Judgments of Facial Expression and Suggestion By Herbert Langfeld

In an experiment upon the judgment of emotions from facial expressions in which the subjects attempted to name the emotion depicted in a number of photographs, it was noticed that it was at times possible under suggestion to read various emotions in the face. An experiment was therefore arranged to determine to what degree one is open to suggestion when reading character. It was also the aim of the experiment to bring out, if possible, individual differences in the power of interpretation of facial expression and in suggestibility. The same pictures were used as in the former experiment. Four of the pictures are reproduced below.


Mild Scorn

Rage With Fear

Significant Smile

These were modified photographs of a talented actor, which appeared in the book by Rudolf entitled 'Der Ausdruck des Menschen.' One hundred and five of the pictures were selected as being the best for the experiment. They covered a wide range of emotions and moods. A few of bodily pain and of the sensations of smelling and tasting, and so forth, were included. Miss Grace Speir conducted the experiment, and tabulated the results. There were five subjects, who were all either members of the advanced experimental course or graduate research students.

The experiment extended throughout the second semester of 1916—1917. Each subject came for one hour a week, and as many pictures as possible were presented in that hour. The subject was shown the picture, and asked to write, down his judgment of the expression. After he had done this, he was told either the artist's title of the picture, or an incorrect title, such as 'inspiration' for a picture of 'distrust,' and asked whether he agreed with this title. Some of the incorrect titles were as opposed as possible to the correct title; others were rather similar. The series of one hundred and five pictures was gone through twice. At one presentation the subject was told the correct title, at another presentation the incorrect. On some pictures the correct title was given at the first presentation, - on others at the second presentation, so that the subject never knew, even if he did suspect the purpose of the experiment, whether a right or wrong title was being suggested.

The results are contained in Table 1. The various groups of emotions used are shown in the first column. In the second column is the number of tests for each group of emotions. As there are five subjects, these numbers must be divided by five to give the number of pictures in a group; for instance, the scorn-contempt group has 80 judgments and 16 pictures. The third column shows the number of times the subjects approximated the actual title of the book.

Table 1

An approximated title was taken as a correct judgment because it could not be expected that the subject would frequently use the exact words of the author in describing the picture, so that 'approximated' means 'equivalent.' Throughout this table, the data are presented both in actual amounts and percentages. In the 80 judgments of the scorn and contempt group, for example, 29, or 36%, were correctly approximated. It will be seen that the laughter group was the most readily interpreted, and that amazement comes next. Scorn and contempt, misgiving, aversion and hate, disgust, anger and rage, anxiety, fear, and terror, all of which are more or less related, are interpreted with about the same accuracy. Inspiration, covetousness, wicked and ill-tempered, begging and entreating were poorly interpreted. It was not expected that the sensation groups would give so low a percentage of correct judgments. In this group were also included pictures of sneezing and yawning, and even these were at times incorrectly named. Of the entire 525 judgments a little over a third were correct.

In the fourth column is the number of book titles approved of those which had been already approximated, that is, if we take the scorn-contempt group, the artist's title for 28, or 96%, of the 29 pictures whose titles had been approximated was approved when subsequently shown by the experimenter. These figures give us a check upon the cleverness of the artist in portraying the desired emotions, and it is the data in the fourth column that are used as a basis in determining the suggestibility. A comparison of the totals in columns three and four shows that of the total of 190 such approximations only seven of the artist's titles were not approved when subsequently shown.

The figures in the fifth column show the degree of suggestibility for the various emotions. In the first group, 28 titles were approved. When the wrong titles came to be suggested for these same 28 pictures (after an interval of a month, on an average) six of these wrong titles were accepted. Of the 183 pictures whose titles had been already approved the suggested wrong title was accepted 34% of the time-Excluding the inspiration and begging-entreating groups, which had only one and two judgments respectively as a basis for calculating the suggestibility, the distrust group offered the greatest opportunity for the effect of suggestibility, and the aversion-hate group was next. With anger and rage, suggestion had no effect.

The sixth column shows the number of times the subjects were unsuccessful at approximating the picture. These numbers are complements of the figures in column three.

Column seven gives the number of artist's titles which were approved of those pictures which the subjects themselves had previously not approximated. For instance, in the first group, of the 51 pictures which had not been approximated, the subjects approved 41 of the book titles. As was to be expected, fewer of the artist's titles were approved of these non-approximated pictures than of the approximated ones; in all only 61%, as against 97% of the approximated titles. This drop is indicative of the artist's failure to reproduce the expression he desired. The pictures that were not approximated, even though the artist's titles were approved, gave more room for suggestion than the approximated pictures. 49% of the suggested wrong titles of the 214 approved, non approximated titles were accepted as against 34% of the 183 approved and approximated. This means that if the subject had not himself judged the title correctly, even though he agreed with the title that the artist gave, he was more open to subsequent suggestion than when he had judged the title correctly in the first place. Distrust is again high in suggestibility. The suggestibility in amazement is as high as in distrust, and in laughter almost as high.

In the next to the last column is the sum of all the book titles which are approved whether they had previously been approximated or not, and is obtained by adding columns seven and four. In the last column is the total amount of suggestibility with the pictures from which the figures of the previous column were obtained, that is, the total amount of suggestibility whether the titles had previously been approximated or not. Of the total number of 397 titles approved, 153 or 41% offered opportunity for suggestion.

Table 2 shows individual differences both in the ability to judge emotions and in suggestibility. In the first horizontal line are the five subjects A, B, C, D, and E.

Table 2

In the second horizontal line are the number and percentage of the book titles approximated by each subject. Subject A is the best and subject E the worst in correctly interpreting facial expressions. In order of merit they rank A, C, B, D, E.

The third horizontal line shows the number of book titles which had been previously approximated, and which were approved by the various subjects. For example, A approximated 58, or 55% of the total number. This column offers no new facts, but is used as a basis for calculating the amount of suggestibility which is shown in the next column.

Here it will be seen that subject E is the most suggestible having accepted 47% of the titles suggested with pictures whose correct titles he had approved; that is, when he was given incorrect titles in connection with the 17 pictures whose correct titles he had previously approved, he accepted eight of them. Subject D was the least suggestible. The ranking of the subjects in suggestibility is E, B, C, A and D.

In the fifth horizontal line is the number of book titles which the subjects did not approximate. This is the complement of the results in the second horizontal line. Therefore the ranking by subjects is the reverse of that of the second horizontal line.

In the sixth horizontal line is the number of book titles which were approved of those which had previously not been approximated. It is seen that subject E is the least discriminating, accepting almost every title shown him, that is, 80 or 90% of the titles suggested for the 87 pictures, previously not approximated. In this group of pictures whose titles were not approximated there are, of course, a great many pictures which very poorly portray the emotion intended. Undoubtedly, when the book titles were subsequently given, some of these which are really inappropriate were accepted through suggestion. If we rank the subjects according to the number of those book titles which they accepted, we shall see that the order closely correlates with that of the ranking according to suggestibility; the most highly suggestible accepting the most titles, and the least suggestible accepting the fewest. Inasmuch as subject E is also the most suggestible to wrong titles, it may be said that he accepts almost anything given him whether right or wrong.

The seventh horizontal line shows the amount of suggestibility to these non-approximated titles according to subjects. It will be remembered from the first table where the results of the five subjects were averaged, that there was more suggestibility in the cases where the book title was not approximated. The figures of this line show that this is so in the case of each individual subject. The ranking according to suggestibility is the same as that previously shown with the approximated titles, being subjects E, B, C, A, D.

The last two horizontal lines present the average of the results of the approximated and non-approximated groups and offer no new facts.

It may be said that by this experiment decided individual differences have been shown among the five subjects in regard to suggestibility and ability to read facial expression. It is evident that subject E has very little ability in reading faces, and is highly suggestible, accepting 65% of the wrong titles shown him, as well as 92% of the book titles, some of which were decidedly poor according to the other subjects.

End of Article

Go back to the main Body Language Articles page

Go From Facial Expressions Back To The All About Body Language Homepage


Michael Jackson

Interesting Psychology

Body Language

Criminal Profiling

Psychology Student Guide

Behavioral Science