-- GenaroRebolledoMendez - 26 Feb 2007

* We need to answer the question: "if you want to teach chlidren social skills, why not have them interact with real children and real adults?"

- Answers, in our environment: child can observe themselves : child can play / experiment freely : we can give consistent rewards : we can have fine grained control of the environment and tasks.....

* We may need a new project partner who is expert on realtime video motion capture?

* reducing stress by simplifying environment

* reducing pressure - computers do not react impatiently

* We need to compare our proposal to Embodied Conversational Agents, e.g. Justine Cassell's SAM: http://www.media.mit.edu/gnl/projects/castlemate/ and DollTalk : http://web.media.mit.edu/~cati/dolltalk.html

* Minor points: we should use real pre-recorded voices at all times, not TTS/speech synthesis. For the room transitions we should have a fade out, but keep the child-avatar on screen at all times (possibly go back to the mirror image?) , and then a fade in on the new room appearing around them.

Just a few more relevant projects:

Affective Learning Companion: http://affect.media.mit.edu/projectpages/lc/ALC/ALC.htm

Affective Mirror: http://affect.media.mit.edu/projects.php?id=753

ESP—The Emotional-Social Intelligence Prosthesis: http://affect.media.mit.edu/projectpages/esp/