Useful Roles of Emotions in Animated Pedagogical Agents. Ilusca L. L. Menezes IFT6261 :: Winter 2006

Transcription

1 Useful Roles of Emotions in Animated Ilusca L. L. Menezes IFT6261 :: Winter 2006

2 Objectives To provide opportunities: To understand the importance of the emotions in learning To explore how to model emotions in pedagogical agents 03/04/2006 Ilusca L. L. Menezes 2

3 Presentation 03/04/2006 Ilusca L. L. Menezes 3

4 Emotion and Cognition 03/04/2006 Ilusca L. L. Menezes 4

5 Emotion and Cognition The traditional view Philosophers like Plato regarded emotion as irrational urges that needed to be controlled through the use of reason (O'Regan, 2003) The recent view Emotion is related to cognition and plays a fundamental role in learning, behaviour and decision making (Salovey et al, 1990; Damasio,1994) 03/04/2006 Ilusca L. L. Menezes 5

6 What is Emotion? Components of Emotion Theories of Emotion Models of Emotion 03/04/2006 Ilusca L. L. Menezes 6

7 What is Emotion? a psychic and physical reaction (as anger or fear) subjectively experienced as strong feeling and physiologically involving changes that prepare the body for immediate vigorous action (Merriam-Webster Dictionary) To understand the emotion is a difficult task: Emotion is a complex term that has no single universally accepted definition 03/04/2006 Ilusca L. L. Menezes 7

8 Components of Emotion (1/2) Emotion can be characterised by three components: 1. Physiological Internal physical changes 2. Behavioural Outward signs of an emotional state 3. Cognitive Thoughts, expectations and beliefs 03/04/2006 Ilusca L. L. Menezes 8

9 Components of Emotion (2/2) Example: Anxiety 1. Physiological Physical changes such as increased heart rate, breathing, sweating 2. Behavioural Facial expressions of tension Closed body posture 3. Cognitive Meaning associated with this emotion, such as I m really worried about this presentation 03/04/2006 Ilusca L. L. Menezes 9

10 Theories of Emotion (1/4) Common sense An event produces the feeling of an emotion This feeling produces physiological changes and behaviour 03/04/2006 Ilusca L. L. Menezes 10

11 Theories of Emotion (2/4) The James-Lange approach (1880) Emotion arises from physiological changes and behaviour 03/04/2006 Ilusca L. L. Menezes 11

12 Theories of Emotion (3/4) The Cannon-Bard approach (1927) We experience physiological changes, behaviour and emotional at the same time 03/04/2006 Ilusca L. L. Menezes 12

13 Theories of Emotion (4/4) The Schacter-Singer approach (1962) Cognitive theory Emotion arises of two factors: physiological changes interpretation of these changes (based on the context) 03/04/2006 Ilusca L. L. Menezes 13

14 Models of Emotion Researchers have developed models of emotion based on cognitive appraisal theory Ortony, Clore and Collins' Structure of Emotion (1988) is an example: it was developed with the aim to implement it in a computer supports twenty-two emotion categories emotions arise from valenced reactions (the reactions of the emotional value associated with a stimulus), that can be positive and negative the stimulus can be induced by events, by agents or by objects 03/04/2006 Ilusca L. L. Menezes 14

15 OCC Model of Emotion 03/04/2006 Ilusca L. L. Menezes 15

16 Emotions affect Learning 03/04/2006 Ilusca L. L. Menezes 16

17 Emotions affect Learning (1/2) Psychologists and educators have been pointed out the way as the emotions affect the learning (Goleman, 1995; Piaget, 1989) Students who are anxious, angry, or depressed don t learn; people who are caught in these states do not take in information efficiently or deal with it well (Goleman, 1995) 03/04/2006 Ilusca L. L. Menezes 17

18 Emotions affect Learning (2/2) Positive emotions: generally enhance motivation facilitate learning and performance Intense negative emotions: can block the thought processes reduce memory capacity (Isen,1993) and inductive reasoning (Idzihowski,1987) can minimize motivation level, interfere with learning, and contribute to low performance 03/04/2006 Ilusca L. L. Menezes 18

19 Architecture of a Pedagogical Agent Animated Modelling Emotions in agents Recognizing Learner s Emotions Designing Emotive Behaviours 03/04/2006 Ilusca L. L. Menezes 19

20 (1/2) When the Intelligent Tutoring Systems (ITS) interact with the learner, they modify their bases of knowledge, they perceive the learner's interventions and they can learn and adapt the teaching strategies according to the learner's performance. To make this improvement, intelligent agents are introduced in this environment and they are called Pedagogical agents 03/04/2006 Ilusca L. L. Menezes 20

21 (2/2) Are cognitive agents (Frasson et al, 1996) Intelligent agents Autonomous Social ability Reactive Instructable Adaptability 03/04/2006 Ilusca L. L. Menezes 21

22 Architecture of a Pedagogical Agent 03/04/2006 Ilusca L. L. Menezes 22

23 Animated Are animated lifelike characters designed to facilitate learning Can communicate with learners both visually and verbally and they utilize different kinds of emotions to do this Play an important motivational role as they interact with learners: can have a strong positive effect on students' learning the persona effect (Lester et al, 1997) 03/04/2006 Ilusca L. L. Menezes 23

24 Examples (1/2) COSMO inhabits the Internet Advisor, a learning environment for the domain of Internet packet routing (Towns et al, 1998) STEVE helps students learn to perform physical, procedural task, such as operating or repairing complex equipment (Rickel et al, 1997) 03/04/2006 Ilusca L. L. Menezes 24

25 Examples (2/2) HERMAN THE BUG inhabits the Design-A-Plant learning environment and helps children learn about botanical anatomy and physiology (Towns et al, 1998) THE PRIME CLIMB AGENT provides hints that help the student to reason about number factorization in Prime Climb, an electronic educational game (Conati et al, 2002) 03/04/2006 Ilusca L. L. Menezes 25

26 Modelling Emotions in agents (1/3) Using OCC Model: we need to define: In the learning environment: the set of events, actions and objects For the agent: we relate: the set of goals, standards and attitudes events with goals actions with standards objects with attitudes the emotions are generated, matching these factors 03/04/2006 Ilusca L. L. Menezes 26

27 Modelling Emotions in agents (2/3) Using OCC Model: For Events: an agent need to have a set of goals which help define his personality these goals may match situations that arise (event) in the simulation when there is a match between some event and the goal-based concerns of agent, emotions are generated events can be desirable or undesirable 03/04/2006 Ilusca L. L. Menezes 27

28 Modelling Emotions in agents (3/3) Example: 03/04/2006 Ilusca L. L. Menezes 28

29 Recognizing Learner s Emotions (1/3) A pedagogical agent needs to decide when and how to intervene the learner. For that, it has to recognize the learner s emotions Which are the mechanisms that can be used to recognize the emotions of a learner? Sensors that can detect the emotions through of the voice intonation, facial expressions, muscular tension and breath We can observe the learner s behaviour the actions of them in system, for example: time of execution of an activity, success or fails in execution of an exercise and order of aid 03/04/2006 Ilusca L. L. Menezes 29

30 Recognizing Learner s Emotions (2/3) How to recognize the emotions of the learner? We can use the OCC model to recognize learner s emotions: we define: the set of events, actions and objects in the learning environment the set of goals, standards and attitudes of the learner we relate: events with goals, actions with standards, objects with attitudes the emotions are generated, matching these factors 03/04/2006 Ilusca L. L. Menezes 30

31 Recognizing Learner s Emotions (3/3) Example: 03/04/2006 Ilusca L. L. Menezes 31

32 OCC Model Positive points: It has served as the basis for implementation of several other computational models It is a simple model Limitations: It focus only on the cognitive structures and mechanisms mediating the interpretation of external stimuli (events, agents, objects). It does not give attention to the physiology and behaviour on emotional processing It assumes fixed goals It does not employ a learning mechanism of emotion 03/04/2006 Ilusca L. L. Menezes 32

33 A DBN for Emotion Recognition (1/2) It is difficult to evaluate precisely the learner s emotions To deal with the high level of uncertainty involved in recognizing learner s emotions, Conati et al use Dynamic Decision Networks (DDNs) that detect variety of affective states based on the OCC model (Conati et al, 2004) There are several reasons for using DDNs to model emotions (Conati et al, 2004): DDNs generate as accurate an assessment on the user emotional state DDNs allow representing the temporal evolution of emotion they provide formal mechanisms based on decision theory to model how an agent can rationally chose among actions with uncertain outcomes 03/04/2006 Ilusca L. L. Menezes 33

34 A DBN for Emotion Recognition (2/2) Situations consist of the outcome of any event caused by either a learner s or an agent s action The nodes Goals are the goals that a learner may have The desirability of an event is represented by the node Goals Satisfied The nodes Emotional States are the emotions that can be generated User goals can depend on User Traits such as personality User goals can influence user Interaction Patterns (Conati et al, 2004) 03/04/2006 Ilusca L. L. Menezes 34

35 Designing Emotive Behaviours (1/2) To design emotive behaviours in an pedagogical agent, it is necessary to create: a general behaviour space populated with emotive behaviours and another with pedagogical speech acts an sequencing engine to dynamically plan full-body emotive behaviours in real time by selecting relevant pedagogical speech acts and then assembling appropriate visual behaviours 03/04/2006 Ilusca L. L. Menezes 35

36 Designing Emotive Behaviours (2/2) Figure: The lifelike pedagogical agent behavior planning architecture (Towns et al, 1998) 03/04/2006 Ilusca L. L. Menezes 36

37 Present Works Conferences 03/04/2006 Ilusca L. L. Menezes 37

38 Emotions play an important role in learning Through animated pedagogical agents, emotions have been incorporated into the learning environment, adding motivation and increasing the learners' performance The OCC model of emotion was showed. We can use it not only to model agent s emotions but also to model learner s emotions 03/04/2006 Ilusca L. L. Menezes 38

39 Present Works Many researchers have been developed computational models of emotion EMA (EMotion and Adaptation) a general computational model of emotion (Gratch et al, 2004) Works have been done with the intention to induce emotions in the learners Inducing Optimal Emotional State for Learning in Intelligent Tutoring Systems (Chaffar et al, 2004) 03/04/2006 Ilusca L. L. Menezes 39

40 Conferences AAAI-06 - The Twenty-First National Conference on Artificial Intelligence: aamas-06 - Fifth International Joint Conference on Autonomous Agents and Multiagent Systems: ITS-06 International Conference on Intelligent Tutoring Systems: FLAIRS-06 - The 19th International FLAIRS Conference: IAT-06 - ACM International Conference on Intelligent Agent Technology: IVA-06 The 6 th International Conference on Intelligent Virtual Agents: IJCAI-07 - International Joint Conference on Artificial Intelligence: ICALT-06 - The 6th IEEE International Conference on Advanced Learning Technologies: ECAI-06 -The 17th European Conference on Artificial Intelligence: TICE-06 -Technologies de l Information et de la Communication dans l Enseignement Supérieur et l Entreprise: 03/04/2006 Ilusca L. L. Menezes 40

41 03/04/2006 Ilusca L. L. Menezes 41

42 (1/4) (O'Regan, 2003) O'Regan, K. - Emotion and E-Learning. Journal of Asynchronous Learning Networks, Vol. 7, No. 3, (Salovey et al, 1990) Salovey, P., Mayer, J.D. - Emotional Intelligence. Imagination, Cognition and Personality, Vol. 9, pp , (Damasio, 1994) Damasio, A. - Descartes s Error: Emotion, Reason, and the Human Brain. Putnam Press, NY, (Goleman, 1995) Goleman, D. - Emotional Intelligence. New York: Bantam Books, (Piaget, 1989) Piaget, J. - Les relations entre l intelligence et l affectivité dans le developpement de l enfant. B. Rimé, K. Scherer (Eds.) Les Émotions. Textes de base en psychologie. Paris: Delachaux et Niestlé, pp.75-95, (Isen, 1993) Isen, A. M. - Positive Affect and Decision Making. Handbook of Emotions, New York: Guilford, pp , (Idzihowski, 1987) Idzihowski, C., Baddeley, A. - Fear and performance in novice parachutists. Ergonomics, Vol. 30, pp , (Ortony et al, 1988) Ortony, A. Clore, G. and Collins, A. - The Cognitive Structure of Emotions. Cambridge: Cambridge University Press, /04/2006 Ilusca L. L. Menezes 42

43 (2/4) (Picard, 2002) Picard, R. W. - Affective Computing. The MIT Press, (Frasson et al, 1996) Frasson, C., Mengelle, T., Aïmeur, E., Gouardères, G. - An Actor-based Architecture for Intelligent Tutoring Systems, ITS 96 Conference, Lecture Notes in Computer Science, No 1086, Springer Verlag, Montréal, pp , (Lester et al, 1997) Lester, J. C. Converse, S. A. Kahler, S. E. Barlow, S.T. Stone, B.A. and Bhogal, R. -The persona effect: Affective impact of animated pedagogical agents. In Proceedings of CHI 97 Human Factors in Computing Systems, pp New York: ACM, (Lester et al,1999) Lester, J., Voerman, J., Towns, S., and Callaway, C. - Deictic believability: Coordinating gesture, locomotion, and speech in lifelike pedagogical gents. Applied Artificial Intelligence 13(4 5): pp , (Johnson et al, 2000) Johnson, W. L., Rickel, J., and Lester, J. - Animated : Face-to-face interaction in interactive learning environment. International Journal of Artificial Intelligence in Education, 11, pp , (Towns et al, 1998) Towns, S., Fitzgerald, P., & Lester, J. - Visual Emotive Communication in Lifelike. In Proceedings of Intelligent Tutoring Systems 98 Conference, Eds. Goettl, B., Halff, H., Redfield, C. & Shute V., Springer-Verlag, /04/2006 Ilusca L. L. Menezes 43

44 (3/4) (Lester et al, 1999) Lester, J., Towns, S., Fitzgerald, P. - Achieving Affective Impact: Visual Emotive Communication in Lifelike International Journal of Artificial Intelligence in Education, 10, pp , (Rickel et al, 1997) Rickel, J., & Johnson, W. L. - Steve: An animated pedagogical agent for procedural training in virtual environments. Workshop on Animated Interface Agents: Making Them Intelligent (August 25) Nagoya, Japan, (Conati et al, 2002) Conati, C., Zhou, X - A Probabilistic Framework For Recognizing and Affecting Emotions. Journal of Applied Artificial Intelligence, special issue on Merging Cognition and Affect in HCI, Vol. 16 (7-8), pp , (Conati et al, 2004) Conati, C., MacLaren, H. - Evaluating a probabilistic model of student affect. Presented at Seventh International Conference on Intelligent Tutoring Systems, Maceio, Brazil, (Conati et al, 2004) Conati, C. and Zhou X. - A Probabilistc Framework for Recognizing and Affecting Emotions. To appear in: Proceedings of the AAAI 2004 Spring Symposium on Architectures for Modeling Emotions, Stanford University, CA, U.S.A, /04/2006 Ilusca L. L. Menezes 44

45 (4/4) (Gratch et al, 2004) Gratch, J. and Marsella, S. - A domain independent framework for modeling emotion. Journal of Cognitive Systems Research,. 5(4): pp , (Chaffar et al, 2004) Chaffar, S., Frasson, C. - Inducing Optimal Emotional State for Learning in Intelligent Tutoring Systems. In Proceedings of Intelligent Tutoring Systems 2004 Conference, Eds. Lester, J., Vicari, R., Paraguaçu, F., Springer-Verlag, Site: Merriam-Webster Dictionary: 03/04/2006 Ilusca L. L. Menezes 45