|
|
SPEAKER: E. Dante Suarez
TITLE: On a Hierarchically Decomposed Agent
AFFILIATION: Professor of Finance, Trinity University
TIME: Wed June 22, 12:30p
LOCATION: 624 Agua Fria Conference Room
Pizza will be available
ABSTRACT
I will informally elucidate my ideas on a general theory of collective behavior
and structure formation, with a resulting architecture that can be broadly
applied. The proposed model represents a decomposition of intent, based on the
idea that an agent?s behavior, whether it represents an individual or a group,
can be seen as an emergent property of a collection of intertwined aims and
constraints. I consider a disentangled agent that is formed by multiple and
relatively independent components. Part of the resulting agent?s task is to
present alternatives, or ?fields of action? to its component selves.
Correspondingly, the composed agent is itself constrained by a field of action
that the superstructure to which it belongs presents. Just as the original
divided agents are modeled as units, the superstructure possesses a certain
amount of cohesion, and can thus be ascribed agency; its independent parts could
be consciously or evolutionally constructed and aligned.
This hierarchical representation intends to reveal behavior as bounded and
bounding, exemplified by an agent that is abstractly defined. The most important
aspect of such an agent is that it maximizes its objective function by
definition, given the constraints of the nature of its components and the
superstructure to which it belongs. According to this proposition, suboptimal
behavior is often misclassified because we do not recognize the actual
maximizing active agent.
The flexibility of this model can be used to describe groups or societies in
interaction. It also allows us to understand experimental evidence that humans
cooperate more then they should according to theories that define them as
unitary. The theory to be developed would explain complex organization taking
into account all aspects of subset reorganization, which depend on the
particularities of the entity. This view must include a measure of group welfare
beyond Pareto optimality, as well as definitions of the concepts of a behavioral
function, optimal scales and design, bordered maximization, cohesion, and
identification processes.
|
Locked
