Clever Hans (The Horse of Mr. Von Osten) / A contribution to experimental animal and human psychology

by Oskar Pfungst

Three words were thereupon whispered in his ear, which he was asked to spell in accordance with the method described on [page 21]. Since he had to indicate first the row, and then the place in the row occupied by the letter, it took two answers to indicate the position of each letter. I acted as questioner. The ordering of the table of letters was unknown to me, except the position of the letter "a", which naturally came first, and the place of the letter "s", concerning whose position I had purposely inquired. The words chosen for this experiment were "Arm", "Rom" (Rome) and "Hans". The horse responded incorrectly in the case of every letter which was unknown to the questioner. "A" and "s" alone were given correctly. Thus in spelling the word "Rom" the horse responded with the series 3, 4; 3, 4; 5, 4; 5, 4; i. e. "jjst", instead of the correct series: 4, 6; 4, 2; 3, 7. I later selected three other words, the spelling of which involved the tapping of thirty-two numbers on the part of Hans, and whose position I had carefully ascertained beforehand. When these were given to the horse to spell, he responded promptly without a single error. Evidently Hans was unable to spell without assistance of some sort from the questioner.

The horse's reputed aptitude in computation was tested in the following way. Mr. von Osten whispered a number in the horse's ear so that none of the persons present could hear. Thereupon I did likewise. Hans was asked to add the two. Since each of the experimenters knew only his own number, the sum, if known to anyone, could be known to Hans alone. Every such test was immediately repeated with the result known to the experimenters. In 31 tests in which the method was procedure without knowledge, 3 of the horse's answers were correct, whereas in the 31 tests in which the method was procedure with knowledge, 29 of his responses were correct. Since the three correct answers in the cases in which procedure was without knowledge evidently were accidental, the results of this series of experiments show that Hans was unable to solve arithmetical problems.

For the purpose of discovering whether the horse could at least count, the Russian kindergarten device, which Mr. von Osten had used in training, was utilized. The machine was placed before the horse, but the experimenter turned his back upon it. Before each test, a number of balls were pushed to one side and Hans's problem was to indicate the number thus separated. Each test was repeated with procedure with knowledge. Of eight such experiments Hans responded successfully every time procedure was with knowledge but failed every time procedure was without knowledge. Thus 7 balls were at one time designated as 9 and later as 14, while 6 were at first designated as 12, and later as 10. Since all these errors could not be accounted for on the ground of miscounts on the part of the horse, it was evident that Hans is quite unable to count.

The memory-test was conducted in the following manner. In the absence of the questioner a number or the name of some day of the week was spoken to the horse. The experimenter would then return and question him. Of 10 responses 2 were correct, 8 incorrect. Among the correct answers were the number 3, a number which, as we shall see, Hans was prone to give under all sorts of conditions, and which therefore meant very little when given as a correct response. The number 2, on the other hand, was consecutively indicated by 7, 9, 5, and 3, 8 was given as 5, 6, 4, and 6, consecutively; and finally Wednesday was indicated as the fourteenth day of the week. After this we undertook the test the horse's far-famed knowledge of the calendar. Dates, such as Feb. 29, Nov. 12, etc., were given to Hans and he was asked to indicate on which day of the week they fell. Sunday was to be indicated by 1, Monday by 2, etc. Of 14 such tests, 10 were unsuccessful, 4 successful. But in the case of these 4 something very interesting occurred. It happened that during this series the keeper of the horse was present, and he happened to know the days on which these dates fell,—as he himself testified. The dates in question were also little more than a week or so from the day of the experiment, so they could easily be determined. But as soon as we took more remote dates both man and beast were hopelessly lost. It was certain that Hans had no knowledge of the calendar. It is needless to say anything of his supposed knowledge of cards and coins. Hans plainly was incapable of the astonishing feats of memory which had been claimed for him.

Finally we investigated Hans' musical ability. In a room adjoining the horse's stall there was a small harmonica, which spanned the once accented octave. On this one or more tones were played. The horse was required to indicate the tone played, the number of tones played and their relation to one another. For testing his general hearing 20 tests were given in which the method was procedure without knowledge. Of the responses only one was correct, and that one was the tone e, for which the proper response was three taps, but we must bear in mind what has already been said of the number 3. The tone b was indicated by 11 taps, although Hans had only learned a scale of one octave and therefore could respond to only seven tones. In the tests in which the method was procedure with knowledge, he again, without exception, was successful. Similar results were obtained in the analysis of compound clangs. In the cases of procedure without knowledge (although the experimenter here knew the correct responses, he purposely refrained from thinking of them) not a single response was correct; while in the cases of procedure with knowledge, all but one were correct. The following were typical responses: Three tones were played and the question was asked, "How many tones were played?" Hans responded first with 4 taps and then with 1. The tones c, e, g, a, (1, 3, 5, 6) were struck and the question asked, "Which tone must be eliminated to make the complex a chord?" In the tests in which the method had been procedure with knowledge, this question had always been answered correctly, but when procedure was without knowledge the responses were first 13, a tone which does not exist for Hans, then 2, a tone which was not given in the clang to be analyzed, and finally 3, which was not the discordant tone. Hans's far-famed musical ability was an illusion.

Taking the results of all the tests into consideration, we find that in the case of procedure with knowledge, 90 to 100% of the responses of the various series were correct, whereas, in those series of procedure without knowledge 10%, at most, of the responses were correct. Under the conditions prevailing during these latter tests, even these 10% must be regarded as due to chance. To be sure Mr. Grabow, a member of the school board and an enthusiastic follower of Mr. von Osten (Zeitschrift für Pädagogische Psychologie, Pathologie und Hygiene, Berlin, 1904, Jahrg. 6, Heft. 6, S. 470), mentions a large number of successful tests, which were supposedly made in accordance with the method of procedure without knowledge. A thorough analysis of his experiments was not possible, because the conditions under which they were conducted were not adequately specified. But I have no doubt that the successful responses of the horse were due solely to the absence of precautionary measures. I, too, could cite a number of seemingly correct responses which demonstrably were due to the absence of adequate precautionary measures. I therefore repeat: Hans can neither read, count nor make calculations. He knows nothing of coins or cards, calendars or clocks, nor can he respond, by tapping or otherwise, to a number spoken to him but a moment before. Finally, he has not a trace of musical ability.

After all this experimentation it was evident that the horse was unable to work alone, but was dependent upon certain stimuli from its environment. The question therefore arose: does the horse get these stimuli while the question is being put, or during his responses, i. e., during the process of tapping.

If Mr. von Osten's opinion was correct, then the process of questioning played an important part in the success of the experiment. Of course, as he said, it was not necessary to ask the question aloud; it was sufficient—curiously enough—that it be inwardly spoken, thanks to the horse's extraordinary auditory sensitivity. If, however, conditions were made such that the auditory sense was eliminated, then the animal would be unable to respond. Such a theory is not quite as absurd as it might seem at first blush. For Hansen and Lehmann have shown that an acute auditory organ is able to respond to such delicate stimulation as is involved in the softest whisper, or even in the so-called nasal whisper in which the lips are tightly closed.[3] They have attempted thus to explain any modes of supposed "thought-transference", (cf. [page 7]). Since experts on horses agree that the horse has acute auditory sensitivity, Mr. von Osten seized upon this fact and tried to establish his theory in the following manner. No response was successfully made on the part of the horse, he said, when the sound waves caused by his (Mr. von Osten's) inner speech were deflected from the ear of the horse. This was the case when he closed nose and mouth while inwardly putting the question, or deflected the waves from the horse's ear by means of a placard held before his mouth while speaking, or finally by applying lined ear-muffs to the horse's ears. If, on the other hand, he closed only his nose and not his mouth while thus inwardly putting the question, or if he held the placard so that there was a possibility of deflecting the sounds to the horse's ear, or if the ear-muffs were of too sheer a material, then Hans could hear and answer the questions which for human ears were inaudible. He demonstrated all this by means of experiments and of 20 tests of the first kind, in which auditory sensations were supposedly eliminated, 95% of the responses were incorrect (Hans would always tap too great a number); whereas of 28 tests of the second kind, not a single answer was wrong, just as had been predicted. Now I have repeated both kinds of tests, but have always found some correct responses in those cases in which the horse, supposedly, was unable to hear, a thing which greatly astonished Mr. von Osten. In fact, the responses of the horse were quite as correct when I did not even whisper the question inwardly. It was quite clear that putting the question in any form whatever was wholly unnecessary. Mr. von Osten's demonstrations to the contrary, which were based upon erroneous physical principles, are to be explained as cases of vivid autosuggestions, (but of this, more in [Chapter V]). After all this experimentation, it was manifest that the cue was not given to the horse while the question was being put; it occurred, therefore, at some time during the process of tapping. But by means of which sense organ was it received by the horse?

We began by examining the sense of vision, and in the following manner. Blinders were applied, and it is worthy of mention that Hans made no attempt to resist. The questioner stood to the right of the horse, so that the animal knew him to be present and could hear, but not see him. Hans was requested to tap a certain number. Then the experimenter would step forward into the horse's field of vision and would put the same problem again. Since, in the tests of the first kind, Hans would always make the most strenuous efforts to get a view of the questioner, and since he would rave and tear at the lines whenever the attempt was made to tie him,—a thing which he had never done hitherto,—it was impossible to determine in some cases whether or not he had seen the questioner during the process of tapping. I am using, therefore, in the following exposition, besides the two categories of "not seen" and "seen", a third which I have called "undecided". A total of 102 tests were made in which large blinders were used. In 35 of these, the experimenter certainly was "not seen" in 56 cases he was "seen" and the remaining 11 are "undecided". Under the first of these categories 6% of Hans's answers were correct (i. e. only two), under the second head 89% were correct and under the third 18% were right. In other words, the horse was at a loss the moment he was prevented from seeing the questioner; whereas his responses were nearly always correct when the experimenter was in sight, certain proof that the horse's failures are to be attributed to the elimination of visual stimuli and not to the general inconvenience occasioned by the blinders. It is evident therefore, that the horse required certain visual stimuli or signs in order to make a correct response.[H]