t from the idea of the "general factor," or g; nor is the task-oriented focus of such a reduction limited by the confines of the machine intelligence theory.
Machine intelligence is a behaviorist view of intelligence, in that it reduces intelligence to a stimulus-response relation. If given the stimulus of "2 + 2," an intelligent first-grader will perform the algorithm involved in solving the problem. Machines perform these tasks extremely well; they have been "taught" or programmed to do math, play chess, and even carry on conversations. The ELIZA program, designed to mimic psychotherapists, supplied responses using a set of rules such as key word association. For example, when told, "My mother drinks too much," ELIZA responded with, "Who else in your family drinks too much?" (Gould, 1981, 369)
The limitations of such a model for describing human intelligence became obvious to researchers. ELIZA was not intelligent in the sense that she understood the conversation; she was merely responding to triggers like the key word "mother." Some element was missing from the machine intelligence theory.
A pattern was slowly emerging: It was easiest to program computers to do types of work that most people found very hard, such as long, detailed mathematical calculations, whereas it was most difficult to program machines to perform operations of which even a small child or mentally defective adult was capable, such as understanding language or recognizing objects or situations (Dehn & Schank, 1982, 353).
Theories of machine intelligence focus on the processes that are involved in performing a task, in any brain or central processing unit. Spearman's g, on the other hand, is a differential theory of intelligence, in that it seeks to define intelligence through the study of individual differences.
Spearman g represents a single underlying entity of intelligence which affects performance on all types of mental tests (Gould, 1981). Spea...