Friday, June 15, 2007

Defining "Cognitive Systems"

In December 2003, the Cognitive Systems Foresight project (commissioned by the Office of Science and Technology, UK government) formally ended. A number of outcomes were achieved, as described by the following bullet points (copied from here):

• Helping to define an emerging field of new science - cognitive systems - to which life and physical scientists are contributing, with the emerging definition of it being:
• Identifying relevant groups of UK-based scientists and engineers, based in industry and academia, and bringing them together for interactive workshops at which common ground was established.
• Identifying exemplar "grand challenges" that constitute fertile ground for further
research.
• Creating an accessible summary of the work done during the year, with the assistance of a consultant science writer (Michael Kenward) and distributing this to a wide range of relevant groups (CUGPOP, members of Research Councils and their Research Boards, selected scientists in other countries working on similar topics). A book is also in preparation.
• Establishing "Foster Parents" (professional societies) to carry work forward in the years ahead, a framework of "Foresight Fellowships" (to be funded by EPSRC and MRC) for small cross-disciplinary pilot projects, and an "Action Plan" to ensure
self-sustaining research networks.


One of the most fundamental discussions that took place as part of the project concerned the definition of "Cognitive Systems": after all, without a generally accepted consensus on this term, it would be difficult to make progress in its name. The definition which was eventually settled on is as follows:

"Cognitive systems are natural or artificial information processing systems, including those responsible for perception, learning, reasoning, decision-making, communication and action"

This is a very broad definition, but having said that, it is also (I think) uncontroversial, and may form the basis upon which further constraints and qualifications may be added (such as, for example, those proposed by Clark and Grush, as I've looked at previously).

No comments: