The Role of General Ability in Cognitive Complexity: A Case Study of Expertise
Stephen J. Ceci and Ana Ruiz
Although theories about generalization (or transfer) have abounded since the time of Aristotle DeAnima, the scientific debate over it did not take shape until the beginning of the 20th century. At that time, researchers at the two main centers of "associationism" in the United States, the University of Chicago and Columbia University, argued about the nature and developmental course of transfer. At Columbia, Edward Thorndike ( 1905) attacked the then-popular belief in the "theory of formal disciplines," which alleged that training in one discipline enabled students to think more rationally in other disciplines. Learning Latin, for example, was thought to lead to a better understanding of English, not simply because these two languages shared many cognates but also because learning Latin was regarded as an exercise that promoted the development of logical reasoning. (Learning chess was also promoted in some quarters for the same reason.) Reasoning is reasoning, so the thinking went, and therefore learning how to reason in one context was thought to transfer to reasoning in other contexts. Thus, it was the presumed ubiquity of transfer that was responsible for its appeal. It represented a parsimonious way of accounting for the obvious fact that humans do not require explicit learning of all matters.
Thorndike ( 1905) challenged the view that transfer was ubiquitous by arguing instead that learning in one context only transferred to another context to the extent that the number of "identical elements" in two contexts overlapped. His model was a neural one, suggesting that the neural pathways that were excited by each element could be important because the overlap in these excited elements was responsible for transfer, when it does occur. His evidence came primarily from subjects' performance on simple perceptual discrimination tasks on which they had received prior training on one form and were then asked to solve a related but physically dissimilar form. For example, subjects who were experienced at estimating the area of a particular geometric shape were not sucessful at estimating the area of other shapes, despite similarities of parameters ( Thorndike & Woodworth, 1901). Thus, Thorndike's view of the limits of transfer stood in contrast not only to the supporters of formal disciplinary training but also to those who favored a broader view of transfer that went beyond stimulus-response, or feature overlap, theory. Foremost in the latter camp was Charles Judd at the University of Chicago.
Judd ( 1908) argued that the degree of transfer was a function not of the number of neural paths shared by the physical features of two tasks but rather of the overlap in meaning attached to the two tasks. His argument ran along the lines of much of contemporary cognitive psychology and seemed to foreshadow Bartlett's ( 1932) notion of "effort after meaning" (i.e., that individuals strive to impose some sort of meaning on their perceptions). For Judd, the individual inter-