international conference on multimedia and expo (july 2000)face/papers/icme00.pdf · 2000-08-05 ·...
TRANSCRIPT
International Conference on Multimedia and Expo (July 2000)
Tutorial on Nonverbal Behavior and Paralinguistic Communication
Emotion and Paralinguistic Communication
• Jeffrey Cohn, Facial Expression in Human-and Human-machine Communication
• Francis Quek, Gesture, Speech, and Gaze in Human Conversational Interaction
• Ryohei Nakatsu, Recent Developments in Multi-Modal Interfaces
• Sidney Fels, Intimacy and embodiment at the interface
Facial Expression in Human- and Human-Machine Interaction
Jeffrey CohnAffect Analysis Group
Department of PsychologyRobotics Institute
Carnegie Mellon UniversityUniversity of Pittsburgh
Pittsburgh, PA 15260 USAhttp://www.pitt.edu/~jeffcohn/jfc.html
http://www.cs.cmu.edu/~face
Overview• What does facial expression express?• Factors that influence production and
interpretation of facial expression• Measurement
– Manual: Facial Action Coding System– Automatic
• Relation of facial to other modes of emotion expression
• Interpretation and user-interface
What Does Facial Expression Express?• Emotion
• Readiness for action
• Social signal (e.g., emblems)
• Meta-communication
• Cognitive activity
• Self-presentation
• Physiology
Basic Emotions
Fear Anger DisgustSadness SurpriseJoy
Others? interest, embarrassment, shame, contempt . . .
Cohn et al. (2000)
Defining Features (From Ekman, 1992)
• Distinctive universal signals • Distinctive universals in antecedent events• Distinctive physiology• Presence in other primates• Coherence among emotional response• Quick onset• Brief duration• Automatic appraisal• Unbidden occurrence
– Posture < face < voice < autonomic response
Emotions Emerge asThey Become Adaptive
• Newborn– Disgust in response to sour
taste– Pain/Distress– Joy
• (Social smile and directed gaze at about 6 wks)
• 6-12 months– Surprise– Sadness– Anger– Fear
Dimensional Approaches
• Emotion circumplex• Valence and arousal
• Positive Affect (PA) and Negative Affect (NA)
• Approach/withdrawal
Circumplex: Self-Reported Emotion
Adapted from Lang et al. (199-) and Watson & Clark, 1985.
Excited
Pleasant
PANA
Unpleasant
Calm
Circumplex: Advantages
• Economical representation
• Consistent with both self-report and facial expression data
• Emphasizes importance of arousal
• Explicitly recognizes commonalities
Circumplex: Disadvantages
• Includes non-emotions, e.g., sleepy
• Assumes relationships among emotions are invariant
• De-emphasizes what is unique
• Puts emphasis on phenomenology of emotion to exclusion of other aspects, such as function
Emotions Have Multiple Components
Prosody
Kinesics
Congruence?
ANS
HR, BP, …
(Arousal)
(Specificity)
FaceCNS
Emotion
Valence
Action Readiness (from Fridlund, 1994)
Readiness to continueContentment
Declaration of superiorityContempt
Readiness to submit or flee
Fear
Readiness to attackAnger
Recruitment of succorSad
Appeasement“Non-Duchenne” smile
Readiness to play, affiliate
“Duchenne” smile
What Else Does Facial Expression Express?
• Meta-communication– Turn-taking
– Back channeling
– Relationship
• Cognitive activity
• Self-presentation
Factors That Influence Production and Interpretation
• Presence of other people and who they are• Display rules
– Minimize– Neutralize– Exaggerate– Mask
• Cultural differences• Gender• Individual differences in
– facial appearance and functional anatomy– Hx of facial expression
•Facial nerve (7th cranial nerve)
•Ipsalateral versus contralateral control
•Pyramidal versus extrapyramidal motor track
Motor Control of the Facial Muscles
From Rinn (199-)
Description of Facial Expression
• Basic-emotions– E.g., EMFACS, AFFEX
• Valence– E.g., Cohn et al. (1990)
• Sign-based approaches– Facial Action Coding System (FACS)
Asymmetry
From Wachtman & Cohn et al. (in press).
• Faces are asymmetric
• ‘True’ versus ‘false’ expressions
• Some expressions are typically asymmetric
• Neuropathology
Facial Action Coding System* Action Units (FACS AUs)
• Upper Face AUs: 12• Lower Face AUs: 18• Eye Position: 6• Head Position: 8• Miscellaneous: 14• Combinations
– Additive– Non-Additive * FACS: Ekman & Friesen
(1978)
AU 12
AU 20
AU 9
AU 10
AU 25
AU 26
AU 27
AU 17
AU 15
Adapted from Farkas & Munro, 1994, 1987
Selected FACS Action UnitsAU 1
AU 2
AU 4
AU 5
AU 6
AU 7
Upper Face AUsAU1 AU2 AU4
Brows lowered & drawn together
Inner portions of the brows are raised
Outer portions of the brows are raised
AU5 AU1+4 AU4+5
Upper eyelids are raised Medial portion of the brow is raised & pulled together
Brows lowered & drawn together & upper eyelids
are raised
AU6 AU7 AU1+2+4+5+7
AU1+2+5+6+7 AU1+2+4+5+6 AU1+2+4+5+6+7
Cheeks are raised & eye opening is narrowed
Lower eyelids are raised
Upper Face AUs
Automated Analysis
• Feature Extraction– Holistic
– Feature based
• Static versus dynamic face image data
• Rigid motion
lip width
39642baby
frame number
6561575349454137332925211713951
change fro
m fra
me 1
.3
.2
.1
0.0
-.1
right lip corner height
39642 baby
frame number
6561575349454137332925211713951
cha
ng
e f
rom
fra
me
1
.2
.1
0.0
-.1
Prosody:Not what you say but how you say
it
ProsodyLoudness ~ intensity Rhythm ~ temporal characteristicsFrequency
Prosodic Features Vary With
• Discrete Emotion– latency longer and more variable in depression– pitch increased in joy, decreased in disgust
• Arousal– speech rate in boredom versus excitement
• Communicative Intent– “Bev loves BOB?”– “Bev LOVES Bob”
• Interpersonal synchrony and attraction– matching of turn-taking duration
• Individual, gender, and regional differences
Summary
• System performance likely to vary with– Quality of ground truth training data
– Arousal
– Emotion • dimensions
• discrete emotion
– Communicative intent
– Presence of other people &their relationship to target person
– Individual differences: gender, culture, situation
Summary cont.
• Convergent measures increase accuracy
• Multi-modal information may yield the more robust recognition
• Mapping of measurements to interpretation will require multi-disciplinary efforts
Further Reading
Bachorowski, J.A. (1999). Vocal perception and expression of emotion. Current Directions in Psychological Science, 8, 53-56.
Clark, M.S. (Ed.) (1992). Review of Personality and Social Psychology, Vol. 13: Emotion and Vol. 14: Emotion and Social Behavior. Newbury Park, CA: Sage.
Cohn, J.F. & Katz, G.S. (September 1998). Bimodal expression of emotion by face and voice. ACM and ATR Workshop on Face/Gesture Recognition and Their Applications, pp. 41-44. Bristol, United Kingdom. http://www.cs.cmu.edu/~face
Damasio, A.R. (1994). Descartes' error: Emotion, reason, and the human brain. NY:Grosset/Putnam.
Darwin, C. (1872/1998). The expression of the emotions in man and animals. Third Edition. NY: Oxford
Ekman, P. (1985). Telling lies. NY: Basic Books.
Ekman, P. (1992). An argument for basic emotions. In N. Stein & K. Oatley, K. (Eds.) Basic emotions. Cognition and Emotion, 6, Nos. 3-4, 161-168. Hillsdale, NJ: LEA.
Ekman, P. & Friesen, W. (1978). Facial action coding system. Palo Alto, CA: Consulting Psychologists. Available from P. Ekman, Human Interaction Laboratory, University of California at San Francisco.
Eibl-Eibesfeldt, I. (1989). Human ethology. NY: Aldine de Gruyter.
Frick, R.W. (1985). Communicating emotion: The role of prosodic features. Psychological Bulletin, 97, 412-429.
Fridlund, A.J. (1994). Human facial expression: An evolutionary view. NY: Academic.
Kanade, T., Cohn, J.F., & Tian, Y. (March 2000). Comprehensive Database for Facial Expression Analysis. Proceedings of the Fourth IEEE International Conference on Automatic Face and Gesture Recognition (FG'00), pp. 46-53. Grenoble, France.
Lien, J.J.J., Kanade, T., Cohn, J.F., & Li, C.C. (2000). Detection, tracking, and classification of subtle changes in facial expression. Journal of Robotics and Autonomous Systems, 31,131-146. . http://www.cs.cmu.edu/~face
Matias, R. and Cohn, J.F. (1993). Are MAX-Specified Infant Facial Expressions During Face-to-Face Interaction Consistent with Differential Emotions Theory? Developmental Psychology, 29, 524-531.
Tian, Y.L, Kanade, T., & Cohn, J.F. (Submitted). Recognizing action units for facial expression analysis. .
Tian, Y., Kanade, T., and Cohn, J.F. (June 2000). Recognizing upper face action units for facial expression analysis. Proceedings of the IEEE Sponsored Conference in Computer Vision and Pattern Recognition, pp. xxx-xxx. Hilton Head, Georgia.http://www.cs.cmu.edu/~face
Additional readings on the topics of facial and prosodic expression and emotion can be found at http://www.pitt.edu/~jeffcohn/emot99.pdf