| 
| Question of the Day                          Thursday, November 21, 1996
| ------------------------------------------------------------------------
| What is the physical symbol system hypothesis?  Who posed it, when and
| what impact did it have on the evolution of AI?
| 


[ The answer below was written by Patrick. It was the only one (what is
happening?!) I received. I think it answers the first part of the 
question very well. As to the second part (impact on AI) I hoped to
generate some discussion...]

The physical symbol system hypothesis was proposed by Simon and Newell in 
1976.  The hypothesis says that a physical symbol system is both 
necessary and sufficient for general intelligence.

(A physical symbol system is one in which a set of physical patterns,
called entities, can be combined into larger entities called expressions; 
the system also contains processes for creating, destroying and modifying
these expression entities.  At any given time the system contains some set
of these entities, and the set of such entities evolves over time. 
Lastly, the system operates in a context wider than itself.)

(Note: My apologies for not answering the second half of the question; I wasn't paying attention. The physical symbol system hypothesis has had a most profound impact on AI, and most researchers either implicitly or explicitly assume that an intelligent system will be one as specified by this hypothesis. It has lead to reactionary writings (by Brooks, most notably) arguing that explicit knowledge does not form a part of biological intelligence, though that writing itself is now being counterattacked (by Albus and others). This hypothesis ranks with the knowledge representation hypothesis as one of the cornerstones of modern traditional AI thinking.)

Back to the Question of the Day Page

Patrick Doyle	November 1, 1996