Clever Hans



I think human beings suck at online discourse. Let me rephrase that – I think social media technology is designed in a way to amplify a particular kind of human suckiness. For starters, it is constantly nudging and persuading us to share more stuff online. Apparently, 70% of our smartphone time is spent inside some form of messaging/social app like Facebook, WhatsApp or Twitter. And the average smartphone user is spending at least 4-5 hours a day checking their phones.

And I don’t buy into the “Oh, the internet was a nicer place a decade ago”. No, it wasn’t. Online discussions have been pointless and nasty since the days steampunk pigeons transmitted usenet messages, but it was just fewer people using it, that’s all. But now, we have American presidents retweeting racist stuff. It’s all part of a predictable path from the heady days of Godwin’s law on Internet forums to drive-by screenshotting on Twitter.

So all in all, we are gloriously magnifying our suckiness by being addicted to technology that is exceptionally good at magnifying our suckiness (Silicon Valley calls this business model “monetizing attention”)

And to pull myself out of this toxic miasma of depressing thoughts, I am going to tell you a story.

It’s a true story about a horse named Hans.

Source: Wikipedia

Hans was an Orlov Trotter Horse, Russia’s most famous breed of horse, one that was famous for its trotting speed (particularly suited to uptempo OP Nayyar songs) and high stamina. The breed was developed as the outcome of love jihad between Arabian stallions and European mares.

But Hans’ core competency was not laying down rhythm tracks for OP Nayyar songs or racing. He was good at mathematics. In fact, he was astoundingly good at it.

He was owned by a Wilhelm Von Osten, a mathematics teacher, amateur horse trainer and phrenologist. Phrenology was one of the most fascinating forms of pseudomedicine in the 19th century, much like homoeopathy. It took a scientifically valid foundational premise, that different areas of the brain have different functions, and then went on to depart liberally from science through buses of creative extrapolation leaving a Mofussil bus terminus. For instance, professional phrenologists would diagnose problems of the mind by merely looking at the shape of a patient’s skull.

While we do not know if Osten practised his skills with Hans’ graceful head, we do know that he managed to “teach” the horse to add, subtract, multiply, divide, work with fractions, tell the time and manipulate calendars.

You could ask Hans an arithmetic question and he would use his foot to tap the answer out. “Hans, what is 11 minus 5?” and he would tap 6 times. “Hans, what day is next Friday?” and he would tap the number of times that represented the date in question.

Understandably, people were astounded. Here was an elegant beast of burden, known for utmost docility and a preference for snacking on apples, demonstrating that humankind’s estimation of the cognitive capabilities of animals was utterly wrong. A sudden rush of “OMG these animals are even closer to human beings than we assumed” guilt washed over his audience, all of whom got to see him for free because Von Osten wasn’t interested in making money with der Kluge Hans (Clever Hans).

He became a nationwide phenomenon in Germany and was even featured in the New York Times.

Understandably, all this public attention got the interest of the German board of education, which appointed a panel of 13 experts, ranging from Circus trainers, psychologists, veterinarians, cavalry officers and school teachers led by philosopher Carl Stumpf (pronounced while attempting to vomit and saying Stoomp at the same time). They were tasked with investigating Von Osten’s claims and determining if he was committing any fraud.

Was he perhaps signaling the answer to the horse so that it knew when to stop tapping?

No. They used different people to ask Hans questions and he aced the answers as usual.

Perhaps Von Osten was using a third party hidden among the spectators to signal to the horse?

No. The horse was able to answer questions even when he was alone with a questioner. He was even able to answer questions that were written down by hand!

The panel was stupefied. They were almost about to declare Hans the Kalki avatar of Vishnu when psychologist Oskar Pfungst (pronounced by trying to spit out mouthwash while saying Fungst), who was presented with the results of the investigation decided to redo some of these tests. And once he put a certain amount of scientific rigour into the experimentation, he came to this most astonishing conclusion.

Clever Hans was not a fraud. He was indeed capable of answering math questions. But, he was able to do so only if the questioner knew the answer to the question he was asking. When written questions were given to a person who had to show it to the horse, but wasn’t aware of the question himself, Hans almost always got it wrong.

Pfungst eventually concluded that Hans was picking up tiny, unconscious body language cues from his questioners. When the number of taps got close to the correct answer, Hans was able to sense tension in his questioners posture and slowed his tapping down. And when he had tapped the right answer and waited a bit, the questioner would declare that he had got it right and applaud or nod his head, so Hans wouldn’t need to tap any further.

So, with that, we quickly reassured ourselves that our math skills were exclusively ours and no apple munching Orlov Trotters were going to take over the planet anytime soon.

But, we got the conclusion wrong.

Of course, Clever Hans wasn’t a mathematical genius, but that’s not the important conclusion. Can you imagine what an incredible feat it was to be able to observe body language cues with such amazing precision to pull off what almost feels like psychic mind reading?

We weren’t overestimating animals’ mathematical abilities. We were grossly underestimating their observation skills.

Pfungst eventually went on to term it the Clever Hans Effect, essentially an Observer-Expectancy fallacy where psychological experimenters tend to influence test subjects with their own cognitive biases, and then come to wrong conclusions based on that.

And now think about this – so much of human communication is non-verbal, and we’ve evolved to both send and receive these unconscious cues when we talk to fellow human beings in person. So when we disagree with someone online, we do it on the basis of literally trying to watch a movie where we only get to see the subtitles translated from another language, a black screen and no audio. So the more we argue and debate by typing on smartphones, the less we are using our amazing evolutionary ability to read a fellow human being beyond the words coming out of their mouth.

In short, you are far more likely to give someone the benefit of doubt in person than online.

And on top of all of this, if what you are getting angry about is a news report where a third party is cherry picking quotes and then extracting the most clickbaity part of what he/she thought someone said and then posting it on social media, I urge you to close your eyes, imagine a magnificent gray horse named Hans, munching on apples and tapping out the answer to “How many more days left before humanity descends into a dystopia where we completely stop talking in person and only fling passive aggressive quote replies at each other on social platforms?”