Suppose we are motivated to develop a theory of rationality because of our awareness of our own fallibility: truth isn’t transparent to us, and so what we should track as doxastic beings is signs or marks of truth, and a theory of rationality is just an account of what these signs or marks of truth involve. When the signs point one way, we should believe the claim in question; when they don’t, we shouldn’t. And we inquire into such matters because truth isn’t transparent.
Suppose, though, that one is a fallibilist about marks of truth as well as truth itself.
That is, suppose that, no matter what condition C a theory of rationality says is strictly correlated with rational belief, condition C is no more transparent to us than is truth itself. (By this, I don’t mean that transparency comes in degrees; I’m only giving a premise that entails that condition C is not transparent on the basis of what we already know about the transparency of truth itself). Then the same motivation for developing a theory of rationality in the first place kicks in again to motivate the development of another normative theory, a theory that allows that a belief can fail to be rational and yet be normatively acceptable in some weaker sense. We might say that the weaker notion is that of excusable belief, but nothing much turns on our choice of vocabulary.
As is obvious, however, regress looms. The epistemological heart can only rest once a transparent property of belief is found, and for full fallibilists, there is no such property. This fact raise the interesting question of where the mistake occurs. I’ll leave that issue to another post, though comments on the issue are always welcome.