E.G.

This post is part of my gradual progress toward an argument against knowledge closure.

Consider four propositions:

1. I know that I saw a zebra at the zoo.
2. I don’t know that what I saw wasn’t really just a cleverly disguised mule at the zoo.
3. If I know that I saw a zebra at the zoo, then I know that what I saw wasn’t a cleverly disguised mule at the zoo.
4. 1, 2 and 3 are inconsistent.

(This is a stolen example.) 

If I’ve just been to the zoo and had a look at the zebras, I probably am inclined to say that the first proposition is true.  But I may also be inclined to assent to 2, at least once 2 is presented to me for consideration.  3 seems plausible (3 is a consequence of knowledge closure combined with plausible background assumptions).  Finally, 4 seems very plausible; 4 appears to be a rather straightforward logical truth.  So there are reasons to accept 1 through 4.  But 1 through 4 are inconsistent, so we will have to deny at least one of them.  Which one should we deny?

Contextualists can deny 4.  Contextualists about knowledge think that the truth conditions for knowledge ascriptions are context-dependent; on the contextualist account, the same ascription of knowledge could be true in one context and false in another.  So, at least in certain contexts (e.g. "everyday contexts"), 1 might be true, even though in other contexts (e.g. "skeptical contexts"), 2 might be true.  In that case, even if 3 is true in every context, 1-3 are not inconsistent, provided that 1 and 2 aren’t both true in the same context.

Invariantism is the denial of contextualism.  If one were an invariantist, would one need to affirm 4?  I want to show that the answer to this question is "no."

Contextualism is obviously true of certain propositions.  "Michael Jordan is tall," for instance, may be true in the context of a discussion about the people on a city bus, but false in the context of a discussion about professional basketball players.  In both cases, the same concept, "tallness," is employed; it just so happens that "tallness" is the sort of concept for which context-dependence is "built in."  Contextualism about knowledge is the view that context-dependence is built into knowledge.  It is this "conceptually built-in" context-dependence which allows the contextualist to affirm 1 through 3 without contradiction.

"A&M Savings is a bank" is a sentence which superficially resembles "Michael Jordan is tall."  If I say "A&M Savings is a bank" in the course of a conversation about "business establishments in which money is kept for saving purposes," then I may well have said something true; but if I say "A&M Savings is a bank" in the course of a conversation about "slopes of land adjoining a body of water, especially a river," then I probably have said something false.  So "…is a bank" appears to be context-dependent in the same way that "…is tall" is.  The crucial difference, however, is that the single word "bank" is used to invoke two different concepts (i.e., the concept of a certain kind of financial business, and the concept of a certain kind of geographical feature), whereas "tall" always invokes the same concept.  "Bank," therefore, has two senses, neither of which is context-dependent, whereas "tall" has a single, context-dependent, sense.  (It may turn out that "bank" does exhibit a certain kind of context-dependence.  If so, it is only because the context in which the word "bank" is uttered may help to fix which sense of "bank" is intended by a speaker. )

One ought not be a contextualist about the predicate "…is a bank."  Rather, one ought to be (what we might call) a "sense pluralist" about "…is a bank."  That is, one ought to think that the word "bank" has multiple senses, each of which is context-independent.

One could be a sense pluralist about knowledge, too.  That is, one could say that there is more than one concept which the word "knowledge" represents.  If that were the case, then 1 and 2 might both be true, even if 3 were also true, so long as the sense of "know" employed in 1 is different from the sense of "know" employed in 2.  So a sense pluralist about knowledge can deny 4 even though she denies contextualism.  But since a sense pluralist is not a contextualist, she is an invariantist.  So it follows that invariantism does not commit one to affirm 4.

If one wants to affirm 1-3, then one needs to deny 4.  Both sense pluralism and contextualism provide the means to deny 4.  But one needs an excuse to deny 4, because ordinary speakers tend to think that 4 is true.  Question: Who can muster the best excuse — the contextualist or the sense pluralist?

The contextualist can produce an excuse to deny 4 by analogy with certain context-dependent adjectives like "tall."  To the ordinary speaker, "Michael Jordan is tall and Michael Jordan is not tall" looks like a contradiction, because ordinary speakers are accustomed to assume that both sides of a conjunction are meant to be understood within the same context.  This assumption is usually correct.  After all, it is "conversationally inappropriate" to say "Michael Jordan is tall and Michael Jordan is not tall"; it is more appropriate to say, e.g.: "Michael Jordan is tall compared to regular people, but not tall compared to typical basketball players."  But conversational inappropriateness is not contradiction, although it sometimes appears to be.

The contextualist can argue that, likewise, the appearance that 1-3 are inconsistent derives from the conversational inappropriateness of their presentation.  1 is true within an "everyday" context, while 2 is true within a "skeptical" context, and it is inappropriate not make these contexts explicit.  But this does not show that there is any contradiction which results from omitting explicit acknowledgement of the differing contexts.  Indeed, if the analogy with "tall" succeeds, then we should think that there probably is not any contradiction.

The problem for the contextualist, however, is that the analogy with "tall" does not seem to succeed.  It is obvious that "tall" is context-dependent; it is not anything like obvious that "knows" is context-dependent.  Many ordinary speakers will be satisfied with our explanation of how "Michael Jordan is tall and Michael Jordan is not tall" does not have to produce a contradiction, but significantly fewer of them will be able to see that 1-3 might not produce a contradiction.  So to come up with an adequate excuse to deny 4, the contextualist cannot just provide an analogy with "tall."  The contextualist must also give an explanation of how the context-dependence of "knows" could be so difficult to see, given that the context-dependence of "tall" can be so obvious.

Now for the sense pluralist’s excuse.  Just as the contextualist can provide an analogy with "…is tall," the sense pluralist can provide an analogy with "…is a bank" to show how the appearance of a contradiction might be generated by 1-3 without there being any actual contradiction.  If she takes this route, the sense pluralist will have a problem similar to the one faced by the contextualist: The sense pluralist will need to explain why sense pluralism about knowledge is so difficult to see, while sense pluralism about "…is a bank" is so obviously true.  But it is here, I think, that the sense pluralist has resources which are unavailable to the contextualist.

Question: How do we know that "bank" has more than one sense?  Suppose I am just learning English.  One day, I hear my parents say that a building where money is kept is a bank.  The next day, I hear my teacher say that a slope alongside a river is a bank.  These might seem to be two very powerful clues that "bank" has more than one sense.  But why should they be?  Why shouldn’t I think that there is just one concept to which the word "bank" points, but that that concept is so broad that its extension can include both a building where money is kept, and a slope alongside a river?  I claim that we would not be able to tell that there is more than one sense of "bank" simply by observing the way competent speakers use that word.  These observations might lead us to guess that "bank" has more than one sense; but I suspect they could lead us instead to guess that "bank" has just one, extremely broad, sense.

So: How do we learn that there is more than one sense of "bank"?  I am not sure, but I suspect it helps that we are able to provide analyses of the two senses of "bank."  We can list features which would make something a "bank" in one sense, and list features which would make something a "bank" in the other sense.  We can then compare the two lists, and notice that they are not the same — that the items on one list do not appear on the other list, and vice-versa.  It is the availability of such differing analyses which, I think, provides one with the means to know that there is more than one sense of "bank."

It is possible — perhaps probable — that no correct, non-recursive analysis of knowledge is available.  If so (and if I am right that analysis is the primary direct way to distinguish different senses of the same term), then it is possible that we simply do not have the tools to determine that there is more than one sense of knowledge.  So, it seems to me, the sense pluralist might have a way to explain why sense pluralism about knowledge is so difficult to see, while sense pluralism about "…is a bank" is so obviously true.  We can see that sense pluralism about "bank" is true because we can analyze the different senses of "bank."  But no analysis of knowledge appears to be possible, so if there is more than one sense of knowledge, we might never be able to tell.