Information Technology Reference
In-Depth Information
Modeling Motivations, Personality Traits and Emotional
States in Deliberative Agents Based on Automated
Planning
Daniel Perez-Pinillos, Susana Fernandez, and Daniel Borrajo
Departamento de Informatica, Universidad Carlos III de Madrid,
Avda. de la Universidad, 30, Leganes (Madrid), Spain
{daniel.perez,susana.fernandez,daniel.borrajo}@uc3m.es
Abstract.
There is a wide variety of applications that require modeling the be-
haviour of virtual agents. Some of these applications aim at human interaction,
such as virtual assistants, and others aim at simulation of human behavior, such
as games or robotics. Most of these applications require not only some level of
intelligent behavior, but also a display of realistic human behavior. This has led
to the definition and use of models that integrate features like emotions, personal-
ity traits, preferences and motivations. Most of this work has been carried out in
the context of reactive architectures. Thus, the reasoning on the emotional state
of agents is only performed for the very next future, generating behavior that is
myopic for middle or long term goals. In this paper, we propose instead a delib-
erative model based on automated planning that integrates all these features for
long term reasoning.
Keywords:
Planning,
Agent,
Emotion,
Personality,
Motivation,
Preference,
Decision making.
1
Introduction
In many domains, the behaviour of any agent can be seen as a sequential decision-
making process, i.e. the cognitive process results in the selection of a course of actions
to fulfill some goals. The decision making process is a continuous process integrated
with the interaction with the environment where individual decisions must be examined
in the context of a set of needs and preferences that the agent has. Recent theories state
that human decision-making is also influenced by marker signals that arise in bioregu-
latory processes, including those that express themselves in emotions and feelings [11].
Probably, this is one of the reasons why the work on reasoning about emotions is be-
coming increasingly relevant, specially in contexts such as assistive technology, user
interfaces, or virtual agents [4,14].
In spite of the wide variety of points of view that have been used to study emotions,
it seems there is some agreement to consider emotion as an inborn and subjective reac-
tion to the environment, with an adaptive function, and accompanied of several organic,
physiological and endocrine changes [17]. Another point of agreement is that emotions
