[ACT-R-users] Gauging Interest for AAAI Fall Symposium on Multi-Representational Architectures

[Unmesh Kurup]

**Apologies if you receive multiple copies of this email**

We are planning on organizing a symposium for the 2009 AAAI Fall 
Symposium series on Multi-Representational Architectures for 
Human-Level Intelligence. The abstract for the symposium is at the 
end of this email. At this time we are trying to gauge interest in 
this topic from the broader AI/CogSci community. If you think you 
would be interested in submitting to/attending such a symposium, 
please send me an email with your name and affiliation.

thanks,
Unmesh Kurup
Postdoctoral Researcher
Dept of Cog Sci, RPI
110 8th St, Troy NY 12180

Chairs: Unmesh Kurup(RPI) and B. Chandrasekaran(Ohio State)
Organizing Committee: Bonny Banerjee(Securboration), John Laird (U. 
Michigan), Scott Lathrop (USMA), Marvin Minsky (MIT Media Lab), 
Samuel Wintermute (U. Michigan)


Abstract
Multi-Representational Architectures for Human-Level Intelligence
---
A multiplicity of representational frameworks has been proposed for
explaining and creating human level intelligence.  Each has been proven
useful or effective for some class of problems, but not across the board.
This fact has led researchers to propose that perhaps the underlying
design of cognition is multi-representational, or hybrid, with different
subsystems with different representations and processes interacting to
produce the complexity of cognition. The first proposals in this mold
were by attempting to integrate connectionism and symbolic representation
frameworks to cover a larger range of phenomena than either could, at
that time.  Since then other notions of multi-representationalism have
been put forward: including versions of Soar and ACT-R, Polyscheme as
well as calls for multi-modal representations as part of architectures.
This seems a good time to review the state of the research on
multi-representational frameworks in producing complex intelligent
behavior.  While the idea of multiple representations seems to gather
widespread agreement, there are puzzling issues about the underlying
theoretical framework:  What count as different representations? Are
there different types of "hybridness", e.g., representational framework A
being a substrate for representational framework for B, vs A and B
working side by side and interacting?  What are the design principles
that govern how a problem is decomposed and allocated to the appropriate
representational subsystem?

The main aim of the symposium is to bring researchers in different types
of multiple representations to explore the above questions. We hope to
examine working applications based on hybrid representations with the
hope of characterizing how the match between the tasks and the
representations in these systems.