From a.rosta@xxxxx.xxxx Sun Dec 19 06:51:47 1999 X-Digest-Num: 315 Message-ID: <44114.315.1739.959273825@eGroups.com> Date: Sun, 19 Dec 1999 14:51:47 -0000 From: "And Rosta" From: "Jorge Llambias" > > Here's another argument against the expansion, > "for all x, John knows that x is...", this time not involving > names. > > Let's say this morning John bought two bottles of > milk and put them in the fridge, and just to make it > simpler let's say the fridge was otherwise empty. > Now, Mary tells John that she took one of the bottles > out of the fridge. She obviously doesn't tell him which > one, because nobody cares which one she took. > > Does John know what is now in the fridge? > Yes, he knows that there is a bottle of milk in the fridge. > > Is it true that for all x, John knows whether x is in the fridge? > No, he only knows that one of the two bottles that he > put in the fridge is there, but he doesn't know which > one (nor does he care). > > So, in this case at least, we cannot take the quantifier out > of the intensional context. Lovely! However, I think this is only a call for a refinement of the proposal, not a rejection of it. Clearly John also doesn't care whether or not the fridge also contains the odd microbe, or 2 micrograms of potassium, or whatever. The solution must reflect the idea that, out of a set of cognitively-relevant possibilities, John knows which is in the fridge. Provisionally: For every x such that x is a cognitively-relevant candidate, John knows whether x is in the fridge Next we need to capture the fact that two different bottles of milk don't count as different candidates if either is the only bottle of milk in the fridge: ... a cognitively-relevant candidate, namely (i) a mass consisting of one bottle of milk such that all bottles of milk in the fridge belong to this mass, (ii) a mass consisting of two bottles of milk such that all bottles of milk in the fridge belong to this mass, (iii) ... (mmmmmmmcccccc) a mass consisting of 453 eggs ... My slightly but not very tentative conclusion, then, is that the solution to the problem you raise is an instance of the general phenomenon whereby most universal quantifications need some (implicit or explicit) restriction to relevant candidates. --And.