Cognitively Useful Fictions

FAIN: FA-50227-04

Catherine Z. Elgin
President and Fellows of Harvard College (Cambridge, MA 02138-3800)

Mainstream epistemology contends that truths are the only things that are cognitively acceptable. I think this is a mistake. Models, idealizations, and approximations are integral to science, philosophy and economics. They afford an understanding of their subjects, but are not strictly true of them. I am writing a book in which I argue that such devices function not as representations of facts, but as fictions. They are cognitively useful because they exemplify, and thereby afford epistemic access to, features they share with their subjects. To make my case, I will develop a theory of fictional language and delineate the role of exemplification in fiction. I will show how fictional devices function in the context of factual theories, what they contribute and how their interpretation is constrained. I will explain why in the context of testable theories such devices advance understanding and why in the context of purportedly factual but untestable theories they do not. Acceptable theories turn out to have a more complicated, textured symbolic structure than epistemologists tend to think.