© 2010, Tim Graf
[From the introduction]
Not all true beliefs count as knowledge; warrant is the technical term for the (possibly logically complex) property that gives you knowledge when you add it to true belief. Warrant is not necessarily the same as justification, or any other epistemological property; all that this stipulative definition seems to entail about warrant is something akin to the following pair of biconditionals:
(1) A true belief counts as a piece of knowledge iff it is warranted.
A warranted belief counts as a piece of knowledge iff it is true.
What exactly warrant turns out to be depends upon the substantive theory of knowledge we adopt. It may involve a belief’s being justified and having an absence of genuine defeaters for that belief, a belief’s being formed through a reliable process, a belief’s being formed through the virtuous traits of an agent’s cognitive character, or some such.
Some epistemologists, however, have attempted to theorize about warrant by offering “nonpartisan” arguments, or in other words, arguments that do not depend upon any substantive theory of knowledge. For example, Merricks (1995, 1997) argues for warrant infallibilism, the claim that a belief cannot be at once warranted and false; Ryan (1996), Howard-Snyder et al. (2003), and Coffman (2008) challenge his arguments without endorsing warrant fallibilism. Pust (2000) argues that some of the claims made about warrant by the philosopher who first used the term as a technical concept in epistemology, Alvin Plantinga, are groundless, given the stipulative definition of the term. Lastly, Michael Huemer (2005) tries to demonstrate several claims about the nature of warrant: that there is no unique warrant property, that warrant fallibilism is true, that some but not all warrant properties are closed under known logical entailment if knowledge is, and that no warrant properties are closed under known logical entailment if knowledge isn’t.
In this paper I will first summarize Huemer’s arguments and then critique his conclusions. Our stipulative definition of warrant needs to be modified somehow to eliminate some of the intuitively problematic candidates for warrant conditions that his analysis allows. What complicates our task is the fact that, as Huemer (p. 117) points out, given the way warrant is defined, it is not the kind of thing we can have clear intuitions about. Or in Pust’s words, “If we had the ability to make independent intuitive judgments about the term’s proper and improper application, there would be no justification, prior to analysis, for stipulating that warrant has a certain functional role. After all, we might find that such a characterization was mistaken,” (p. 51). All we can know about warrant at the outset is that it fills the functional role of making up the difference between knowledge and mere true belief. We may have intuitive ideas about what does that, but the thing that actually fills that functional role may turn out to be far different from what we originally thought.
Graf, Tim (2010) "Restrictions on Warrant Properties," Res Cogitans: Vol. 1: Iss. 1, Article 11.