The year is 1904. The spectators hold their breath in anticipation as a horse, Hans, stamps his hoof on the blackboard—on which the whole alphabet was written—in front of him, picking out the letters H, N, and A. The audience sighs in relief. Dohna—the name of the officer standing before Clever Hans and his owner Herr Wilhelm von Osten, and to whom they have been introduced earlier. Von Osten reaches out his hand to Hans to give the horse a treat.
Everyone claps, but they look in bewilderment at each other as if to ask—how? Von Osten claims that he taught Hans just as any person might teach a child: Hans is instructed to “count” the number of balls by tapping his hoof—at the same time, von Osten counts out loud with him. Thus, supposedly, Hans associates the auditory signal provided by von Osten with the visual signals, and the hoof taps are a way for Hans to communicate the visual signals he perceives.
Even philosopher-psychologist Carl Stumpf, who formed the Hans Commission in order to investigate the truthfulness of the phenomenon in 1904, concludes that Hans is, truly, capable of performing mathematical operations such as addition, fractions, and even square roots. Three years later, Professor Oskar Pfungst, a psychologist, studies the Orlov Trotter closely in his laboratory, and comes upon this conclusion: Clever Hans is clever, indeed, but not in the way his admirers, and even his trainer, know.
Although Hans could generally answer the questions posed by people other than von Osten, the accuracy of his answers drops when the questioner was far, or completely obstructed from Hans’s view, and when the questioner did not know the answer to the question beforehand. Upon even closer investigation, Pfungst realizes that Hans derives his answers from the very subtle body language of his questioner. When Pfungst asks Hans a question, the psychologist looks down to observe the number of stamps the horse makes with his foot.
When Hans arrives at the correct number of stamps, the researcher is satisfied and looks ever so slightly away, unwittingly signaling to the horse that he has arrived at the correct answer. The year is 1907, and Professor Oskar Pfungst discovers what will be known as the Clever Hans effect. *** It is now 89 years later. An experimenter observes a subject walk along a hallway. He presses the stopwatch; it reads at around 8 seconds.
He catches up with the participant and tells her the complete objective of the experiment—to see if participants primed with stimuli relevant to a stereotype would unconsciously embody the stereotype in some way. Minutes ago, this participant was made to manufacture a sentence out of a set of words given to them. Some of the words in this sentence were relevant to the elderly stereotype, such as “wise” and “wrinkle”. In some participants’ sentences, the words were completely neutral, age-wise.
The experimenter tells the participant that they are investigating language proficiency in their participants. After the experimenter (partially) debriefs and thanks the participant for their time, a second experimenter takes note of the amount of time it takes for the participant to walk down the corridor—specifically, 9. 75 m away from the doorway of the experiment room. As expected, the experiment appears to have significance—the participants who have been exposed to the elderly-stereotype-relevant words walk for an average of 8. 8 seconds—significantly more than those who read age-neutral words, who walk for an average of 7. 30 seconds. A replication of the experiment shows the same result: those who are primed walk an average of 8. 20 seconds, while those who are not walk an average of 7. 23 seconds. The year is 1996, and John A. Bargh, Mark Chen, and Lara Burrows have just shown that priming effects can still occur even when the subject is not aware of its presence. *** The setting appears to be taken straight from Bargh et al. ’s study, but it is 2011.
An experimenter calls to the participant who is just about exit the building, says that he has forgotten to administer one final task. He poses questions designed to measure the participant’s awareness of the study’s manipulation, and awareness of the similarity of the current study to the 1996 study. The participant shakes her head: no, she had no idea that she was expected to have walked slower, and, no, she had not heard of a similar study. Stéphane Doyen, Olivier Klein, Cora-Lise Pichon, and Axel Cleeremans might be replicating Bargh et al. s study, but there are still some many remarkable differences between the two studies—for one, Doyen et al. ’s study has twice as much participants as Bargh et al. ’s. Secondly, while Bargh et al. manually measured the time it took for the participants to walk to the end of the corridor, Doyen et al. use infrared sensors to measure their participants’ walking speeds, both coming to and from the experiment room. But of course, the most outstanding difference between the two is this: the primed subjects of the 2011 walk no slower than the non-primed subjects.
Doyen et al. speculate two possible causes for the differences: first, the differences may be because of the manual time-taking method in Bargh et al. ’s study. The second cause—and perhaps the more controversial—is that Bargh et al. ’s experimenters had some expectations regarding their subjects’ performances and had, unwittingly, adapted their behavior accordingly, affecting, in turn, their participants’ behavior. In short: Bargh et al. ’s experimenters and participants might have been Clever Hans-ed. ***
It is still 2011, except now an experimenter once again takes note of the time it took a participant to walk to the end of the hall, much like in the 1996 study. The experiment is a success: the hypothesis is significant, and the data mimics that of Bargh et al. ’s. But unlike the previous studies, Doyen et al. manipulate not only their 50 participants, but also the 10 experimenters they have hired. The experimenters are briefed: half are told that the prime would cause participants to walk faster; the other half, slower.
According to the time taken manually by the experimenters, those who are primed walk significantly slower than those who are not—but if and only if the experimenter believes that the primed subjects should walk slower. If the experimenter is made to believe that the primed participants should walk faster, the data reverses itself: the primed subjects, indeed, walk faster than the non-primed subjects—that is, only if one takes the manual (subjective) timing into consideration: the fast-briefed experimenter is less accurate in timing the primed participants compared to the rest.
The slow-experimenter-slow-subject result also appears in the results, even when the time reading used is the reading measured by the infrared sensors; however, when the experimenter is given the fast briefing, the infrared sensors indicate no significant difference between the speeds of the participants. In general, however, the participants assigned to the slow experimenter, regardless of whether they are primed or not, walk significantly slower than those assigned to the fast experimenter. *** It is 2012. The psychology (and psychology enthusiasts) community is abuzz.
Eyebrows raise in response to Bargh et al’s experiment’s failure to replicate in most studies, and its successful replication in Doyen et al’s deliberate Clever Hans-like manipulation. Ed Yong, a blogger at Discover Magazine, however, does not completely discount the possible validity of Bargh et al’s experiment. Yong points out that Doyen et al’s experiment does, in fact, show that their participants were primed—not by the age-relevant words presented in the scrambled word task, but by the experimenters’ behavior.
Indeed, Doyen et al’s experiments show that their subjects may have been primed to be biased towards old age. In the debriefing part of Doyen et al’s experiments, participants were made to choose between four pictures: a picture of an athletic person, a picture of a handicapped person, a picture of an Arabic person, and a picture of an elderly person. Doyen et al’s experiments—including even the one where they do not manipulate the experimenters’ behavior—show that the primed subjects preferred the picture with the elderly person too frequently for it to be considered chance, while the non-primed subjects did not.
Thus, it is possible that the 2011 participants were primed—and its effects were manifested in a different manner than expected, leaving Bargh et al’s findings to be still, at the end of the day, plausible. References (and additional reading): Bargh, J. A. , Chen, M. , & Burrows, L. (1996). Automaticity of social behavior: The effects of trait construct and stereotype activation on action. Journal of Personality and Social Psychology, 71(2), 230 – 244. Bellows, A. (2007). Clever Hans the math horse. Damn Interesting. Retrieved from http://www. damninteresting. com/clever-hans-the-math-horse/ Bower, B. 2012). The hot and cold of priming. ScienceNews. Retrieved from http://www. sciencenews. org/view/feature/id/340408/description/ The_Hot_and_Cold_of_Priming Doyens, S. , Klein, O. Pichon, C. L. , & Cleeremans, A. (2011). Behavioral priming: It’s all in the mind, but whose mind? PLoS ONE. 7(1). Retrieved from http://www. plosone. org/article/ info%3Adoi%2F10. 1371%2Fjournal. pone. 0029081 New York Times, The. (1904). Berlin’s wonderful horse: He can do almost everything but talk—how he was taught.
The New York Times. Retrieved from http://query. nytimes. com/mem/archive-free/ pdf? es=9E02E2D91F3AE733A25757C0A96F9C946597D6CF Yong, E. (2012). Primed by expectations: Why a psychology experiment isn’t what it seemed. Discover Magazine. Retrieved from http://blogs. discovermagazine. com/notrocketscience/2012/01/18/primed-by-expectations-why-a-classic-psychology-experiment-isnt-what-it-seemed/#. UTVgQnyPhRh Yong, E. (2012). A failed replication draws a scathing personal attack from a psychology professor. Discover Magazine. Retrieved from http://blogs. discovermagazine. com/notrocketscience/2012/03/10/ failed-replication-bargh-psychology-study-doyen/#. UTVgQ3yPhRi