Low Recognition of Emotion from Spontaneous Facial Expressions of New Guineans

1
Pamela Naab & James Russell* Boston College Abstract The current study (N=50) examined recognition from 20 spontaneous expressions from Papua New Guinea photographed, coded, and labeled by Ekman (1980). For the 16 faces with a single predicted label, average endorsement was 24.2%. For four faces with two predicted labels (blends), average endorsement was 38.8%. Spontaneous expressions do not achieve the level of recognition achieved by posed expressions. Introduction Can emotions be easily read from facial expressions? (e.g., Ekman, 1980; Tomkins, 1962; Izard, 1971). Limitations of these studies: Small number of emotions studied Blends not included Results cannot specify whether Emotions not translated into facial expressions in the first place? Or, could it be that observers failed to recognize the emotion? The Study Stimuli- Spontaneous facial expressions photographed, analyzed, and labeled by Paul Ekman (1980). South Fore of Papua New Guinea Isolated from Western culture: “There was an enormous advantage to being with a people who were not camera- shy. They did not know what a camera did so they were not self-conscious about it, and much of their social life was outdoors and easily seen” (Ekman, 1980, p.11). Ekman (1980) labeled the emotion conveyed by each expression, based on: Knowledge of the expresser’s situation Analysis of the facial muscle movements visible in the photograph Emotions: Emotions: States: States: happiness anger contempt happiness anger contempt hesitation hesitation sadness fear interest sadness fear interest perplexedness perplexedness surprise disgust embarrassment surprise disgust embarrassment relaxation relaxation Offered observers a greater than usual number of emotion labels Included blended expressions Ekman (1980) provided a clear prediction about the level of agreement to be expected: “Since these pictures show universal facial expressions, the message conveyed by each face will usually be quite obvious. In the captions to the plates I have added brief explanations of exactly how these emotions are registered on these faces (or for that matter, any face)” (p. 11). Method Participants 50 Boston College undergraduates. Materials Facial expressions were shown as still black and white 5” x 7” photographs of 20 spontaneous facial expressions of members of the South Fore of New Guinea from Face of Man (Ekman, 1980). Intensity Ratings. We asked whether there was a significant difference between the intensity ratings of Ekman’s predicted label and a comparison label For 2 faces, Ekman’s term was selected most often For 8 faces, the comparison label was selected most often And the remaining 12 faces showed no differences Consensus Scoring Endorsement of Modal Responses. Another measure of “recognition” sets aside the predicted label and simply relies on the modal response within this sample. Modal response endorsement = 42.2% Chance = 8.3%. Discussion Low Recognition of discrete emotion from spontaneous facial expression raises questions about past research using posed expressions. Endorsements were higher than chance, showing that our judges were not random. Perhaps they were able to assess the positivity of an expression and select emotion labels based on those criteria. Because recognition of emotion and non- emotion states (perplexed, hesitant, relaxed) did not differ, this suggests that the face may portray information beyond emotion. Procedure Instructions: 1. Circle the single best word for that emotion. 2. Please respond to every emotion word by circling a number. Each number represents the intensity to which the emotion is present in the face, ranging from 0 (none) to 4 (maximum intensity). None None Maximum Maximum Intensity Intensity I------------------------------------------ I------------------------------------------ I I 1. Happiness 1. Happiness 0 1 0 1 2 3 4 2 3 4 2. Fear 2. Fear 0 1 0 1 2 3 2 3 4 4 3. Interest 3. Interest 0 0 1 2 1 2 3 3 4 4 4. Anger 0 4. Anger 0 1 1 2 3 2 3 4 4 5. Embarrassment 0 5. Embarrassment 0 1 2 1 2 Results Ekman’s Predictions Endorsement. The most common method for assessing ‘recognition’ is the percent of participants who endorse the predicted label, in this case Ekman’s label. Single Emotion Mean Endorsement = 24.2% Random selection = 8.3% Blends 38.8% selected one of the two predicted labels Random selection = 16.7%. Modal Responses. Another measure of recognition is whether the modal response corresponded to Ekman’s predicted label. Single Emotions- 5 modal responses matched Ekman’s label (out of 16) With the 4 blended expressions, the two most frequently chosen labels matched two predicted labels for one face one of the predicted labels for two faces *Contact: [email protected] Study Descripti on Inductio n Emotions Recognition Motley & Camden (1988) Posed & Spontaneo us Photos Elaborat e interact ions & posed Happines s Sadness Anger Surprise Disgust Confusio n Posed- 81.4% All spontaneous- 26% Wagner, MacDonal d, & Manstead (1986) Dynamic spontaneo us expressio ns Emotion- elicitin g slides Happines s Sadness Anger Surprise Disgust Fear Neutral -Above chance: Happy- 48.4% Disgust- 22.69% -Chance levels (anger, sadness, fear, neutral) -Below chance (surprise) Yik, Chinese Lab- Happines Happiness- 31.5 42.2 84.4 0 10 20 30 40 50 60 70 80 90 100 Ekm an's Term M odal Response Posed Expressions Chance 8.3% Percent of Ss Point of Comparison Endorsement for a Particular Term Spontaneous expressions have obtained lower recognition than posed expressions: Which are representative of the facial expressions encountered everyday? Spontaneous vs. Posed Expressions Em barrassm ent 0 10 20 30 40 50 60 Happiness Relaxed Interest S urprise Perplexed Hesitant Contem pt S adness Embarras Disgust Fear A nger Em otion Label % E ndorsem ent Perhaps there are other things people read into a face, such as social messages (Fridlund, 1997) or action tendencies (Frijda & Tscherkassof, 1997). References Ekman, P. (1980). The Face of Man. New York, NY: Garland Publishing, Inc. Fridlund (1997). The new ethology of human facial expressions. In J.M. Russell, & J.M. Fernandez-Dols (Eds.), The psychology of facial expression (pp.78-102). New York: Cambridge University Press. Frijda, N. & Tcherkassof, A. (1997). Facial expressions as modes of action readiness. In Russell, J. & Fernandez-Dols, J. (Eds.), The psychology of facial expression (pp. 103-129). New York: Cambridge University Press. Izard, C. (1971). The face of emotion. CT, US: Appleton-Century-Crofts. Motley, M. & Camden, C. (1988). Facial expression of emotion: A comparison of posed expressions versus spontaneous expressions in an interpersonal communication setting.Western Journal of Speech Communication, 52, 1-22. Russell, J., Lewicka, M., & Niit, T. (1989). A cross-cultural study of a circumplex model of affect. Journal of Personality and Social Psychology, 57, 848-856. 24.3 40.6 24 29.2 0 5 10 15 20 25 30 35 40 45 Ekm an's Term M odalResponse Em otion States Emotion attributions for Ekman’s ‘embarrassment face (2 faces) Anger 0 10 20 30 40 50 60 Happiness Relaxed Interest S urprise Perplexed Hesitant Contem pt S adness Embarrass Disgust Fear A nger Em otion Label % E ndorsem ent Results Across Facial Expressions with the Same Predicted Emotion Emotion attributions for Ekman’s ‘anger face (1 face) Recognition of Emotion and Non-Emotion States % Endorsement

description

Low Recognition of Emotion from Spontaneous Facial Expressions of New Guineans. Pamela Naab & James Russell* Boston College. *Contact: [email protected]. Limitations of these studies: Small number of emotions studied Blends not included Results cannot specify whether - PowerPoint PPT Presentation

Transcript of Low Recognition of Emotion from Spontaneous Facial Expressions of New Guineans

Page 1: Low Recognition of Emotion from Spontaneous Facial Expressions of New Guineans

Pamela Naab & James Russell*

Boston College Abstract

The current study (N=50) examined recognition from 20 spontaneous expressions from Papua New Guinea photographed, coded, and labeled by Ekman (1980). For the 16 faces with a single predicted label, average endorsement was 24.2%. For four faces with two predicted labels (blends), average endorsement was 38.8%. Spontaneous expressions do not achieve the level of recognition achieved by posed expressions.

IntroductionCan emotions be easily read from facial expressions? (e.g., Ekman, 1980; Tomkins, 1962; Izard, 1971).

Limitations of these studies:Small number of emotions studiedBlends not includedResults cannot specify whether

Emotions not translated into facial expressions in the first place? Or, could it be that observers failed to recognize the emotion?

The Study

Stimuli- Spontaneous facial expressions photographed, analyzed, and labeled by Paul Ekman (1980).

South Fore of Papua New GuineaIsolated from Western culture:

“There was an enormous advantage to being with a people who were not

camera-shy. They did not know what a camera did so they were not self-conscious about it, and much of

their social life was outdoors and easily seen” (Ekman, 1980, p.11).

Ekman (1980) labeled the emotion conveyed by each expression, based on:

Knowledge of the expresser’s situation Analysis of the facial muscle movements visible in the photograph

Emotions:Emotions: States:States: happiness anger contempt hesitationhappiness anger contempt hesitation sadness fear interest sadness fear interest perplexednessperplexedness surprise disgust embarrassment relaxationsurprise disgust embarrassment relaxation

Offered observers a greater than usual number of emotion labels

Included blended expressions

Ekman (1980) provided a clear prediction about the level of agreement to be expected:

“Since these pictures show universal facial expressions, the message conveyed by each face will usually be quite obvious. In the captions to the plates I have

added brief explanations of exactly how these emotions are registered on these faces (or for that matter,

any face)” (p. 11).

Method

Participants

50 Boston College undergraduates.

Materials

Facial expressions were shown as still black and white 5” x 7” photographs of 20 spontaneous facial expressions of members of the South Fore of New Guinea from Face of Man (Ekman, 1980).

Selection of Photographs- We selected facial expressions in an effort to include as many different emotions as possible.

Intensity Ratings. We asked whether there was a significant difference between the intensity ratings of Ekman’s predicted label and a comparison label

For 2 faces, Ekman’s term was selected most often For 8 faces, the comparison label was selected most oftenAnd the remaining 12 faces showed no differences

Consensus ScoringEndorsement of Modal Responses. Another measure of “recognition” sets aside the predicted label and simply relies on the modal response within this sample.

Modal response endorsement = 42.2%Chance = 8.3%.

Discussion

Low Recognition of discrete emotion from spontaneous facial expression raises questions about past research using posed expressions.

Endorsements were higher than chance, showing that our judges were not random.

Perhaps they were able to assess the positivity of an expression and select emotion labels based on those criteria.

Because recognition of emotion and non-emotion states (perplexed, hesitant, relaxed) did not differ, this suggests that the face may portray information beyond emotion.

Procedure

Instructions:1. Circle the single best word for that emotion.2. Please respond to every emotion word by circling a number. Each number represents the intensity to which the emotion is present in the face, ranging from 0 (none) to 4 (maximum intensity).

None None Maximum Maximum Intensity Intensity I------------------------------------------II------------------------------------------I1. Happiness1. Happiness 0 1 2 3 4 0 1 2 3 42. Fear2. Fear 0 1 2 3 0 1 2 3 4 43. Interest3. Interest 0 0 1 2 3 1 2 3 4 44. Anger 04. Anger 0 1 2 3 1 2 3 4 45. Embarrassment 05. Embarrassment 0 1 2 3 1 2 3 446. Disgust6. Disgust 0 0 1 2 3 1 2 3 4 47. Sadness7. Sadness 0 0 1 2 3 1 2 3 4 48. Contempt8. Contempt 0 0 1 2 3 1 2 3 4 49. Relaxed9. Relaxed 0 0 1 2 3 1 2 3 4 410. Surprise10. Surprise 0 0 1 2 3 1 2 3 4 411. Perplexed11. Perplexed 0 0 1 2 3 1 2 3 4 412. Hesitant 012. Hesitant 0 1 2 3 1 2 3 4 4

ResultsEkman’s PredictionsEndorsement. The most common method for assessing ‘recognition’ is the percent of participants who endorse the predicted label, in this case Ekman’s label.

Single Emotion Mean Endorsement = 24.2%Random selection = 8.3%

Blends 38.8% selected one of the two predicted labels Random selection = 16.7%.

Modal Responses. Another measure of recognition is whether the modal response corresponded to Ekman’s predicted label.

Single Emotions- 5 modal responses matched Ekman’s label (out of 16)

With the 4 blended expressions, the two most frequently chosen labels matched

two predicted labels for one faceone of the predicted labels for two facesneither predicted labels for one face

*Contact: [email protected]

Study Description Induction Emotions Recognition

Motley & Camden (1988)

Posed & Spontaneous Photos

Elaborate interactions & posed

Happiness Sadness Anger Surprise Disgust Confusion

Posed- 81.4% All spontaneous- 26%

Wagner, MacDonald, & Manstead (1986)

Dynamic spontaneous expressions

Emotion-eliciting slides

HappinessSadnessAngerSurpriseDisgustFearNeutral

-Above chance:Happy- 48.4%Disgust- 22.69%-Chance levels (anger, sadness, fear, neutral)-Below chance (surprise)

Yik, Meng, & Russell (1995)

Chinese babies spontaneous expressions

Lab-eliciting procedures

Happiness Sadness Anger Surprise Disgust Fear

Happiness- 74%Other emotions- 23%

31.542.2

84.4

0102030405060708090

100

Ekman's Term ModalResponse

PosedExpressions

Chance 8.3%

Perc

en

t of

Ss

Point of Comparison

Endorsement for a Particular Term

Spontaneous expressions have obtained lower recognition than posed expressions:

Which are representative of the facial expressions encountered everyday?

Spontaneous vs. Posed Expressions

Embarrassment

0102030405060

Happiness

Relaxed

Interest

Surprise

Perplexed

Hesitant

Contempt

Sadness

Embarras

Disgust

Fear

Anger

Emotion Label

% E

nd

ors

emen

t

Perhaps there are other things people read into a face, such as

social messages (Fridlund, 1997)

or action tendencies (Frijda & Tscherkassof, 1997).

ReferencesEkman, P. (1980). The Face of Man. New York, NY: Garland Publishing, Inc.Fridlund (1997). The new ethology of human facial expressions. In J.M. Russell, & J.M. Fernandez-Dols (Eds.), The psychology of facial expression (pp.78-102). New York: Cambridge University Press.Frijda, N. & Tcherkassof, A. (1997). Facial expressions as modes of action readiness. In Russell, J. & Fernandez-Dols, J. (Eds.), The psychology of facial expression (pp. 103-129). New York: Cambridge University Press.Izard, C. (1971). The face of emotion. CT, US: Appleton-Century-Crofts. Motley, M. & Camden, C. (1988). Facial expression of emotion: A comparison of posed expressions versus spontaneous expressions in an interpersonal communication setting.Western Journal of Speech Communication, 52, 1-22.Russell, J., Lewicka, M., & Niit, T. (1989). A cross-cultural study of a circumplex model of affect. Journal of Personality and Social Psychology, 57, 848-856.Tomkins, S. (1962). Affect, imagery, consciousness: The positive affects. Oxford, England: Springer.Wagner, H., MacDonald, C., & Manstead, A. (1986). Communication of individual emotions by spontaneous facial expressions. Journal of Personality and Social Psychology, 50, 37-743.Yik, M., Meng, Z., & Russell, J. (1998). Adults’ freely produced emotion labels for babies’ spontaneous facial expressions. Cognition and Emotion, 12, 723-730.

24.3

40.6

24

29.2

0

5

10

15

20

25

30

35

40

45

Ekman's Term Modal Response

EmotionStates

Emotion attributions for

Ekman’s ‘embarrassment’

face (2 faces)

Anger

0102030405060

Happiness

Relaxed

Interest

Surprise

Perplexed

Hesitant

Contempt

Sadness

Embarrass

Disgust

Fear

Anger

Emotion Label

% E

nd

ors

emen

t

Results Across Facial Expressions with the Same Predicted Emotion

Emotion attributions for

Ekman’s ‘anger’ face (1

face)

Recognition of Emotion and Non-Emotion States

%

End

ors

em

ent