Navigation Bar - National Geographic Magazine - Home
Header - Survey 2005: In Your Face Results - Angry, Angrier, Angriest: Can you Tell?


David Matsumoto
San Francisco 
State University

Paul Ekman
The Ekman Group, LLC

James Witte
Clemson University

Roy Pargas
Clemson University
More than a year ago, you participated in Survey 2005, an online global study on discerning human facial expression conducted in partnership with the University of California at San Francisco, San Francisco State University, and Clemson University. The results you’ve been waiting for are in. Learn what researchers have to say about how well we measure anger and why.

Emotions are incredibly important psychological phenomena. They tell us about the relationship between us and whatever triggered them. They prepare our bodies to react in specific ways. Anger, for example, prepares us to fight, while fear prepares us to flee. Our displays of emotion inform others of our internal state and behavioral intentions, evoke emotions in those who witness them, and, in turn, influence their behavior. Emotions, therefore, are evolutionary-based, psychobiological reactions that help us solve complex social problems.

In psychology, it is a well-known finding that a small set of facial expressions of emotion is universal. People all over the world—regardless of race, culture, gender, or religion—express and recognize the emotions of anger, contempt, disgust, fear, happiness, sadness, and surprise in facial expressions. These are known as basic emotions because they are universal and because we share them with our nonhuman primate relatives (who also had to solve social coordination problems).

Because basic emotions are conveyed primarily through facial expressions, much research on emotional expression has focused on the face, the most complex signal system we have. Our facial musculature includes more than 40 functionally anatomical units that can move independently of each other. Additionally, each unit can have its own timing characteristics (in terms of onset, apex, and offset), intensity, and laterality or symmetry. This allows us to produce literally thousands of expressions, depending on the exact combinations of single units that are moving in the face at any one time and their exact timing, intensity, and symmetry profiles.

Despite the fact that the face can produce thousands of different expressions, researchers examining judgments of facial expressions have generally used only full-face, high intensity expressions known as prototypes. Research on facial expression production has shown that each emotion is associated with a core set of facial muscles that are innervated when that emotion is aroused. When all of the core facial units associated with an emotion are innervated together at high intensity, they produce the prototype expression. But in reality, the face produces many variants of each emotional expression, because the core elements occur by themselves in partial expressions, in combination with core elements of other emotions (known as blends), or in combination with other facial actions not associated with emotion.

Survey 2005 Results

Although there have been many studies testing the pancultural recognition of facial expressions of emotions, to date there has never been a study testing the recognition of different variants of basic emotions. The study you participated in through National Geographic Magazine Online a little more than a year ago is the first in the world to do so. Approximately 50,000 individuals responded. All of the expressions you judged were variants of the prototypical expression of anger, that is, they all had some of the core elements of anger. They all differed in certain ways, and there are many ways in which we can examine the responses.

As a first look at the data, we examined how well people did in judging anger when relatively more or less of the prototypic core elements of anger were included in the expression. For example, take a look at the four expressions in Table 1. One of the ways in which they differ is in the number of core elements of anger. The first expression contains the highest number and is the most intensely expressed. The second contains the next highest, the third the next, and the fourth the lowest. The fourth expression, in fact, is a quite subtle version of anger, but the clue for anger is definitely there in the pursing and tightening of the lips. As you can see from the right column of Table 1, the vast majority judged the first expression as angry; this percentage declined in the second, third, and fourth expressions. One of the interesting findings was that only a very small percentage of individuals were able to pick up the clues to anger in the fourth expression.

The results also showed reliable differences in responses based on sex. Women were slightly more accurate in their recognition of the emotion in each of these expressions (Table 1). This sex difference has been well replicated in the psychological research literature. Women are simply more sensitive to these cues than men (although the difference is not very large), even under conditions of minimal stimulus information (for example, when the faces are presented extremely fast, in a blink of the eye). We still don’t know why this sex difference exists. It could be that women around the world are socialized to decode emotions better than men from such an early age that the ability to do so is automatic by adulthood. Or, female brains may be better equipped to decode emotions from birth.

Table 1 - Percentage of observers judging each of these expressions as anger
Title - Expression
Title - Judges Title - % Anger Judgements
Image - Expression 1
Male 69.6
Female 74.2
All 72.5
Image - Expression 2
Male 55.9
Female 59.6
All 58.2
Image - Expression 3
Male 33.8
Female 37.0
All 35.8
Image - Expression 4
Male 6.0
Female 7.2
All 6.8

Title - Table 2 - Percentage of observers judging the two most intense expressions as anger as a function of type of employment
Title - Expression
Title - Type of Employment Title - % Anger Judgements

Image - Expression 1

Professionals (e.g., scientists, teachers, nurses, lawyers) 75.4
Technicians and associate professionals (e.g., engineers)


Legislators, senior officials, and corporate managers 72.9
Clerks (e.g., secretaries, mail clerks, cashiers, tellers) 72.5
Craft and related trades workers (e.g., miners) 72.4
Service workers and shop and market sales workers 71.9
Sales, services, agricultural and construction labor 70.2
Plant and machine operators and assemblers 63.6
Skilled agricultural and fishery workers 61.3
Armed forces 58.8

Image - Expression 2

Professionals (e.g., scientists, teachers, nurses, lawyers) 62.6
Legislators, senior officials, and corporate manager 59.2
Technicians and associate professionals (e.g., engineers) 58.2
Clerks (e.g., secretaries, mail clerks, cashiers, tellers) 56.2
Sales, services, agricultural and construction labor


Service workers and shop and market sales workers 56.0
Craft and related trades workers (e.g., miners) 55.4
Plant and machine operators and assemblers 54.0
Armed forces 53.3
Skilled agricultural and fishery workers 52.5

Another interesting finding in our study concerned differences in recognition ability as a function of type of employment (Table 2). In Table 2, we classified the various types of employment of the respondents and calculated their average accuracy in recognizing the two most intense expressions as anger. The differences were pretty large. These data, however, cannot tell us the source of the difference, only that there is a difference. It could be, for instance, that certain types of jobs—such as being a scientist or a legislator—involve more instances in which it is necessary for individuals to recognize and deal with other people’s anger, and thus those who stay in those jobs get better at recognizing it when it occurs. Other jobs, such as factory workers or military personnel, may not facilitate the development of such skills because it is not as necessary. Or it could be that these latter types of jobs involve recognizing emotions to such an extent that people build a higher threshold of tolerance for recognizing true anger. Perhaps people simply have inborn individual differences in their ability to recognize emotion that is related to some kind of genetically based differences in personality and that people, on average, with these kinds of differences gravitate to certain types of jobs.

All of these possibilities lead to very interesting questions that scientists can pursue in the future, and they are heavily informed by the data from this study. Thank you for doing your part for science.

We are indebted to the assistance of Laurie Margot Ross and Sanae Nakagawa in the preparation of the study.

For more information on research or training on emotion, go to or
For more information on OnQ, the web survey authoring tool used to develop and deploy this survey, click here.
Copyright 2006 National Geographic Society. All rights reserved.
Button -  National Geographic Magazine home