Information Technology Reference
In-Depth Information
at how humans handle their desires, motivation theory [6] bridges the gap between a
being's personality and its desires it wishes to satisfy. A
motive
such as “
benevolence
”
or “
greed
” is a basic marker of an agent's personality that can be used to create (or
abandon) some desire. We illustrate this intuition with a simple example.
Example 1.
Given an agent competes with other agents for food, the motive “
greed
”
would generate the desire of acquiring as much food as possible, while motive “
benev-
olence
” would generate the desire of acquiring just as much food as really needed and
to ensure that other agents acquire as much food as they need. Both motives might very
well be present in the agent's personality but with differing strengths.
In this paper, we develop a formal account for incorporating motivation into the
BDI
approach. Instead of assuming desires to be given we assume that an agent comprises of
some set of basic motives that drives its behavior. Each motive of the agent is equipped
with some weight and each motive is coupled with a set of desires that can be positively
or negatively influenced by the motive. We give a formal account on the aggregation
of the weights of the motives and these couplings in order to determine the desires
the agent is motivated to follow. Furthermore, using the notion of
reliability
[4] we
investigate how beliefs about actions (
know-how
) and beliefs about the world might
influence the deliberation on motivation. More precisely, the motivation to follow some
desire is strongly influenced by the uncertainty of the world and the knowledge of the
agent that some course of action might not be reliable achievable.
The rest of this paper is organized as follows. In Section 2 we begin by giving some
background on
BDI
agents and a formal account on representing motives and motiva-
tion that derives from psychology. In Section 3 we elaborate a simple agent model that
integrates handling of motivation. We go on in Section 4 by giving a formal account on
the dynamics of motivation and in Section 5 we consider the unreliability of the envi-
ronment and discuss its influence on the motivational model of the agent. In Section 6
we review related work in Section 7 we conclude.
2
Agents and Motives
The
BDI
approach is a well-established approach to model rational agents. This model
distinguishes between
beliefs
,
desires
,and
intentions
in order to represent human-like
reasoning and behavior. In this model, beliefs represent the agent's (subjective) knowl-
edge about itself, the world, and other agents. Desires describe what the agent is longing
for in its environment and intentions account for its currently pursued goals and the in-
tended course of action. Given some percept from the environment the agents usually
employs some form of belief revision or update in order to incorporate the new infor-
mation into its own beliefs. Afterwards, by taking the current state of the world into
account, the agent considers its desires and selects some desire to be pursued as a
goal
.
It appropriately updates its intentions using means-end reasoning and planning tech-
niques [3], selects some course of action, and eventually performs some action in the
environment. In general, this process is repeated indefinitely.
While most formalizations of the
BDI
model [10] assume the agent's desires to be
given, a more natural as well as flexible and powerful approach, demands for the possi-
bility to generate desires. The research in psychology describes how desires are created
