Emotional Facial Expressions in Advertising

Peter Weinberg, University of Paderborn
Franz-Josef Konert, University of Paderborn
ABSTRACT - Market research and communication strategies require standardized methods to measure emotional consumer behavior. The empirical study reported on here is devoted to the question of whether or not the operationalization of criteria for the mimic depiction of emotions may lead to a more valid evaluation than histrionic intuition alone. If so, such a catalogue of criteria will provide testable indicators for the depiction of emotions in advertisements.
[ to cite ]:
Peter Weinberg and Franz-Josef Konert (1984) ,"Emotional Facial Expressions in Advertising", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 607-611.

Advances in Consumer Research Volume 11, 1984      Pages 607-611


Peter Weinberg, University of Paderborn

Franz-Josef Konert, University of Paderborn


Market research and communication strategies require standardized methods to measure emotional consumer behavior. The empirical study reported on here is devoted to the question of whether or not the operationalization of criteria for the mimic depiction of emotions may lead to a more valid evaluation than histrionic intuition alone. If so, such a catalogue of criteria will provide testable indicators for the depiction of emotions in advertisements.

The results confirm the possibility of defining specific emotions by means of criteria for the mimicry in advertisements. Thus, histrionic intuition can be supported by the pre-setting of such criteria. The technical advertising experience is being embedded in empirically verified findings of consumer research in order to improve the briefing for pictorial information processing.


Market research and advertising have not yet been provided with standardized methods and comparable procedures For the valid and reliable observation and presentation of the expressive emotional behavior in buying decisions or in advertising messages. The following study shows how one may proceed methodologically in order to give consideration to consumer emotions in the marketing concepts of explanation and influencing (Weinberg 1981).

Future market research must therefore devote itself more strongly than before to the analysis of non-verbal behavior in the marketing area. In doing so, special interest is placed on the non-verbal measurement of emotions by facial and body languages - on the one hand, during the consumers decision- process and, on the other hand, during the influencing process of advertising.

A growing importance is being assigned to the emotional influencing of consumers by means of advertising on fully developed and saturated markets. This importance is being met by designing the advertising media in accordance with findings of the social sciences. Modern advertising practices employ a multitude of proven social techniques (Kroeber-Riel and Meyer-Rentschel 1982, pp. 49).

In most strategies of emotional influencing by advertisements, emotions are communicated visually. It is a known fact, that, when compared with verbal information, the acquisition and processing of pictorial information has essential advantages:

- Pictures can emotionalize more strongly than texts.

- Pictures can say more than texts and are more complex.

- Pictures are observed before texts, are then learned better and also remembered longer.

These findings and experiences suggest that -within advertising- the visual communication of emotions should be intensified. Practical approaches for this are provided by the research findings on non-verbal communication (Scherer and Wallbott 1979).


Summarizing, (Weinberg 1983) it may be said that facial expressions can be understood as distinct indicators of emotions. It is known that the muscles of the face permit a great number of distinguishable possibilities of emotional expressions.

Ekman (1973) assumes that there are genetic programs for emotions which also influence emotional expressions. They determine the interplay of facial muscles in specific emotions and are fairly similar with all people. Cultural differences have led to convent ions if feelings may be shown or must be masked.

A lot of research has tried to systematize the distinguishable emotions in facial expressions. Despite semantic inaccuracies and methodical shortcomings, evaluation or decoding studies frequently were able to discern seven categories of emotions for which the relevant literature has synonyms (Ekman, Friesen and Ellsworth 1974, p.59).

Other investigations (Weinberg and Gottwald 1982) showed that in addition to a multitude of dimensions of emotion, the direction and the intensity of emotions can be read from the face. In the component or encoding studies, the face is segmented into individual areas, and one attempts to interpret each facial component separately. The most important approaches appear to be the FAST-Method (Ekman, Friesen and Tonkins 1971) as well as the FACS-Technique (Ekman and Friesen 1976).

Nevertheless, it is still unclear whether different feelings are expressed from different areas of the face or whether a feeling is reflected in the total face. A component analysis can give detailed information for the representation of emotions in advertisements. However, the determination of emotions involved in the buying decision will be easier by using known categories. Ekman et al. (1974 and 1978) have concurrently determined the following emotions as categories or components: happiness, surprise, anger, sadness, disgust and fear. A multitude of linguistic synonyms is known for these possibilities of mimical expressions.


Hands, head and feet may express various gestures. It appears appropriate to distinguish between such gestures which are connected with speech and such which the individual relates to himself. Gestures of the second type also serve the expression of emotions (Argyle 1979, PP. 946).

For the differentiation of emotional qualities gestures of the hands are not so suitable as mimical expressions. They more likely provide information on the intensity of emotion (activation) of the individual.

In sum, it is advisable to use gesturing for the interpretation of verbal and mimical indicators in the analysis of buying behavior as well as in the presentation of persons in advertisements. In both cases it must be tested whether they correspond with the affective and cognitive message. Verbally and mimically expressed emotions are linked mentally, and the movements of the hand which accompany them must be perceived as appropriate and adequate.

The posture of the body can be divided into three basic positions: "sitting, standing and lying "; it expresses emotions, attitudes and status relationships of the individual. The orientation of the body characterizes the positioning of the body towards a partner in interaction. Finally, the movement of the body is understood as the space-time changes in which the entire body is involved (Scherer and Wallbott 1979. D. 146).

In the narrow sense of the word, these three aspects of body language can't be separated from one another in the analysis of the consumer decision processes and in the presentation of consumers in advertising. Interest focusses on body posture, with or without interactive partner, in longitudinal as well as cross-sectional analysis. Thus, orientation and movement of the body characterize situational conditions for body posture.

Ekman (1965) and Ekman and Friesen (1967) determined empirically that the head and facial expression express more the type of emotion, whereas its intensity is more likely conveyed by body posture. Apparently strong emotions can be read from conventional body postures.

Argyle (1979, pp. 261) believes that the entire body may express specific feelings. In his opinion, movements of the body can also be viewed as "extensions" of gestures. In that case, both underline verbal communication.


The experimental study reported here is devoted to the question of whether the pre-setting of criteria for the mimical presentation of emotions may lead to more valid interpretations than histrionic intuition. Such a catalogue of criteria will provide testable indicators for the representation of emotions in advertisements.

Kroeber-Riel (1980, pp. 55) defines emotions to be subjectively perceived arousals. If one follows this concept of emotion, it appears appropriate to distinguish three components of emotions for measurement:

- Intensity of emotion: This is the strongness of the inner arousal, i.e. the objectively present, physiologically measurable activation.

- Direction of emotion: Emotions are experienced as pleasant or as unpleasant, i.e. in accordance with the plus/minus sign of their direction. The direction of the emotion may depend on the intensity of emotion.

- Quality of emotion: Emotions are perceived subjectively and, depending on linguistic usage, identified in manifold ways. Here, the question is that of the type of emotions. Qualities of emotions correspond with the directions of emotions.

To measure these components of emotions there are basically three possibilities:

- Recording physiological indicators

- Collecting self-perceptions by means of interview

- Observation of motoric behavior.

The interpretation of physiological, verbal and motoric indicators requires a prior explanation as to which emotions may be recorded at all. Help in this instance is provided by Izard (1980, pp. 106) and Plutchik (1980. pp. 139).

Both of them found seven concurring "basic emotions": interest or expectation, joy, surprise, sadness or sorrow, anger, disgust and fear. In addition, there are numerous mixtures of emotions which can be composed from these basic emotions.


Aims of the Study and Origin of Data

Advertising agencies are confronted daily with the task of translating emotions into publicity. This also involves the mimical presentation of specific feelings or blends of emotions, for example for emotional benefits which are to be communicated by advertisements.

If specific emotions are pre-set in a briefing it will depend particularly on the histrionic capability of the actor as to whether the intended emotion is communicated or not. For this reason, it is to be tested experimentally if greater agreement can be reached in the evaluation of emotional facial expression when a catalogue of criteria (Weinberg 1983) is used for the presentation of emotion.

In June 1982, a six-day decoding experiment using an categorical experiment was conducted at the University of Paderborn. The experimental stimuli were the three emotions (surprise, anger and joy) which were represented in the catalogue of criteria. The emotions were to be presented by participants in the experiment and not by actors since literature reports (Ekman, Friesen and Ellsworth 1974, p. 27) that there is more likely a tendency to take a "posed" facial expression for a spontaneous one if this emotion is presented by untrained Persons.



At the first meeting (control group) the individuals were asked to be seated in front of a (white) film screen. After a certain adaption phase, they were informed that in a few moments they had to express a specific feeling by using their faces. A story was read which contained a certain emotion that was to be shared by the participants and represented by facial expression (i.e. surprise, rage or joy)

At the second meeting (experimental group), which took place several days later, the encoders had to present the same emotion. The difference to the first meeting was that this time they were confronted with specific pre-set data (from the catalogue of criteria) as to which mimics were to be used in presenting the emotion. The following instructions were selected from the above catalogue:

on Surprise: mouth slightly opened, raised eyebrows, eyes opened widely.

on Rage: staring look, tightly closed mouth, vertical forehead wrinkles, eyebrows contracted.

on Joy: broad closed smile, narrowed eyelids, dreamy look. .

Several photos were made of each person and were later evaluated by neutral persons. They had to select that slide front the experimental as well as from the control group which best represented the intended emotion.

Students from the University served as untrained observers of the selected photos. The control and the experimental group (40 persons each) were equally distributed under several characteristics (sex, subjectively perceived ability to recognize emotions, personality traits).

"Agreement" among the raters required a special operationalization. Since the emotions of "joy, surprise, anger" dispose of linguistic synonyms, 7 synonyms could be pre-set for each category of emotion (Schmidt-Atzert 1980, pp. 34, Ekman, Friesen and Ellsworth 1974, ?. 52). The measures used for agreement were:

- Strong agreement (value 1)

Upon the question of which concepts correspond best to the emotion shown, the test person names only one concept and this concept is identical to the category represented on the slide.

- Middle agreement (value 2)

The test person names, by order of rank, several concepts which all stem from the category represented on the slide.

- No agreement (value 3)

The test person names at least one concept which stems from a category not represented on the slide.

Selected Findings for the Interpretation of Mimical Expression

The testing of the hypothesis was done by means of three one-way analyses of variance. The independent variable was the respective facial expression presented in the experimental and control group. The dependent variables were the concurring agreements of the observers. In the below tables, those slides are listed with which the experimental stimulus exerted an influence in the sense o f the hypothesis.

Since agreement was measured by means of a transformation, the findings were checked additionally by contingency analysis.



After finishing of the experiment a similarity test of the 12 stimuli (individually for the experimental and control group) was conducted on the basis of a pair comparison. 60 students (experimental group = 30, control group = 30) were questioned as to the similarity of the emotions presented in the face with those shown on the respective slide. This aimed at:

1. It should be clarified whether three different emotions had been presented, and, if so, this had to be confirmed by three homogeneous segments.

2. Further, it was of interest whether these segments differed from each other within the two groups. In this context, the distances between the stimuli in the evaluation within the experimental and the control group were of particular importance.

The analysis took place on an individual as well as on a compressed data basis. The following remarks refer to the MINISSA model, which is the formal equivalent of the model developed by Kruskal.

For the graphic interpretation the two-dimensional solution is shown. Figures in the respective plot-prints have the following meaning:

1, 2, 3, 4: Emotions from category "Surprise"

5, 6, 7, 8: Emotions from category "Joy"

9, 10, 11, 12: Emotions from category "Anger"







On MINISSA Positioning:

The control group contains three clusters of the respective categories of emotions which, however, are marked by relatively large distances from each other. In the experimental group these distances are much smaller with the significant slides. Calculations were done on the basis of metric parameter 2 (Euclidean Distance). It is common to interpret the distance between two objects in models of MDS as an indicator of their similarity. In the foregoing case this points to a greater similarity of the slides in the three emotion categories of the experimental group than of those in the control group

The results of the experiment show that the ability to express by the face varies. Thus, in the categories of "surprise" and "anger" the agreement of the evaluators could be improved with 3 out of 4 slides. For "joy" this was the case only for one slide. For this emotion the agreement in evaluation can probably be improved most if gesturing is taken into account (Hypothesis; emotion "joy" + hands and arms raised = emotion "enthusiasm"). Consequently, the research design will have to be expanded so as to include additional signal classes in order to standardize the pre-setting of criteria.

Content validity of the experiment was assured by randomization. Testing of the validity of the variable "agreement" was done by means of criterion validation since the variable "agreement" was identically operationalized for the free and the pre-set answers (1 = strong agreement, 2 = middle agreement, 3 = no agreement). The following table shows the significant contingency coefficients (for X < 0.05) between the two variables.



Testing of reliability was done according to the variable "date of participation in the experiment" which was dichotomized. No significant differences could be determined on the 5% level (analysis of variance). Using an additional analysis of variance, it was tested whether the sex of the actors and the raters exerted a significant influence.

No significant effects could be determined on the 5% level. As was already explained that the sequence of the slides varied. Since this variable was coded, it could be tested by means of a t-test (a < 0.05) whether significant differences resulted. That was not the case with the remaining slides.


When summarized, the above results confirm the possibility of defining specific emotions by criteria for facial expressions in advertisements. Histronic intuition can thus be supported by pre-setting such criteria. Experience in advertising management is thus embedded in empirically tested findings in order to set a better briefing for the presentation of emotions

In the future, emotional advertising will gain importance, especially in saturated markets of fully developed products. Strategies of emotional conditioning can satisfy the need for sensual product benefits. For this reason, such a system of categories for the advertising design of emotions by means of non-verbal indicators may be able to provide a valuable contribution.


Argyle, Michael (1979), Korpersprache und Kommunikation (Bodily Communication), Paderborn: Junfermana.

Ekman, Paul (1965), " Differential Communication Affect by Head and Body Cues," Journal of Personality and Social Psychology, 2, 726-735.

Ekman, Paul (Ed.) (1973), Darwin and the Facial Expression: A Century of Research in Review. New York: Academic Press.

Ekman, Paul and Friesen, Wallace V. (1967), "Plead and Body Cues in the Judgement of Emotion: A Reformulation," Perceptual and Motor Skills, 94, 711-724.

Ekman, Paul and Friesen, Wallace V. (1976), "Measuring Facial Movement," Environmental Psychology and Nonverbal Behavior, 1, 56-75.

Ekman, Paul and Friesen, Wallace V. (1978), The Facial Action Code: A Manual for the Measurement of Facial Movement. Palo Alto: Consulting Psychologist Press.

Ekman, Paul, Friesen, Wallace V. and Tomkins, Silvan S. (1971), "Facial Affect Scoring Technique: A First Validity Study," Semiotica, 3, 37-58.

Ekman, Paul, Friesen, Wallace V. and Ancoli, Sonia (1980), "Facial Signs of Emotional Experience," Journal of Personality and Social Psychology, 39, 1125-1134.

Frois-Wittmann, Jean (1930), "The Judgement of Facial Expression," Journal of Experimental Psychology, 13, 113-151.

Harrison, Randall P. (1974), Beyond Words: An Introduction to Nonverbal Communication, Englewood Cliffs: Prentice Hall.

Izard, Carroll F.. (1977), Human Emotions, New York: Plenum Press.

Kroeber-Riel, Werner (1980), Konsumentenverhalten. 2. Auflage. (Consumer Behavior. Second Edition), Munchen: Vahlen.

Kroeber-Riel, Werner and Meyer-Hentschel, Gundolf (1982), Werbung. Steuerung des Ronsumentenverhaltens, (Advertising. Strategies for Influencing Consumer Behavior), Wurzburg-Wien: Physica.

Plutchik, Robert (1980), Emotion: A psychoevolutionary Synthesis, New York: Harper & Row Publishers.

Scherer, Klaus R. and Wallbott, Harald G. (Eds.) (1979), Nonverbale Rommunikation. Forschungsberichte zum Inter-aktionsverhalten, (Nonverbal Communication. Research of Interaction Behavior), Weinheim-Basel: Beltz.

Schmidt-Atzert, Lothar (1980), Die verbale Kommunikation von Emotionen. Eine Bedingungsanalyse unter besonderer Berucksichtigung physiologischer Prozesse, (Verbal Communication of Emotion. An Analysis under Consideration of Physiological Processes), Dissertation GieBen.

Weinberg, Peter (1981), Das Entscheidungsverhalten der Konsumenten, (Consumer Decision Behavior), Paderborn: Munchen-Wien-Zurich: Schoningh.

Weinberg, Peter and Gottwald, Wolfgang (1982), "Impulsive Consumer Buying as a Result of Emotions," Journal of Business Research, 10, 43-57.

Weinberg, Peter (1983), Beobachtung des emotionalen Verhaltens. Innovative Marktforschung. Tirsg. von der Forschungsgruppe Konsum und Verhalten. (Observation of Emotional Behavior. Innovative Marketing Research. Ed. by Forschungsgrunpe Konsum und Verhalten), Wurzburg-Wien: Physica, 45-62.