A philosopher goes into the armchair and brings back knowledge. What world have they been exploring? What is this knowledge of, and how did they find it? These are questions that philosophy, the most methodologically self-conscious of all the disciplines, can’t help but ask itself over and over again. They put the armchair philosopher in a rather uncomfortable spot. It seems to be difficult, even impossible, to find answers on which armchair philosophical investigation is a respectable pursuit—answers according to which it is both reliable (or at least, justified) and able to uncover substantive facts.
I’ll focus on a particular form of armchair inquiry, sometimes called the method of cases, sometimes conceptual analysis, but which I prefer to call philosophical analysis. The goal of philosophical analysis is to take some category—such as knowledge or causality—and to inquire into what I call its “essential nature”, namely, what makes it the case that a specimen belongs to the category. Analysis tests hypotheses about essential natures by considering particular cases and comparing what a hypothesis says about the case— whether it counts as knowledge or causation, say—with what seems to be the intuitively correct judgment. The hypothesis is considered correct only if it correctly reproduces the intuitive judgments about every possible specimen.
Let me pose my opening questions again, this time with philosophical analysis specifically in mind. What kinds of things are facts about essential natures? How can we come to know them in the armchair, without any input from the senses?
One response proposes that our mental representations of philosophical categories are built around definitions that spell out what it takes to belong to the category, that is, that stipulate the category’s essential nature. Being stipulations, they cannot be mistaken—and there they are, lying in the head, ripe for the taking. For some reason, however, they will not yield to direct introspection. (This idea goes back to the early days of conceptual analysis in the 18th century.) That is not to say that they are mentally impotent. On the contrary, they direct our judgments about individual cases. We judge something to be a member of a category if and only if it satisfies the corresponding mental definition. Our case judgments are, as a consequence, a reliable guide to the contents of our definitions, and so to the essential natures of the categories.
If this is the way that analysis works it is not so clear, however, that the essential natures will be of any great interest. They are determined by mental definitions that are by assumption set up in advance of any concerted philosophizing. As such, it seems that they are little more than a first pass at taxonomizing the subject matter—at picking out interesting or significant features of the epistemic, moral, or metaphysical landscape. Just as our Paleolithic ancestors’ first attempts to categorize marine life may have yielded a “fish” concept that included dolphins and whales but excluded seahorses and eels, so you might expect that our preliminary attempts to carve up the epistemic or metaphysical domains would fall short of grouping together those things that go together deep down. In that case, philosophical analysis, though feasible, would give us facts whose significance is more anthropological than philosophical. By understanding philosophical analysis as the excavation of mental definitions, we secure the reliability of analysis at the cost of its substance.
The alternative is to suppose that the mental representations guiding our judgments about cases are not definitions but rather corrigible hypotheses about the category in question. This opens up the possibility that our philosophical concepts work more like natural kind concepts than like definitional concepts—that they have a tendency to latch onto something interesting, even if our initial beliefs about the category in question are impoverished or false.
On this view, philosophical analysis might after all be worth doing; the essential natures learned through analysis would, on the whole, be significant matters of fact. But in gaining substance, we seem to have lost reliability. If our beliefs about a category are fallible, then the case judgments made using those beliefs are also fallible. When we conclude that a Gettier case does not exemplify knowledge, we might simply be mistaken. In the case of natural kinds, we can rely on experience to correct false beliefs and so to render our case judgments more reliable over time. But in armchair philosophy, we have only the judgments themselves to go on. It seems we can get no epistemic traction.
My new book Thinking Off Your Feet (Harvard University Press, 2019) embraces the analogy with natural kinds and argues that philosophical analysis conceived in this way can be (and can be seen to be) reliable as well as substantial. I’ll sketch some of the book’s key moves on the Brains blog this week. Tomorrow I’ll draw on recent work in cognitive psychology to say what it means for a philosophical concept to be in the relevant respects like a natural kind concept. Wednesday I’ll confront the problem of reliability, and I’ll tackle substance on Thursday. Friday will be a surprise.