Daniel Greco (Yale University), "Idealization in Epistemology: A Modest Modeling Approach"
Oxford University Press
In Idealization in Epistemology: A Modest Modeling Approach, I argue that model-building is ubiquitous in epistemology, and that this fact is of central epistemological significance. Of course, epistemologists have been building models for decades—representing belief change using the mathematical formalisms of probability theory and modal logic is a well established practice. But while the practice of modeling in epistemology is old, metaepistemological reflection on the significance of that practice is not. In my book I try to demonstrate that the fact that epistemologists build models isn’t merely a metaepistemological observation that can leave first-order epistemological debates untouched. Rather, once we view epistemology through the lens of idealization and model-building, the landscape looks quite different.
“It’s just a model” is sometimes used as a shield to blunt the force of an objection, but it’s a shield that can seem to diminish its bearer. In using it, one seems to concede that one’s sights are less lofty than they might be. “I’ll leave to other, more ambitious thinkers, the task of coming up with the final, comprehensive philosophical theory of knowledge (or justification, or evidence, or rationality, etc.); I’m just building models.” This can sound like an abdication of the mete and proper role of philosophy.
To the contrary, I argue that that model-building is likely the best we can do as epistemologists. Once we start using epistemological categories like belief, knowledge, and confidence—even when we don’t embed them in explicitly formalized mathematical models—we’ve already entered the realm of idealization and model-building. So a non-idealized, comprehensive and complete theory of these phenomena is an oxymoron—akin to a non-idealized theory of frictionless planes. We can object to a model of knowledge by pointing to a better model, but in the absence of a better model, the fact that a framework for epistemological theorizing involves simplifications and idealizations—the fact of its status as a model—is not in itself objectionable.
I distinguish between two ways we can see model-building as fitting into inquiry. The first approach, which I call “ambitious,” involves aspiring—at least in principle—to the attainment of a complete and fully accurate theory of the domain being modeled, and hoping that the models one builds in the meantime will provide some insight into what that merely hypothetical theory might look like. The ambitious modeler thinks that a perfect model exists somewhere in Plato’s heaven, and hopes that the imperfect, idealized models she herself builds will provide some clues as to what that Platonic model might look like. On the second, “modest” approach to modeling, we can be agnostic or skeptical concerning the existence of a perfect model, and we don’t need to think of the tractable models we construct and analyze as providing clues about its nature. Rather, the modest modeler is content to work with a collection of models, each partial and less than fully accurate, without holding out hope for a grand unification on the horizon.
I draw on extant work in philosophy of science to point out that modest modeling is generally treated—by both philosophers of science, and by practicing scientists—as a reasonable default throughout the special sciences, while ambitious modeling is much rarer; arguably, it exists in fundamental physics but almost nowhere else. I go on to argue that in the relevant respects, epistemology is much more like economics and other social sciences than like physics; beliefs are more akin to prices than to particles. As a result, I argue that modesty in epistemological modeling is a reasonable default. This makes available to me an argument schema that I use throughout the book, applied to a variety of debates within epistemology. It goes something like this.
First, I interpret views about phenomena of epistemological interest—information, belief, knowledge—as modeling frameworks. Chapter 3 concerns the possible worlds framework for modeling information, chapter 4 concerns the Bayesian framework for modeling belief and learning, Chapter 5 considers the relationship between decision theory and folk psychology, and chapters 6 and 7 consider modeling frameworks that allow for trivial iteration of higher order belief and knowledge (Chapter 6) as well as the possibility of common knowledge (Chapter 7). In each case, I interpret objections that are ordinarily intended to show that some theory is false—that some general principle has a counterexample—as attempts to show that some modeling framework cannot adequately capture its target phenomenon.
Next, I argue that these objections can be met—the phenomena can be captured—so long as we’re willing to be modest about the scope of our models. The objections show that no single model built by the framework in question can capture all the cases we want to. But we can capture the phenomena the objections point to by using different models to handle different cases.
My argument schema bears a family resemblance to contextualist views in epistemology; contextualists will often defend a principle by arguing that putative counterexamples involve a kind of subtle equivocation; while it initially seems as if they’re committed to jointly inconsistent claims, once they note that the claims are made it different contexts, the contradiction is resolved. But while my strategy is superficially similar to that of contextualists, my explanation for why that strategy is legitimate is quite different. Traditional versions of contextualism are motivated by linguistic analogies—contextualists argue that “knows” behaves like other context-sensitive pieces of language such as “tall” or “rich”, while their opponents point to striking dissimilarities. By contrast, I don’t focus on linguistic analogies between epistemological vocabulary and other context-sensitive pieces of language. Rather, I focus on analogies between the theoretical roles of epistemological categories, and the theoretical roles of similar categories in sciences where a modest modeling approach is attractive. The economist who uses different, incompatible models of the macroeconomy in different theoretical contexts, or for different practical purposes, needn’t embrace any distinctive semantic theses about “macroeconomy”, “growth”, or “inflation.” Rather, she merely accepts that the macroeconomy is an intractably complex system, and that no single model can best illuminate every aspect of it. The biologist who uses a plurality of models of species, genes, or organisms, needn’t embrace any distinctive semantic theses about “species”, “gene”, or “organism”. Similarly but more controversially, I claim, the epistemologist who uses different, incompatible models of knowledge in different theoretical contexts, or for different practical purposes, needn’t embrace any distinctive semantic thesis about the word “knows.”
Finally, in chapter 8 I turn my focus to debates about the relationship between “ideal theory” and “non-ideal theory” in epistemology. I argue that because idealization is inescapable in epistemology, there isn’t a feasible, general project of “de-idealizing” the field. While particular models can fruitfully be made more realistic or detailed along this or that dimension, such projects always need to be defended one at a time, on a case-by-case basis; we’re all ideal theorists, and inevitably so.
thanks