Clever Hans — the horse that could not count but could read a face

In Berlin around the turn of the twentieth century, a horse named Clever Hans appeared to do arithmetic, tell time, read, and spell, answering questions by tapping out numbers with his hoof. He could do none of those things. In 1907 the psychologist Oskar Pfungst demonstrated that Hans was not calculating at all but reading tiny, involuntary movements in the body and face of whoever asked the question — cues that told the horse when to start tapping and, decisively, when to stop. The arithmetic lived in the questioner, not the horse. The case is unusual among hoaxes because there was no hoaxer: the owner, Wilhelm von Osten, sincerely believed in his horse’s intellect and was himself among those deceived.

That sincerity is central to the file. Von Osten, a retired Berlin schoolteacher with an interest in animal intelligence, had spent years patiently “teaching” Hans, an Orlov Trotter stallion, using number boards and drills. He charged no admission, sought no fraud, and submitted the horse freely to investigation. In 1904 an official commission of thirteen — assembled under the philosopher and psychologist Carl Stumpf and including a veterinarian, a circus trainer, schoolteachers, and the director of the Berlin zoo — examined Hans and concluded there was no trickery. They were right that von Osten was honest and wrong about what the horse was doing.

The decisive work came from one of Stumpf’s young associates, Oskar Pfungst, who treated the horse not as a wonder to be admired but as a problem to be controlled. Through a careful series of tests published in 1907, Pfungst varied two things the commission had not properly isolated: whether the questioner knew the answer, and whether Hans could see the questioner. The results were stark. When the questioner knew the answer and was visible, Hans was right the great majority of the time; when the questioner did not know the answer, or when Hans could not see them, his accuracy collapsed. Hans was reading people, and could only “know” what his human knew.

Pfungst then identified the mechanism precisely: as the hoof-taps approached the correct count, the questioner unconsciously increased in tension and made a minute postural or facial change, releasing it at the right number, which the horse had learned to use as a signal to stop. The effect was so reliable that Pfungst could make Hans stop on any number by performing the cue deliberately — and found he could not suppress the cue in himself even when he tried. From this came the enduring term “the Clever Hans effect,” now a foundational warning in every science that studies behaviour: the danger of an experimenter unconsciously cueing the result they expect.