I’m currently in the midst of studying for a candidacy exam on a very specific topic: idealized scientific representations. A scientific representation or model is considered idealized, according to one influential account, if (a) it includes or asserts some falsehood, (b) this falsehood simplifies the model in some relevant way, e.g. to make calculation easier, or to make explanations clearer, and (c) the falsehood approximates some relevant truths about the system(s) being represented. Idealization, then, is a specific type of misrepresentation, distinct from useful fictions and plain falsities (since idealizations must at least approximate the truth). Examples of idealizations in science include (by no means exhaustively): Bohr’s model of the hydrogen atom, the ideal gas law, the model of the ideal pendulum, astronomical models that treat the Earth as a volumeless point-mass, ecological models that assume strict regional barriers, the economic assumption of perfect access to information about prices, and just about every line of best fit ever drawn.
There is a fast-growing literature on this topic, which is understandable given that all modern sciences seem replete with idealized representations. Indeed, according to some people working in this area, the methodological novelty that Galileo (arguably the founder of modern science) hit upon was not simply the use of mathematics to represent nature, but the use of idealized mathematical representations. Regardless of its origins, idealization seems to be a practice that lies near the foundation of today’s scientific practice. Given the pervasive use of idealized representations in science, there are many worthwhile philosophical questions to be asked about it. For example, what notion of “approximate truth” is appealed to here? How can we learn things by making false assumptions? What are we learning about the world through idealized representations?
The issues surrounding idealization that interest me most usually involve the application of idealized representations in critical settings, e.g. when crafting public policy on the basis of predictions derived with idealized models. Unsurprisingly, applying scientific representations that are known to include simplifying falsities makes everyone feel a bit uneasy, especially when the success of such applications really matters, and even more so given that we have some clear cases of idealization gone awry.
Continue reading →