Page 231 - Socially Intelligent Agents Creating Relationships with Computers and Robots

214 Socially Intelligent Agents

in nature, and empirical studies on exploratory learning environments [16] have
shown that they tend to be effective only for those students that already possess
the meta-cognitive skills necessary to learn from autonomous exploration (such
as self-monitoring, self-questioning and self-explanation).
In this chapter, we discuss how to improve the effectiveness of educational
games by relying on socially intelligent agents (SIAs). These agents are active
game characters that can generate tailored interventions to stimulate students’
learning and engagement, by taking into account the student’s cognitive states
(e.g., as knowledge, goals and preferences), as well as the student’s meta-
cognitive skills (e.g., learning capabilities) and emotional reactions.

2. SIAs as Mediators in Educational Games
We argue that the effectiveness of educational games can be increased by
providing them with the capability to (i) explicitly monitor how students interact
with and learn from the games; (ii) generate calibrated interventions to trigger
constructive reasoning and reflection when needed.
However, this must be done without interfering with the factors that make
games fun and enjoyable, such as a feeling of control, curiosity, triggering of
bothintrinsicandextrinsicfantasies, andchallenge[12]. Thus, itisnotsufficient
to provide educational games with the knowledge that makes more traditional
Intelligent Tutoring Systems effective for learning: an explicit representation
of the target cognitive skills, of pedagogical knowledge and of the student’s
cognitive state. It is fundamental that the educational interventions be delivered
withinthespirit of the game, bycharactersthat (i) areanintegral part of thegame
plot; (ii) are capable of detecting students’ lack of engagement, in addition to
lack of learning; (iii) know how to effectively intervene to correct these negative
emotional and cognitive states.
Basically, these characters must play, in the context of the game, the medi-
ating role that teachers and external instructional activities have played during
the most successful evaluations of the EGEMS prototypes. The requirement
that these agents be socially intelligent is further enforced by the fact that
we are currently interested in investigating the educational potential of multi-
player computer games to support collaborative learning. In the last few years
there has been increasing research on animated pedagogical agents and there is
already empirical evidence of their effectiveness in fostering learning and mo-
tivation [17]. Our work extends existing research toward making pedagogical
agents more socially apt, by enabling them to take into account users’ affective
behaviour when adapting their interventions and to engage in effective collab-
orative interactions.