Software Blending by a Connector-based Multiagent
System
Fauconnier and Turner have proposed a mechanism, known
as Cognitive Blending, to explain key cognitive processes involved in the
construction of meaning and the organization of new knowledge in humans.
In January, 2003 a software research team at the Naval Postgraduate School
used a computational model based on multiple agents to achieve limited blending
by software. This presentation will explain the extent to which blending
is taking place, describe the methods and techniques that were combined to
achieve this result, list the software problems that were overcome thus far,
and outline the objectives for the next phase of our work. This result
was part of an application-level project that explored the feasibility of
using software-based Cognitive Blending to complement Rational Choice
models as a means of anticipating the decisions of others. The presentation
will conclude with a description of work-in-progress for this application.
John Hiles
U.S. Naval Postgraduate School
jhiles@mindspring.com