The value of introspection to the designer of mechanical problem solvers

Abstract

Lately there has been a flurry of interest in constructing artifacts that will behave like living organisms in some useful respect. Since some living organisms are remarkably efficient and successful in performing certain tasks, e.g., sensing very weak signals, recognizing very complex patterns, or traveling over very rough terrain, we are advised to look at Nature's design for suggestions about how to build machines to do such things. But we are immediately up against an awesome obstacle: Nature's design is no easy thing to get at. I t would seem that God is a master at concealing and protecting company secrets. The obstacle seems to be especially awesome for those so brash as to wish to provide machines with intelligence. How do we get at Nature's design of intelligence? Psychologists study this design-though not for the purpose of copying it. This paper considers borrowing and adapting, for just this purpose, a method which psychologists have developed for observing how human beings solve problems.