GRAND CHALLENGE PROJECT (GC-5)
ARCHITECTURE OF BRAIN AND MIND


THIS FILE IS also at
http://www.cs.bham.ac.uk/research/cogaff/gc/otherbackground.html
It contains material that was previously in this file http://www.cs.bham.ac.uk/research/cogaff/gc/

NOTE:

Since this file was produced several of the documents referred to in it have been removed or moved. This is particularly true of the DARPA documents. Apparently DARPA has no concern to preserve the history of its activities and deliberations.

Other background material

Submissions to Grand Challenges Workshop

109 proposals were submitted prior to the November workshop, and a subset of the proposers were invited to the workshop to discuss the offerings and come up with a small collection of grand challenge proposals. They were divided into four panels. I was in Panel D which was concerned with biological modelling. Two proposals emerged from that panel, proposal GC-1 (In Vivo <=> In Silico), and Architecture of brain and mind This web site is about GC-5.



Grand Challenge proposal from Panel D
Architecture of brain and mind

A subset of the members of Panel D, led by Prof Mike Denham at Plymouth University agreed to work on developing a proposal based around the idea of designing a child-like robot, combining low level and high level approaches.

Simon Colton at Imperial College, London, set up a web site for this project here: http://www.doc.ic.ac.uk/~sgc/grand_challenge
which includes the summary edited by Mike Denham (14th Jan 2003) http://www.doc.ic.ac.uk/~sgc/grand_challenge/proposal.pdf

DARPA Cognitive Systems workshop (November 2002)

There is a Cognitive Systems programme funded by DARPA in the USA, whose goals are closely related to this proposed grand challenge. Information about it is available here
http://www.darpa.mil/ipto/
http://www.darpa.mil/ipto/Solicitations/PIP_02-21.html
http://www.darpa.mil/body/NewsItems/pdf/iptorelease.pdf

One of the documents http://www.eps.gov/spg/ODA/DARPA/CMO/BAA02-21/SynopsisP.html states:

Cognitive systems have a number of differentiating characteristics, but in short, they can be characterized as systems that know what they are doing. Such systems will have the ability to reason about their environment (including other systems), their goals, and their own capabilities; this reasoning will be able to take advantage of substantial amounts of appropriately represented knowledge. Cognitive systems will be able to learn both from their experiences and by being taught. They will be capable of natural interaction with humans, and will be able to explain their reasoning in natural terms. They will be able to form teams that succeed at goals that cannot be achieved as individuals. And, notably, they will be robust in the face of surprises provided by the environment; the avoidance of the "brittleness" of the expert systems of the past will be a distinguishing feature

A small workshop to discuss the DARPA project was held in Virginia 3-6 November 2002.

The white papers and slide presentations are (or were) available at the workshop web-site: http://www.dsic-web.net/meetings/oy8guwod/index.html


Related work at MIT

Closely related work is being done by Marvin Minsky and Push Singh at MIT. See their symposium report (PDF)
http://web.media.mit.edu/~push/StThomasSymposium.pdf

And a paper by Minsky, Singh and Sloman arising out of that symposium http://web.media.mit.edu/~push/StThomas-AIMag.pdf

Push Singh's thesis and other related documents http://web.media.mit.edu/~push/


Maintained by A.Sloman
Updated: 29 Dec 2006