Watching Out for the Clever Hans Effect

During her summer holidays, I tried teaching carryover and borrowing (in maths) to my daughter. Like everyone else, she picked up carryover easily. But borrowing proved a harder nut to crack (no surprise there either). I’d be sitting next to her as she practiced and at one point, it looked like she had learnt borrowing too. A couple of days later, I wrote down a couple of subtraction questions and stepped away to attend to a few other things. When I returned, I found she had made many mistakes in subtraction/borrowing. What had happened? How could she have forgotten given that she was practicing every day?

Then I remembered this horse called Clever Hans. In the late 19^th century, Wilhelm Von Osten started teaching his horse to count by tapping his hooves. Amazingly, the horse could soon do many maths problems, including fractions, multiplication, even working out dates! Sure, he wasn’t always right, but the accuracy rate was high enough for Germany’s board of education to take notice. They wanted to investigate closely, and Von Osten agreed. In 1904, the panel concluded Clever Hans was genuine.

But one psychologist, Oskar Pfungst, remained unconvinced. So he conducted additional tests, trying to eliminate and/or identify variables that might be affecting the outcome. Strangely, if the owner stood farther away when asking the questions, the accuracy of the answers decreased. But why? If the owner didn’t know the answer himself, the accuracy dropped down to… zero! Even if the owner knew the answer but was hidden behind a screen, Clever Hans’ accuracy fell to… zero. Was the owner signaling to the horse? Was all this just a fraud? But then why would the owner agree to so many tests? In fact, the same pattern of accuracy would occur even with others who asked the question instead of the owner! So what was going on?

Pfungst now turned his attention to the questioners. And voila!

“He almost instantaneously noticed certain shifts in the posture, facial expressions and breathing of the questioners whenever Hans tapped his hoof. With every tap, their tension seemed to increase; when the correct answer had been reached, it would disappear… Hans was taking these subtle shifts in tension as his cue to stop.”

The even stranger part?

“The most fascinating part of this was that both Von Osten and any other questioner involved had absolutely no idea that they were giving Hans cues. It was all completely unconsciously done.”

Today, the conveying of such unintentional cues is called the “Clever Hans Effect”.

Though I’ve not done any such rigorous study, I suspect this is exactly why my daughter was able to do her sums right… until I left the room. So now I give her the sums and walk out. If she gets it right now, (I hope) it means she’s learnt it!