Ernest Sosa

Geoffrey Sayre-McCord puts before us an interesting and original line of thought. Here is its main structure: Naturalist semantics would bring important benefits to ethics. But it has very high costs. Fortunately, we can secure such benefits without the costs, by substituting, for the natural kinds of naturalist semantics, a set of moral kinds determined not by scientific but by moral theory. I find myself stumped by the preliminaries at , however, which need further support, or so I will argue in section I. Section II will assume that the concerns of section I can be met, and will discuss the rest of the argument independently of that

Belief is considered a kind of performance, which attains one level of success if it is true (or accurate), a second level if competent (or adroit), and a third if true because competent (or apt). Knowledge on one level (the animal level) is apt belief. The epistemic normativity constitutive of such knowledge is thus a kind of performance normativity. A problem is posed for this account by the fact that suspension of belief seems to fall under the same sort of epistemic normativity as does belief itself, yet to suspend is of course precisely not to perform, certainly not with the aim of truth. The paper takes up this problem, and proposes a solution that distinguishes levels of performance norrmativity, including a first order where execution competence is in play, and a second order where the performer must assess the risks attendant on issuing a first-order performance. This imports a level of reflective knowledge that ascends above the animal level.