In the previous post, I defended a restricted form of modal skepticism and I concluded that many traditional philosophical issues could not be resolved and should be set aside. One may wonder what is left for philosophers to do: Am I suggesting to close philosophy departments?
Fear not, dear reader, there is a lot of things left for philosophers, as I explained in this last post. Many philosophical projects do not require any knowledge of metaphysical necessities: non ideal theories in moral and political philosophy, rational reconstruction (a la Reichenbach in philosophy of science), critical theory, for instance. What’s more, modally immodest philosophical issues have philosophical counterparts that do not require any knowledge of metaphysical necessities, and those should be what we philosophers focus on. I illustrate these counterparts in Chapter 6 of Philosophy Within Its Proper Bounds by examining various ways of studying causation and multiple realizability.
In addition to these modally modest philosophical projects, Philosophy Within Its Proper Bounds defends a philosophical activity that was once central to philosophy, but has since been much maligned: conceptual analysis.
Now, the version of conceptual analysis I defend—naturalized conceptual analysis—has little to do with the versions of conceptual analysis that have been debated in recent decades (Jackson, Chalmers). On my view, concepts are psychological entities; analyzing them often requires empirical, including experimental means (what I call “the method of cases 2.0”); and conceptual analysis does not result in a priori knowledge. It is nonetheless philosophically significant.
In line with Doing Without Concepts, I take concepts to be psychological entities. Furthermore, I delineate them psychologically: I distinguish the beliefs or belief-like states (which I call “bliefs”) that constitute them from those that don’t using a psychological property of our access to memory. The bliefs that are constitutive of a concept are accessed by default, that is, roughly, in a context-insensitive manner. This psychological distinction between the concept-constitutive bliefs and those that are not contrasts with the appeal to semantic (analyticity) or epistemic (a prioricity) properties that is common in philosophy. Because concepts are psychologically delineated, analyzing them leads no a priori or analytically true knowledge.
Analyzing concepts then just is characterizing the bliefs that are constitutive of a concept. People will have slightly different concepts of, e.g., justice, but at least in a linguistic community these concepts will be very similar. Because concepts are not transparent, empirical research is needed to determine what bliefs constitute them.
What is the point of conceptual analysis? Conceptual analysis matters for several philosophical projects, but the project I find the most important is therapeutic: Naturalized conceptual analysis helps philosophers identify concepts that are in some sense or other invalid. Philosophers have long been concerned with particular forms of invalidity: concepts that lead to sorites, TONK-like concepts, ambiguous concepts, etc. Naturalized conceptual analysis is the heir to this tradition. I also highlight another form of invalidity, which I call “empirical invalidity”: Concepts are empirically invalid when their introduction and elimination conditions result in unreliable inferences. Identifying empirical invalid concepts requires both conceptual analysis 2.0 and empirical knowledge about the domain these concepts are about. The conceptual analysis of the lay concept of innateness is used to illustrate these claims in Chapter 7 of Philosophy Within Its Proper Bounds
In addition to naturalized conceptual analysis, philosophers ought to engage in conceptual engineering (which I call, following Carnap) “conceptual explication.” Conceptual engineering is the modification of an existing concept. I distinguish Carnapian from Gramscian explication. The former is concerned with remedying epistemic flaws in our concepts: obscurity, imprecision, etc. The latter is concerned with remedying non-epistemic flaws: Perhaps some concepts are intrinsically morally wrong or they may be morally wrong because of the thoughts and actions they lead to. Ether way such concepts should be identified and modified, and this is the job of the conceptual engineer. This brand of conceptual engineering differs from the kinds currently discussed in the lively debate about conceptual engineering (Cappelen, see also this recent conference) in being psychological and in avoiding reference to the meaning or semantic content of our concepts.
Again, empirical, including experimental, work is important for successful conceptual engineering: It plays a role in identifying those concepts that need to be explicated, and it is important in the assessment of our explications. Many, including Carnap, hold that similarity with the engineered concept is a virtue of an explication, perhaps because of the concern that the more different an explication is from the concept to be explicated, the more likely it is to have undesirable side effects (a form of conservatism). In any case, experiments can play a role in assessing the similarity to the original concept of various competing explications (Schupbach).
So, philosophers, stop worrying about metaphysical necessities and modally immodest philosophical issues. Analyze concepts, engineer concepts!