February 10, 2004

Emptying and Knowing

I was rereading Knowledge and the Flow of Information for the paper on testimony and agency, and I ran across a passage in which Dretske compares knowledge with being empty. To be empty is to contain no objects, and to know is to be free of uneliminated doubts. But what counts as an object, and what counts as a doubt, depends on your purposes.

Well, we were just discussing some of Jason Stanley's arguments against contextualism. One was that, unlike other context-dependent words like "tall" or "big," "know" doesn't admit of degree modifiers--you can't say "I know that 2 + 2 = 4 more than you do."

Could the verb "empty" be a word that is clearly contextually dependent but doesn't admit degree modifiers? It seems odd to say

(1) *I emptied the salt shaker more than I emptied the pepper shaker

unless you mean "more often." But "empty" is clearly context-dependent. Whether "I emptied the salt shaker" counts as true will vary, depending on whether you meant to put as much salt in your food as possible (in which case it's OK to leave grains of salt clinging to the sides) or whether you mean to fill it with sugar (in which case you'd better have got those grains out). So maybe "empty" could serve as a model for the context-dependence of "know."

It might make sense to say

(2) I emptied the salt shaker more thoroughly than the pepper shaker

But doesn't that constitute an admission that you didn't really empty the pepper shaker? Similarly, I don't find it completely weird to say

(3) Jamal knows when the movie is more securely than Jerome does

say, if Jerome has read the listing in the paper, but Jamal has actually called the theater. At least, I don't know what else you'd say about such epistemic overkill.

Posted by Matt Weiner at February 10, 2004 07:25 PM

I think I point out in "On the Linguistic Basis for Contextualism" that embedded questions under "know" (like "know where Texas is" or "know how to ride a bicycle" or "know when the movie is") are gradeable, since what is being said when one grades such constructions is that someone has a more complete answer to the embedded question than another person.

Posted by: Jason Stanley at February 11, 2004 06:33 AM

If Bruce knows how to ride a bicycle better than Sheila knows how to ride a bicycle, does this really mean that Bruce has a 'more complete answer' to the question, 'Do you know how to ride a bicycle?' Surely, for both Bruce and Sheila, the answer is just 'yes'. (Unless, perhaps, Bruce can trump all by adding, 'yes: I can do a no-hands wheelie and 10 foot jumps riding backwards', or something suitably flashy. Let us suppose, however, that Bruce and Sheila are merely suburban day riders.)

I suppose that Bruce might answer, 'yes: *very well*, thank you', whereas Sheila may answer, 'yes, but *not as well as Bruce*'. But do these answers class as more or less 'complete' in any real sense? Perhaps in this case by 'more complete' we mean 'of a greater degree of conviction', for instance, as regards Bruce's response to the question. However, then the notion of gradeability has not been explicated by the 'more or less complete answer' analysis, for the notion of gradeability (of conviction, for instance), appears in the anlysis itself.

Posted by: Will Davies at February 11, 2004 08:08 AM

We're obviously impinging on Jason's paper on "knowing how" with Timothy Williamson. I don't know that paper well, so I'll have to let Jason speak for himself about how that plays in.

Except I should say that I think the embedded question is "How do you ride a bicycle?" not "Do you know how to ride a bicycle?"

It seems to me that the gradability of knowing how is more clear than the other cases. On the received view, on which knowing how is connected with (or synonymous to) ability, it's simple to account for that; Bruce knows how to ride a bicycle better than Sheila does because Bruce can ride a bicycle better than Sheila can. The more complete answer view seems like it can handle this as well.

I also see the applicability of the more complete answer analysis to "knows where Texas is"--one person can say "In the southwestern U.S." and the other can say "South of Oklahoma " etc. But it doesn't seem to me to apply to (3)--both Jamal and Jerome can give no more complete an answer to "When is the movie?" than "At 7:30." On the other hand, "?Jamal knows when the movie is better than Jerome does" sounds deviant to me, so maybe this supports

Posted by: Matt Weiner at February 11, 2004 09:21 AM

I agree that the relativity oddity of grading knowing-when locutions is due to the fact that usually one just knows that a certain time is the time in question, and it's odd to think of one person knowing that information in a more complete manner than another person.
On "emptied" -- one can say "He completely emptied the trash can"...I'll have to think more about comparisons with "emptied" though...

Posted by: Jason at February 18, 2004 08:27 AM