Information Technology Reference
In-Depth Information
Fig. 2.
Agent decision-making in a feedback loop of behavior observation, expectation, action
selection, and operation
commitments) provides a promising foundation for self-organization as a paradigm for
multiagent coordination.
Nevertheless, in the aforementioned approach [1,14], expectations reflecting and
guiding agent behavior are modeled by the system designer as an external observer.
However, self-organization requires organizational structures to emerge from the sys-
tem's operations without external intervention; i.e., the mentioned feedback loop must
be closed within the multiagent system. Thus, the next section introduces the notion
of
double contingency
which describes the emergence of mutual expectations structur-
ing communication flows between agents appearing as black boxes. In the following,
this concept is operationalized in order to demonstrate its ability to enable autonomous
coordination of agent communication systems.
4
Expectation-Based Self-Organization
According to the sociologist Niklas Luhmann,
double contingency
denotes both the
fundamental problem of social order generation as well as its own solution leading to
the emergence of social order [11, pp. 103-136]. Referring to Parsons and Shils [15], he
points out: Given two entities
alter
and
ego
, mutually appearing as black boxes to each
other, ”if alter makes his action dependent on how ego acts, and ego wants to connect
his action to alter's“ [11, p. 103], they reciprocally block their ability to act at all.
However, the solution to that problem lies in the interdependency of actions, as well.
As soon as alter or ego behave in whatever way, action becomes not only possible, but
social structures emerge from the self-referential circle of mutually dependent activi-
ties. Those structures consist of expectations evolving from, e.g., ego's observation of
his own actions as well as of alter's behavior. These expectations, in turn, guide ego's
selection of subsequent actions. Hence, a feedback loop of observation, expectation,
selection, and operation (action) emerges as depicted in Figure 2.
In the context of multiagent systems, double contingency can be viewed analogically
as the problem of determining interaction opportunities. It also denotes its own solution
through the emergence of expectations guiding agent communication as the fundamen-
tal operation in MAS. As a starting point serves the simulation model by Dittrich et
al. [3]: They simulate and analyze Luhmann's concept of double contingency in a sce-
nario of two agents interacting with each other by exchanging messages with varying
