Towards Emotionally-Intelligent Pedagogical Agents

Research shows that emotions play an important role in learning. Human tutors are capable of identifying and responding to the affective states of their students; therefore, for ITSs to be truly affective, they should also be capable of tracking and appropriately responding to the emotional state of their users. We report on a project aimed at developing an affect-aware pedagogical agent persona for an ITS for teaching database design skills. We use the dimensional approach to affective modeling, and track the users' affective state along the valence dimension as identified from tracking the users' facial features. We describe the facial-feature tracking application we developed, as well as the set of rules that control the agent's behavior. The agent's response to the student's action depends on the student's cognitive state (as determined from the session history) as well as on the student's affective state. The experimental study of the agent shows the general preference towards the affective agent over the non-affective agent.

[1]  Clifford Nass,et al.  Wired for Speech: How Voice Activates and Advances the Human-Computer Relationship , 2005 .

[2]  M. Csíkszentmihályi.  Flow: The Psychology of Optimal Experience , 1990 .

[3]  Winslow Burleson,et al.  Gender-Specific Approaches to Developing Emotionally Intelligent Learning Companions , 2007, IEEE Intelligent Systems.

[4]  Igor S. Pandzic,et al.  MPEG-4 Facial Animation , 2002 .

[5]  Takeo Kanade,et al.  Recognizing Action Units for Facial Expression Analysis , 2001, IEEE Trans. Pattern Anal. Mach. Intell..

[6]  Timothy W. Bickmore.  Unspoken rules of spoken interaction , 2004, CACM.

[7]  Jonathan Klein,et al.  This computer responds to user frustration: Theory, design, and results , 2002, Interact. Comput..

[8]  Antonija Mitrovic,et al.  Feedback Micro-engineering in EER-Tutor , 2005, AIED.

[9]  M. Bradley,et al.  Measuring emotion: Behavior, feeling, and physiology , 2000 .

[10]  Judy Kay,et al.  Learner Control , 2001, User Modeling and User-Adapted Interaction.

[11]  Punya Mishra,et al.  Etiquette and the design of educational technology , 2004, CACM.

[12]  Arthur C. Graesser,et al.  Toward an Affect-Sensitive AutoTutor , 2007, IEEE Intelligent Systems.

[13]  B. Bloom.  The 2 Sigma Problem: The Search for Methods of Group Instruction as Effective as One-to-One Tutoring , 1984 .

[14]  Algirdas Pakstas,et al.  MPEG-4 Facial Animation: The Standard,Implementation and Applications , 2002 .

[15]  Rob Reilly,et al.  An Affective Module for an Intelligent Tutoring System , 2002, Intelligent Tutoring Systems.

[16]  L. Nadel,et al.  Cognitive neuroscience of emotion , 2000 .