebo he will sway the results significantly.
Regards,
Jonas
My gripe with these screening tests is that false positive and false negative rates are never reported and hence makes them unlawful.
I’ve blogged about this sort of thing, here and suggested that the reason that there is a difference between medical screening tests and forensic screening tests is because medical tests are for us, whilst forensic tests are for them, ie criminals, see Us and Them.
Great post. I am more impressed with the horse’s cognitive abilities now than I was before. Sadly people only get excited when animals appear to mirror human “intelligence”.
This reminds me of a story which reported that Cyclosa mulmeinensis, an orb spider “creates imitations of iself” as a decoy to predators.
Now, how would a spider know what it looks like? This is the problem with animal behavior – interpretation. I assume the spider was basing its decoy on other individuals, who just happen to be the same species. That also makes more sense from a perspective of selective pressures, does it not?
Maybe one day we can figure out what was really going on with the supposedly psychic octopus – and the even more shocking behavior of our own species who thought such an intelligent animal should be kept in a glass tank.
A sniffer dog’s job is to collaborate with a handler to find suspicious items… dogs don’t work alone (with a few experimental exceptions). It is unsurprising that a handler’s bias is passed on to the dog in operation. The handler has got the experience, overview and intelligence the dog lacks, and the combination of the dog’s nose and the handler’s thinking and cues is what gets the sniffer job done.
Another question about the Clever Hans-effect & sniffer dogs:
When a drug smuggling criminal sees a drug detection dog, does his expectations to the dog’s ability to smell the drugs trigger a psycho-somatic reaction (heart rate, sweating, body language e.t.c) that cues the dog?
(and maybe cues the handler too?)
Yes, indeed, Emmy – strange to see how humans get all fired up over their own specialities and neglect others’. Perhaps this is inevitable and it wouldn’t really matter, except it so often involves us in unethical behaviour.
I watched this show on “Animal Intelligence” and it seemed to me that the show was to make the animal behaviour that mirrored Humans obvious. It’s so egoist and human-centric to think that animal’s are only intelligent when compared to humans. Uh, hello- I’d like to see a human go live in a jungle- I’d bet they would learn more from the animals!
This is a great article, and also shows something that is greatly important to us as a species- the empathy of animals and their ability to read and interpret our emotions and thoughts. Maybe we should start a bit of a two way street there?
@ PA,
Great comment. Re. empathy & ability to read humans, it also depends on the animal species’ evolutionary history (with humans).
Dogs are specialised in interpreting humans emotions due to their evolutionary history as work & companion species to humans through > 15,000 years.
I read about an experiment which compared wolf and dog puppies’ ability to understand human body language and interact with humans:
Dog and wolf puppies were brought up under the same conditions, before their ability to understand human body language and interact with humans was tested (not sure how).
The outcome was that 8 weeks old puppies could easily understand human body language and interacted naturally with humans, while the wolf puppies could not. The result suggested that dogs are somewhat ‘designed’ to be co-operative with humans through thousands of generations of selective breeding.
As for horses, while their human-cooperation history is shorter, they share several characteristics with dogs.
For example, horses are by nature socially advanced, emotional and very communicative animals that live in big social groups with a strong hierarchy. Humans relied on them for generations to get vital work done (farming, hunting, transport, war), and the horses’ ability to read human signs and co-operate with us was and is vital to their usefulness to us. Their human-relationship skills have been selectively bred for through many generations, and the horse is to day primarily a companion animal ~ empathy counts as high as ever.
I think that explains both Clever Hans’ and Sniffer Dogs’ tendency to pick up signals from their handler.
How interesting. It’s very funny that the humans were more distracted by their distractor than the dogs were by theirs!
This is especially interesting given that current legal status of police searches involving dogs does a bit of an end-run around the constitution:
http://reason.com/archives/2011/02/21/the-mind-of-a-police-dog
Every time I’ve seen sniffer dogs working, their handler is directing orders either verbally or with hand signals. So I would think the dog is very aware of the handlers conscious and uncoscious behaviour while working.
A dog-human work team is an ekvipage, just like a horse and its rider is… a working unit. Not just a dog with a handler or a handler with a dog.
The human influences the dog while the dog influences the human while the human influences the dog and so on and so on… like a mirror reflecting in a mirror.
The fine tuned co-operation and combined sniffer ability and human judgement is what makes the team effective… NOT the dog’s nose alone. Why would anyone think so?
Same with guide dogs… they guide the blind in co-operation with the blind. The blind guides the dog while the dogs guides the blind, and at the end of the day, the human is responsible for judging the situations the dog is there to alert about.
Same with all other working dogs… There is nothing new under the sun. Dogs were never independent ‘equipment’ (neither are we, or any other social animals I know of).
What is the fuss about?