Unrevisability

Christopher Hill (Brown University)

Many philosophers have been persuaded by Quine’s arguments that all sentences are empirically revisable, and have been led by these arguments to share his skepticism concerning the idea that human beings have a priori knowledge. This paper is a critical response to Quine’s views about revisability. In building my case, I will be relying heavily on the idea that words play roles of various kinds in such cognitive activities as explaining, predicting, classifying, problem solving, planning, reasoning, evaluating, and telling and appreciating jokes.  Many of these roles are empirical in character, in the sense that they either presuppose empirical claims or have such claims as their outcomes. This is true, for example, of roles associated with explanation and prediction. I will maintain, however, that various other roles are empirically innocent, and that the beliefs and dispositions that support these innocent roles are often safe from empirical challenge. An example is the role played by "true" in enabling generalized endorsements of sentences. We would have need of a term that played this role no matter what evidence we encountered. Because of this, and because the T-sentences must be in place in order for "true" to play the role, it behooves us to hold the T-sentences immune to empirical revision.