L. S. Carrier

Bredo C. Johnsen1 misconceives my strictures concerning acceptance of the following principle : If A both knows that p and knows that p entails q, then A can come to know that q.Johnsen seems unaware that my criticism was intended to apply only after is made to appear in its most plausible light; that is, only after its consequent is interpreted as: ’It is logically possible for A to know that q.’ Without this interpretation might be dismissed simply on the grounds that A suffers from some physical or psychological disability that prevents him from drawing inferences from what he knows.Properly interpreted, remains acceptable as long as the propositions substituted for p and q are such that it is at least logically possible for A to get evidence enough to make them known. Agreement on this point is itself enough to render Johnsen's own examples irrelevant. For instance, even though it may be physically impossible for A to get adequate evidence that in the constellation Andromeda there is a planet intermediate in size between Venus and Earth, the foregoing is still a fit substitution instance for q; but since such a q does not suffice to falsify the consequent of, it does nothing to generate any skeptical argument, either.

In this article it is argued that it is impossible to give a reductive analysis of knowledge, given that knowledge is an "epistemic" concept with these marks: (1) like necessity, it is only partially truth-functional; and, (2) unlike necessity, it includes an "intentional" component (belief) which is completely non-truth-functional. a reductive analysis would have to contain at least one extensional component, one intentional component, and none that is itself epistemic. but any plausible analysis then turns out either to be non-reductive, e.g., causal, or else falls prey to "gettier" counterexamples. my positive suggestion is that we eschew reductive analyses of epistemic concepts in favor of non-reductive analyses.