Computer Science as Empirical Enquiry:

Symbols and Search

Allen Newell and Herbert Simon


What is a physical symbol system?

The answer to this question is best described in Newell and Simon's own words.

"A physical symbol system consists of a set of entities, called symbols, which are physical patterns that can occur as components of another type of entity called an expression or symbol structure."

Two important notions have also been defined:

Designation- An expression designates an object if, given the expression, the system can either affect the object itself or behave in ways depending on the object.

Interpretation-The system can interpret an expression if the expression designates a process and if, given the expression, the system can carry out he process.

At any instance of time the system contains a number of such symbol structures including processes that operate on the expressions so as to create, destroy or modify them or other routine jobs such as "move so many steps as given in this symbol". I shall proceed to use the above notions with an example. The Random access memories of microprocessors are good examples of such systems. Typically, successive locations have "instructions" (symbols) and "data" (also symbols). The system may "designate" the data value as a memory location. Interpretation means carrying out the designated process. One such symbol may be "interpreted" as an instruction that asks the microprocessor to take the symbol in the next location and place it in a temporary location such as the accumulator.

The authors include symbols from human symbol systems in their definition of symbol. Thus they include humans in their definition of a physical symbol system.


What is the Physical Symbol System Hypothesis?

The authors hypothesize that a physical symbol system has the necessary and sufficient means for general intelligent action. What they mean is that (1) any system that exhibits intelligent behavior may be shown to be a physical symbol system. Also (2) a physical symbol system, organized well enough, can exhibit intelligence in the same scope as seen in human action.

What role does the study of people have in testing this hypothesis?

Some of the contemporary models on human intelligence have the characteristics of a physical symbol system. Researchers in the fields of information-processing, cognitive science, cognitive psychology and other fields relating to the functioning of the brain, benefit from using the "information-processing" or "computer" metaphor. A number of theories exist that model human behavior with symbol systems. This has even had a big impact on research in AI, since AI researchers seek to use the very same processes on their machines that are being uncovered and defined by cognitive psychologists. The symbiosis offers strong supporting evidence to part (1) of the physical symbol system hypothesis.

What role does computer science have in testing this hypothesis?

Developments in Computer Science offer strong evidence that support part (2) of the claim that is made in the physical symbol system hypothesis. In fact, the field of Artificial Intelligence may be viewed as a quest to build computer systems that can exhibit intelligence in the same scope as seen in human action. Lot of contemporary examples of such systems exists and any intelligent action maybe mimicked, at least in theory. Many successful AI programs corroborate the hypothesis. What remains to be seen is whether there are any limitations in the Physical symbol system model, by virtue of which, the system is incapable of exhibiting certain kinds of "intelligent behavior". Such examples would undermine the hypothesis.