The ability to simulate complex cognitive processes is a major accomplishment in itself. But what is more significant is that information-processing models developed for quite different problems (for example, a model to prove theorems in geometry versus a model to describe neurotic behavior) have many common component processes.

This commonality of processes among model suggests that a general theory of complex cognitive processes may not be too far away, Simon and Newell (1964) were the first to isolate some of the components common to many information- processing models and pull them together into a single model, which they dubbed the General Problem Solver (GPS).

They propose that we should be able to combine the special information of any particular task (chess playing), theorem proving, music composition) with GPS and come up with a composite program that can solve the task, using strategies and tactics of the type employed by human beings.

The GPS simulates in a formal manner what the individual is believed to do when attacking problem. The programs are very complex and can be characterized here only in the barest outline. The actual program for GPS is built around two basic processes that follow each other in repeated cycles until the problem is solved, or abandoned as too difficult or insoluble.

ADVERTISEMENTS:

The first process, part of what is called the problem-solving organization, is to set sub goals that might be appropriate to the solution of the problem. These sub goals are then evaluated, and a promising one is selected for exploration. Note that this is a kind of “executive”, or “decision-making”, function, including both a search and an evaluation phase.

A sub goal, for example might, be the solution of a simplified version of the more general problem. Once the executive routine selects a sub goal, the process known as the mean-end analysis applies relevant heuristics to reach the sub goal.

This requires that the information-processing mechanism begin with data that are given and follow permissible transformations as in ordinary problem solving. Because the heuristic approach does not guarantee a solution, if the initial approach fails, the executive routine then searches for sub goals that appear more productive.

The approach to complex cognitive processes exemplified by GPS is quite promising. If it turns to that information-processing models based on only a few basic methods of symbolic representation and a small number of elementary information processes can simulate complex human behavior, then we have truly advanced our understanding.