Information Technology Reference
In-Depth Information
Representing Emotion and Mood States for
Virtual Agents
Luis Pena
1
,Jose-Mar ıa Pena
2
, and Sascha Ossowski
1
1
Universidad Rey Juan Carlos
{
luis.pena,sascha.ossowski
}
@urjc.es
2
Universidad Politecnica de Madrid
jmpena@fi.upm.es
Abstract.
Emotional agents are an active research domain, with direct
application in several industrial fields such as video games, interactive
environments, or enhanced human computer interactions. Emotional be-
haviour should consider both the representation of the emotions and the
mood states. The two most widely accepted and used models for this pur-
pose, the OCC and the PAD models, are rooted in cognitive psychology.
Upon this background, this paper puts forward two main contributions:
firstly, it discusses the use of a common representation for both mood
states and emotions; secondly, it introduces the concept of the Mood
Vector Space and analyses its the properties.
Keywords:
Emotional Model, OCC, PAD, Mood Vector Space.
1
Introduction
The creation of more believable agents for virtual environments and simulation
scenarios is putting the problem of emotion synthesis and analysis into the spot-
light. Many theories have been created for this purpose. Some are rather complex,
based on numerous parameters derived from empirical data, but are inapplica-
ble from an engineering perspective; others are tractable from a modelling and
a computational point of view but have less psychological basis.
The main objectives that emotional synthesis has to deal with are (1) emotion
creation based on environment analysis, (2) projection of these emotions into
actions of the agent and (3) modelling of the influence of personality over the
entire process. Setting out from the PAD Temperament and Emotional models
created be Mehrabian [8,7] and their psychological foundations, in this paper we
present an approach for the projection and representation of emotions within
a continuous space. In particular, we define the Mood Vector Space to model
emotions and transitions among them. Its target are architectures for agents
acting in complex and dynamic environments such as video games, which require
This work was supported in part by the Spanish Ministry of Science and Innovation
through the projects
AT
(Grant CONSOLIDER CSD2007-0022, INGENIO 2010)
and
OVAMAH
(Grant TIN2009-13839-C03-02).
