The Productive Roles of Falsehoods in Science
The use of falsehoods is a common, if rarely recognized, part of many good scientific practices. In many cases, there is no mystery behind the use of a falsehood. For instance, when scientists approximate some physical value—say, that an electron has a charge of exactly 1.6EE-19 coulombs, rather than 1.60217....EE-19—they are incorporating something that is not exactly true. This use of falsehoods is, of course, not a problem—it is widely understood that the scientist has rounded, and, for her part, the scientist has very good reasons to suppose that the approximation will not significantly affect her results. Similar reasoning permits the use of such classic elements of models as a frictionless plane, a massless pulley, a point mass, etc. These idealizations are conceptual analogues to numerical approximations: we have good reasons to suppose that the differences between our model and the world are negligible. My interest lies predominantly in those falsehoods which cannot so easily be classified as conceptual idealizations or numerical approximations—what philosophers have come to call fictions.
Most philosophers currently studying scientific fictions (e.g. Suarez, Winsberg, Teller, Giere) identify fiction by pointing to properties of the scientific community—a model is fictional just if it is used as fictions are used, or if it was created with the intent that it be used as a fiction, or if most scientists regard it as a fiction, etc. While I find such analyses illuminating, I also find them unduly limited to well established fictions, and thus must remain silent when there is no scientific consensus about whether something is (or ought to be used as, or is intended for use as) a scientific fiction. To complement these sociological accounts, I have examined those properties of a (part of a) model that indicate that it is properly thought of as an idealization—and so should be interpreted as, in some way, approximately true—or as a fiction—in which case it is more properly thought of as merely empirically adequate.
My central project is to come up with a set of epistemic tools that allow us to analyze a model to determine the nature of the various irrealities involved so that we may determine the appropriate epistemic attitude to have regarding that model. Specifically, I look at how a model's ability to accurate predict, explain, or describe the world is affected by making the various posits that constitute that model more or less realistic. I use this tool to differentiate between idealizations, different kinds of fictions, and abstractions in such a way that we can say more about the model than how the community of researchers use it. This investigation is driven by a case-based approach, in which actual scientific practices in classical mechanics, mechanical engineering, quantum statistical mechanics, and evolutionary game theory are used as a foil each step along the way, illustrating the appropriate and inappropriate uses of the various ways that models may fail to describe the world, but, nonetheless, be truth-conducive.
Purves, G. M. (2012). The productive roles of falsehoods in science. (Unpublished doctoral dissertation). University of South Carolina, Columbia.