From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 10 19:07:58 1999 Date: Wed, 10 Feb 1999 19:07:54 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Do names have primary intensions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Some thoughts on the "minutes". >Our discussion seemed to focus on two issues, broadly conceived. >(1) What are primary intensions? We spent a lot of time trying to >work out whether proper names, such as London, have primary >intensions, and if so, >> what are they. There was some discussion of >whether 'London' and 'Londres' have different primary intensions, if >they have primary intensions at all. Actually, I discuss the 'London'/'Londres' case a bit in "The Components of Content". Everyone should read that paper, since it gives a lot more detail on the 2-D framework, though it is primarily cast in terms of concepts, not words. Do proper names have primary intensions? As I said yesterday, they arguably don't have PIs that are universal to all users of a name. It may be that two different users have different PIs for "London", just as for "Bill Clinton", etc. So this might lead one to say that names don't have PIs as a matter of "semantics", where the semantics of a term is supposed to be universal to all users of the term. Still, I think that it is plausible at least that any name has a PI for any user on any occasion of use. We might put this by saying that "name tokens" have PIs (or are associated with PIs), even though "name types" may not. If so, we can arguably see PIs at least as part of the semantics of an "utterance" (cf. Reimer's distinction), and certainly as part of the semantics of an underlying thought. Why think that name tokens have primary intensions? Basically, because like any referring word, a name (token) must have some pattern of application across epistemic possibilities. Just say I use the name "Godel". Then I can consider all sorts of epistemic possibilities about the way the actual world might be, and I can see how the name will refer with each of them. Think of Kripke's own methodology in Lecture 2 -- we consider e.g. the epistemic possibility that the incompleteness of arithmetic was proved by a guy called "Schmidt" and stolen by a guy called "Godel" who published it, moved to Princeton, etc. Kripke notes, in effect, that if that epistemic possibility is actual, our term "Godel" refers to the Princeton guy. That's to say that the primary intension of "Godel" picks out the Princeton guy in the relevant centered world. One can do the same thing for a very wide range of worlds considered as actual (i.e. considered as epistemic possibilities): for a large numbers of such worlds, there seem to be clear facts about what our words will pick out if those worlds are actual. We can think of the primary intension as the "reference-fixing" conditions of our terms, if we like, though actually I prefer to think of it as giving the epistemic application-conditions of our terms, or the "epistemic profile" (as opposed to the "modal profile" of the secondary intension). It's central to the way language works that any referring term has an epistemic profile (at least for a given user on a given occasion): there are facts about how it applies to different epistemic possibilities, and about how it will refer if those possibilities turn out to be actual. How does one evaluate the primary intensions of names that one uses? Basically, take a name, e.g. "London", take a given centered world, and ask oneself "to what does the name refer if that world is actual"? Here, one considers the world as actual, i.e. considers it as an epistemic possibility: "what if the world actually turns out to be that way?". And for a very wide range of such worlds, we have clear intuitions about how the name refers. E.g., if the actual world turns out to contain XYZ in the oceans etc, then we'll say that "water" refers to XYZ. If the actual world turns out to have the Princeton guy stealing the proof from Schmidt, we'll say that "Godel" refers to the Princeton guy. If the actual world turns out to have a non-whale-eaten guy at the other end of a causal chain from our use of "Jonah", we'll say that "Jonah" refers to that guy. Etc, etc. Summing up a primary intension in langauge is often difficult, and the same goes for these cases; what really matters is the function from worlds to referents, not any capsule summary. But if I were to try to get at some of the things that are involved e.g. in the PI of my name "Godel", we might try: "the guy called 'Godel' who's at the other end of a causal chain from my use of the name", or something like that. That's imperfect, as Kripke argues -- e.g. one can consider epistemic possibilities in which the guy my term refers to wasn't called "Godel" at all (I've got his name wrong, or some such). But that just means that we have to refine our view of the PI. Importantly, Kripke's own methodology here relies precisely on evaluating how the term will refer if a given epistemic possibility turns out to be actual, i.e., on evaluating the PI of the term at a world. As for "London" and "Londres": think of Pierre's situation. For him, there are lots of epistemic possibilities. The world he thinks he is in is one with a beautiful faraway city by the name of "Londres", and an ugly one close at hand called "London". Pierre quite reasonably says that if that epistemic possibility is actual (as he believes it is), then his terms "London" and "Londres" name two different cities: "London" picks out the faraway city and "Londres" the city close at hand. So in that centered world, the PIs of "London" and "Londres" give different results. On the other hand, another epistemic possibility for Pierre (at least a broad epistemic possibility in the sense articulated earlier) is the actual possibility: i.e., that the people he got the term "London" from were actually referring to the same city he's living in now, and that it has both beautiful and ugly parts, etc. If Pierre were confronted with the hypothesis that this epistemic possibility is actual, he should rationally conclude that under that hypothesis, his terms "London" and "Londres" pick out the same city. So that's to say that in this centered world (Pierre's actual world!), the PIs of "London" and "Londres" give the same results. So, Pierre's "London" PI and his "Londres" PI give the same result on some centered worlds (including the actual world), but different results on other worlds (e.g. the one he thinks is the case). So they are at least slightly different intensions overall. How to summarize these intensions in language? Again, it's difficult and imperfect, but we might make a first attempt by saying that his "London" PI picks out "the beautiful city I've heard of under the name 'London'", and his "Londres" PI picks out "the ugly city I'm living in", or perhaps "the city I've heard of under the name 'Londres'". Getting the details right will depend on careful consideration of cases, and will depend on just how Pierre's conceptual system is set up, etc. But this gives us enough to see how the PIs might pick out the same extension in one world (the actual world) but different extensions in some other worlds. The fact that Pierre's PIs are different precisely reflects the fact that it is epistemically possible for him that London isn't Londres. This suggests a general principle: when "A=B" is a posteriori (for a user), A and B have different PIs. To see this, one can go through the following reasoning. (1) "A=B" is a posteriori; so (2) it is (broadly) epistemically possible that A is not B, so (3) there is some epistemic possibility in which my terms "A" and "B" pick out different things (when that possibility is considered as actual) so (4) there is some world in which the PI of "A" and the PI of "B" yield different extensions so (5) A and B have different PIs. I hope all the steps here make sense. You can illustrate it by thinking of the London/Londres case, or the Hesperus/Phosphorus case. Note that the principle here is equivalent to: when "A=B" is a posteriori, "A=B" has a contingent primary intension. (N.B. the primary intension of a statement is just the obvious generalization of the primary intension of a term -- in the book I call this a "primary proposition", but I now tend to use the unified terminology.) If A and B have different PIs, there's a centered world where they pick out different extensions, and in which the PI of "A=B" is false. And to say that "A=B" has a contingent PI is just to say that there's a world in which the PI is false. This is an instance of a very important general principle: if a statement S is a posteriori, S has a contingent primary intension. It's fair to say that this is at the heart of the two-dimensional account of a posteriori necessity. The central claim of this account is if a statement S is necessary a posteriori, S has a contingent primary intension and a necessary secondary intension. That's not too hard to illustrate by working through the standard Kripke cases. I note that there are just possibly some counterexamples to this principle which someone might put forward, though they probably won't be the standard Kripke cases. We'll be discussing potential counterexamples later in the course, but feel free to suggest any now. But anyway, one can see how all this hooks in with the need for names to have PIs. Presumably a lot of statements involving names will be a posteriori for many or all users, e.g. "London = Londres", "Cicero = Tully", etc. That's just to say that for a given user, there is the epistemic possibility that the statement is false, i.e., there are some scenarios such that if they turn out to be actual, the statement turns out false. So the very a posteriority of the statement indicates that the names in question have distinctive patterns of applications to epistemic possibilities, i.e. that they have (distinct) PIs. That's enough for one message. Hopefully it gives useful background for thinking about issues in the rest of the minutes. More on the rest later. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 12 01:05:35 1999 Date: Fri, 12 Feb 1999 01:05:21 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: Primary intensions and beliefs To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Brad writes: >This question I think is partially addressed by Dave in his last >posting, but here goes. I was wondering what constitutes a primary >intension. In the examples we've discussed, such as "London" and >"water", it seems that it is the beliefs that an agent has involving a >particular concept which are constitutive of that concept's primary >intension. So, for example, it is because Pierre believes that Londres >is pretty and perhaps has various perceptual beliefs from photos (I >think those kind of details were in the thought experiment?) that the >primary intension of "Londres" differs from that of "London" (about >which Pierre has very different beliefs, such as "it is ugly" and "it is >dirty" and "the food there is awful"). Actually, I wouldn't say that beliefs "constitute" a primary intension, though there is a relationship between certain core beliefs and the primary intension. To determine what a PI is, one doesn't look at beliefs, but at the application-conditions of a concept to the world. What matters is what the concept picks out in any given epistemic possibility. So in Pierre's case, we look at what "London" would pick out for him if various epistemic possibilities turned out to be actual. That will be the primary intension. Nevertheless, there is some sort of relationship between an agent's beliefs involving a concept and the concept's primary intension. We can think of a primary intension is roughly corresponding to the agent's "a priori beliefs" or "unrevisable beliefs" involving the concept. For example, if a user's primary intension for "one meter" picks out the length of a certain stick, this corresponds to the fact that it is a priori for that user that the stick is one meter long. This might be seen as a "core belief" that corresponds to the primary intension. Note that not all of an agent's beliefs involving a concept will correspond so closely to a primary intension. Indeed, most beliefs won't be reflected in the PI at all. Most beliefs about meters, say -- that Olympic athletes run 100 meters, that there are about 1600 meters in a mile, etc -- and aren't reflected in the PI structure. This reflects that the beliefs in question are a posteriori, and could easily be given up if an appropriate epistemic possibility turns out to be actual. In Pierre's case, just which of his "London" and "Londres" beliefs will be "core" and which will be "incidental" to his PIs depends to some extent on his psychology. I actually suspect that his belief that Londres is pretty isn't reflected in the "Londres" PI at all. It could epistemically turn out that the place he had heard of in Paris is actually ugly, but he'd still say that "Londres" refers to that city. If so, the PI (the application-conditions of the term) doesn't require prettiness. Perhaps a core belief might be something more like "Londres is the city I heard of a while ago under the name 'Londres'", or something like that; it's harder to see an epistemic possibility in which he'd give that up. As for "London", I suspect that ugliness and dirtiness will likewise be incidental. Maybe "London is where I've been living recently" is a core belief (i.e. he picks out the referent precisely as the place he's been living), or maybe "London is the place people call 'London'" is, depending on just how Pierre's psychology works. In any case, I think it's a mistake to think about a PI too much in terms of beliefs. It's true that one can try to find core beliefs in the vicinity of a PI, but the fundamental thing is the pattern of application across epistemic possibilities. The beliefs that turn out to be "core" are just those that are seen to hold across any of these epistemic possibilities; but I'd argue that it's this pattern of application that tells us that the beliefs are "core", rather than vice versa. Otherwise, one can be confused by surface conditions of "coreness" that don't really reflect the deep epistemic profile of the concept. So beliefs are really secondary here. >There is an article by Georges Rey (1983, I think) in which he argues >roughly that we must distinguish the meaning of a concept from the >beliefs that an agent has *about* that concept. I gather that on the 2d >view Dave is giving, that distinction or separation isn't quite right. >Some beliefs (and this is made clear in Dave's last posting) are >constitutive of a concept's primary intension. And from Dave's last >posting, it seems to be the beliefs that the agent him/herself takes >(perhaps implicitly) to be constitutive. Well, yes and no, as I think the above makes clear. A few things to say. (1) "Beliefs about the concept" taken literally are fairly irrelevant -- e.g., my belief that I have the concept "London". It's beliefs *involving* (or *using*) the concept that are at issue here -- e.g., my belief that London is pretty. I take it that's what you and Rey mean. (2) I agree with Rey that what really matters is the meaning (the PI), not the beliefs, although there may be a correspondence between the meaning and certain beliefs. (3) As above, even at the level of correspondence rather than constitution, it's only some beliefs that are relevant (the core ones). All that being said, I think you're on the right track with what you say about the beliefs that matter being the ones that the agent implicitly takes to be constitutive. But again, I wouldn't want to *define* a PI in those terms. >I think that, in the end, this seems ok to me, though I had previously >been compelled by Rey's point. Perhaps Rey's point is applicable only >to secondary intensions, which seem to be what Rey means by "meaning". >There is the odd possibility that a person (say a chemist) comes to be >so entirely convinced that water is H2O that the belief that water is >H2O becomes constitutive of the concept "water". If future advances in >science reveal that water is really XYZ (suppose that it turns out that >the theory of the elements is entirely mistaken), on the 2d analysis my >hypothetical chemist would have to conclude that water does not exist. Hmm. I suspect that you're probably right that Rey's intended "meaning" is more like a SI. I think the best way to think about the chemist is not as someone who becomes really convinced that water is H2O. Even such a chemist (normally understood) would presumably concede that she was wrong if the stuff out there turned out to be XYZ. That would suggest that her PI isn't really so H2O-ish. Better to think of her as a theoretical chemist who has never had acquaintance with the stuff out there, who only knows molecular chemistry, and who uses "water" more or less as another name for H2O. Then if the stuff in the lakes turns out to be XYZ, this chemist won't care at all -- she might go on using "water" as a name for H2O (assuming H2O is still taken to exist on the new theory). For her, "water is H2O" really will be a priori. [Technical note: The issue here is complicated by semantic deference, as in Burge's arthritis cases. If the chemist is like many speakers in intending to defer to usage in her surrounding community, then I guess it couldn't be that "water is H2O" is a priori for her, and on making the XYZ discovery, strictly speaking her utterance of "water is H2O" would be false. But let's say the chemist doesn't much care about the way others use the term, and uses it her own way (with her own full mastery of her own concept). Then the PI will work the way I suggested. The residual issue (in the non-deferring case) is whether she could truly be said to have our word "water", or a different word altogether (one which happens to have the same sound and the same referent in the actual world). Some might argue that her failure to semantically defer, combined with her different usage, makes it a different word with a different meaning. Others might argue that the commonality in actual reference is enough, given that it's a natural kind term. They could arguably gain support from the way names work (it's epistemically possible that my "Paderewski" and your "Paderewski" could turn out to refer to different people, but as long as they actually refer to the same person, we're arguably using the same name). What to say about natural kind terms here is really a very subtle and difficult question. Fortunately it doesn't matter too much for our purposes.] >Someone following Rey, however, would not have to say this. Instead, >one could claim that the chemist's *beliefs* about water are entirely >distinct from what his concept of "water" actually means. So it doesn't >matter how strongly the chemist believes that water is H2O--if water is >really XYZ, the chemist simply has a false belief *about* water. Well, I'll agree with Rey that the mere strength of belief isn't enough, but I guess I do think that some beliefs matter somewhere. In the first chemist case, we probably would say the chemist has a false belief about water, because it would turn out that her PI lies elsewhere. Similarly for the deferential theoretical chemist. But in the crucial case, the nondeferential theoretical chemist, if the XYZ scenario turned out to be actual, we might say that she doesn't have a false belief about water at all, just a true belief about something that isn't water (though she calls it "water"). (Note that in all three of these cases, there's arguably some match between PI and belief. Arguably, the PI of the first case would involve "watery stuff" and of the second case might involve "what people in my community call 'water'. These plausibly correspond to *other* beliefs of the chemist, e.g. the belief that water is watery stuff or that it's called 'water' around here, which would then turn out to be the core beliefs. Of course in the third case the core belief would be that water is H2O, and the PI would pick out the H2O-ish thing in all worlds.) >I certainly don't think that the above worry is a serious problem for >the 2d account. It seems to me that the above might simply show that we >find it difficult to imagine (or rather, that it is unlikely) that the >primary intension of a person's concept of water would in fact come to >be constituted by a belief that water is H20. What beliefs *are* >constitutive of primary intensions, in general? That is, is there some >general account to be given of the difference between constitutive and >"incidental" beliefs (beyond the fact that the difference is that an >agent *takes* certain beliefs to be constitutive and others to be >incidental)? Perhaps the beliefs that are involved in the descriptions >used for "reference fixing"? The latter is more or less what I'd say. They can be seen as those involved in reference-fixing conditions, and might equally be seen as those that the agent would hold to be true no matter which broad epistemic possibility turned out to be actual. But again, the order of priority is not: give an account of what it is to be a core belief, and then define a PI in terms of that. Rather, its: define a PI in terms of application to epistemic possibiities (i.e. to worlds considered as actual), and then define core beliefs in terms of that. >On a related note, I wonder if anyone can explain the following footnote >from "The Components of Content" (my page 6, but I assume that varies): > >"But note that even Kripke's (1980) argument against description >theories presupposes something like a primary intension. It proceeds by >considering the referent of a concept such as "Godel" at various >actual-world candidates, thus evaluating the primary intension, and >arguing that any given description gives the wrong result. ...." > >I'm not entirely clear how this case shows that Kripke presupposes >primary intensions. Doesn't the "right result" for the reference of >"Godel", on Kripke's account, turn out to be that the referent has to be >*that man*--Godel! If I'm understanding Kripke correctly here, I can't >see how this uses something like primary intensions rather than simply >using something like a secondary intension--the referent in the actual >world. Hmm, remember that by this point, Kripke has moved away from the "modal" considerations in Lecture 1, and is now on to the "epistemic" considerations in Lecture 2. His point with "Godel" isn't to make the Aristotle/Hesperus/etc point that it picks out the same guy in all counterfactual worlds (that's the point about secondary intensions). His point is rather one about descriptions in the fixation of reference, and about epistemic possibility (so it's a point about primary intensions). In particular, he is taking a given description D (in this case "the person who proved the incompleteness of arithmetic" or some such), and he is arguing that it is not a priori that Godel is D. To do this, he considers an epistemic possibility (the one involving the proof being stolen) and suggests that if that possibility is actual, "Godel" picks out the stealer, not the prover. So he is in effect making the point that the primary intension (the epistemic application-conditions) of "Godel" are not given by D. And similarly for any other description, he claims. Note that his methodology presupposes something like a primary intension, in considering epistemic possibilities and evaluating reference in them. His claim is in effect that that the PI isn't given by a description, but rather works by returning whatever lies at the end of an appropriate causal chain from the word in question. (Of course Kripke himself wouldn't put things in those terms.) Note that this epistemic point, that it is not a priori that Godel is D, is very different from the modal point, that it is not necessary that Godel is D (i.e., that the secondary intension of "Godel" isn't given by D). The latter point Kripke certainly accepts, and corresponds to the sort of point he makes in Lecture 1, but by now he is making a different point. (See the beginning of lecture 2 for explicitness on this.) I take it that your point about "Godel" referring to *that man* across possible worlds is an instance of modal point, not the epistemic point. (In effect, it's the claim that "Godel" is a rigid designator.) The two points arguably come apart for a name like "Jack the Ripper", for which the modal point holds but the epistemic point arguably doesn't. Arguably, reference is fixed as "the person who committed the murders", and it's a priori that Jack the Ripper committed the murders (if he exists). Note that Kripke will still say that "Jack the Ripper" picks out *that very man* across possible worlds, so the modal point will still hold, even though in this case the epistemic point doesn't. Here, the primary intension of "Jack the Ripper" can be captured by a description, even though the secondary intension can't. Translated into the 2-D framework, the modal point is that for all names, the secondary intension doesn't correspond to any description (except trivial ones). The epistemic point is that for most names, the primary intension doesn't correspond to any description. As I noted earlier, the modal point is less controversial than the epistemic point. The distinction between the modal point and the epistemic point is subtle at first, but it's quite crucial to a full understanding of what Kripke's up to and of the 2-D framework, so people should fire away with questions if they're not quite clear on it. --Dave. P.S. Sorry about the unintended e-mail earlier. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 10 23:16:00 1999 X-Accept-Language: en Date: Thu, 11 Feb 1999 00:09:35 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Brad Thompson Subject: Primary intensions and beliefs To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO This question I think is partially addressed by Dave in his last posting, but here goes. I was wondering what constitutes a primary intension. In the examples we've discussed, such as "London" and "water", it seems that it is the beliefs that an agent has involving a particular concept which are constitutive of that concept's primary intension. So, for example, it is because Pierre believes that Londres is pretty and perhaps has various perceptual beliefs from photos (I think those kind of details were in the thought experiment?) that the primary intension of "Londres" differs from that of "London" (about which Pierre has very different beliefs, such as "it is ugly" and "it is dirty" and "the food there is awful"). There is an article by Georges Rey (1983, I think) in which he argues roughly that we must distinguish the meaning of a concept from the beliefs that an agent has *about* that concept. I gather that on the 2d view Dave is giving, that distinction or separation isn't quite right. Some beliefs (and this is made clear in Dave's last posting) are constitutive of a concept's primary intension. And from Dave's last posting, it seems to be the beliefs that the agent him/herself takes (perhaps implicitly) to be constitutive. I think that, in the end, this seems ok to me, though I had previously been compelled by Rey's point. Perhaps Rey's point is applicable only to secondary intensions, which seem to be what Rey means by "meaning". There is the odd possibility that a person (say a chemist) comes to be so entirely convinced that water is H2O that the belief that water is H2O becomes constitutive of the concept "water". If future advances in science reveal that water is really XYZ (suppose that it turns out that the theory of the elements is entirely mistaken), on the 2d analysis my hypothetical chemist would have to conclude that water does not exist. Someone following Rey, however, would not have to say this. Instead, one could claim that the chemist's *beliefs* about water are entirely distinct from what his concept of "water" actually means. So it doesn't matter how strongly the chemist believes that water is H2O--if water is really XYZ, the chemist simply has a false belief *about* water. I certainly don't think that the above worry is a serious problem for the 2d account. It seems to me that the above might simply show that we find it difficult to imagine (or rather, that it is unlikely) that the primary intension of a person's concept of water would in fact come to be constituted by a belief that water is H20. What beliefs *are* constitutive of primary intensions, in general? That is, is there some general account to be given of the difference between constitutive and "incidental" beliefs (beyond the fact that the difference is that an agent *takes* certain beliefs to be constitutive and others to be incidental)? Perhaps the beliefs that are involved in the descriptions used for "reference fixing"? On a related note, I wonder if anyone can explain the following footnote from "The Components of Content" (my page 6, but I assume that varies): "But note that even Kripke's (1980) argument against description theories presupposes something like a primary intension. It proceeds by considering the referent of a concept such as "Godel" at various actual-world candidates, thus evaluating the primary intension, and arguing that any given description gives the wrong result. ...." I'm not entirely clear how this case shows that Kripke presupposes primary intensions. Doesn't the "right result" for the reference of "Godel", on Kripke's account, turn out to be that the referent has to be *that man*--Godel! If I'm understanding Kripke correctly here, I can't see how this uses something like primary intensions rather than simply using something like a secondary intension--the referent in the actual world. From owner-modality@LISTSERV.ARIZONA.EDU Sun Feb 14 22:48:15 1999 Date: Sun, 14 Feb 1999 23:47:22 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Erik J Larson Subject: Re: Epistemic and modal issues To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Hello, I know we're not directly discussing PI's in relation to reference anymore, but lately I think I've managed to unravel what little understanding I (thought I had) on this issue. So, apologies for re-opening some discussion, but here goes. Alot of what I took Thony's concern with PI's to be centered on was the convergence of PI's to descriptive content when using them to fix reference. I take it that Kripke (obviously) wants to avoid this possibility, requiring only of PI's that they are part of a causal chain that begins in the referent and links up to the subject (Chalmers too--see p.59 of TCM). My problem is that, pulling PI's away from descriptive accounts and towards causal ones (or at least calling them "compatible" with such accounts) does not make the application of PI's in reference fixing any less mysterious. I'm not sure that we can have cognitive access to things like causal chains, and if so, I'm not sure how any of this would work. To put it simply, I don't see how something like "watery stuff" can fail to be descriptive (and therefore adopt all the problems of descriptive theories or reference), or (assuming it is not) how calling in something like a "causal chain" makes the application of a PI any clearer. If this is any objection at all then I suppose it would be sufficiently general to make reference fixing for SI's puzzling too. My general problem is that I can't seem to make sense of what we mean when we say "watery stuff" if it is not shorthand for some description of the properties of that stuff. If we want to call it a causal connection (involving some initial dubbing ceremony), then I guess I don't know how we have access to those sorts of connections such that we can accomplish reference fixing. The point of this is just to express my confusion with exactly what is meant when someone begins talking about a primary intension. "Watery stuff" is clear enough, I suppose, but when I think about what I mean by this, I end up thinking of a bunch of descriptive conditions. There are two ways to run this, I guess. First, the confusion applies to PI's in particular, not SI's. This seems strange, since it would seem very arbitrary that "surface" properties but not the internal structure,of something would be subject to description-theory objections (i.e., Kripke's). Second, fixing reference is equally problematic for PI's and SI's. This doesn't seem right either, because we want to preserve the notion that concepts like "watery stuff" apply in all possible worlds (to water) while "Water is H20" is different in that it could have been something else (XYZ) say, before the reference was fixed. So I suppose the modal distinction is preserved or insulated from epistemic worries about fixing reference, but I'm still troubled by the use and applications of things like PI's in the discussion so far, and I'm not sure what the point of the modal issues are if we can't make sense of PI's and SI's epistemically. Is this all clear (or has it been clarified already?). If so, disregard... Erik L. From owner-modality@LISTSERV.ARIZONA.EDU Mon Feb 15 01:08:49 1999 Date: Mon, 15 Feb 1999 01:08:25 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: Epistemic and modal issues To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Erik asks some important questions: >Hello, I know we're not directly discussing PI's in relation to reference >anymore, but lately I think I've managed to unravel what little >understanding I (thought I had) on this issue. So, apologies for >re-opening some discussion, but here goes. Alot of what I took Thony's >concern with PI's to be centered on was the convergence of PI's to >descriptive content when using them to fix reference. I take it that >Kripke (obviously) wants to avoid this possibility, requiring only of PI's >that they are part of a causal chain that begins in the referent and links >up to the subject (Chalmers too--see p.59 of TCM). My problem is that, >pulling PI's away from descriptive accounts and towards causal ones (or at >least calling them "compatible" with such accounts) does not make the >application of PI's in reference fixing any less mysterious. I'm not sure >that we can have cognitive access to things like causal chains, and if so, >I'm not sure how any of this would work. Actually, this posting is right on topic for this week's discussion of PIs and the 2-D framework. Are PIs like descriptions? On my view, in a sense yes, and in a sense no. PIs needn't be linguistic formulae, or correspond to such formulae. A PI is a function from centered worlds to extensions. Sometimes such a function will be completely captured by a description, such that the PI picks out whatever satisfies the description in world: e.g. "the length of such-and-such stick", for the case of 'meter', or "the individual at the center of the world", in the case of 'I'. But this sort of neat expressibility may be quite rare, and certainly isn't universal. The conditions of application in a PI may be too unruly to be captured by any simple formula, or perhaps by any finite formula at all. So we shouldn't expect to always be able to express PIs as simple descriptions. But still, there's a looser sense in which PIs behave something like a description. Like a description, a PI captures the conditions that an object in the actual world has to satisfy in order to be the referent of one's term or concept. And like a description, it's plausible that a PI is cognitively accessible in principle, at least in the sense that if a given centered world is specified, a thinker can in principle figure out the value of the PI in that world (by figuring out what the extension of our concept or term would be if that world were actual). Furthermore, one can at least attempt to *approximate* the conditions involved in a PI by a description. Any short description may be imperfect -- there may be cases where the description and the PI give the wrong results, as e.g. in the Kripke cases. But arguably one can give better and better approximations to the PI by giving longer and more involved descriptions, to handle exceptions and the like. It's not clear that any finite description will be perfect: maybe it will be like the case of "knowledge", where one seems to get longer and longer descriptions (cf. Chisholm's thirteen clauses, and Shope's _The Analysis of Knowing_), and one gets the sense that one needs some sort of infinite "fractal" structure to get all the details just right. But in any case one can arguably give better and better descriptions that cover more and more of the cases. How does all this play out, if something like the causal theory of reference is correct? Well, I wouldn't say that the PI in this case is a causal chain: you're right that we don't have access to causal chains. But note that even in the Kripke cases, we still have *conditional access* to extension. Kripke's very own methodology is to consider various scenarios as epistemic possibilities, and to figure out what the extension of our terms (e.g. "Godel", "Feynman", "gold") would be if those scenarios turn out to be actual. So we still seem to have cognitive access from a function to from scenarios to extensions. And that's a primary intension. How can one cash out the primary intension in descriptive form here? Well, as usual, it's not obvious that one can. But one can at least provide approximations. E.g., one might approximate the PI of 'Godel' as "the individual at the other end of a causal chain from my use of the term 'Godel'", or "the person that experts in my community call 'Godel'", or "the person I've heard of under the name 'Godel'", or some such. Some of these are vague (e.g. the first -- what sort of causal chain is required?) and some may be imperfect (e.g. the second -- maybe I misheard and evryone else calls him "Girdle", not "Godel"?), but at least they give a loose approximation. And arguably they can be refined into better and better approximation. Is there a perfect description? Well, that's still a topic of some debate. Some think that something like the third description above can do the job (metalinguistic descriptivism). And some think that if the causal theory of reference is right, one ought to be able to bundle it all into a description, something like "whatever individual is causally connected to me in the manner specified by the causal theory of reference" (causal descriptivism). Even Kripke thinks that something like the latter might be done (see p. 162 and footnote 38), though he thinks it is trivial. Whether a less trivial description can do the job -- maybe a "cluster" description with causal and metalinguistic elements -- is still arguable. Anyway, the PI framework doesn't need to take a stand on whether the PI can be captured by a description. I think by now you can see in what sense it is "description-like", in that it in effect gives the conditions a referent must satisfy, and is cognitively accessible. And in any given case, one can at least try approximating the PI by a description. I often give descriptions to express something about a PI myself, but one shouldn't forget that the description is usually just an approximation. But in any case, given that we have conditional access to extension in a given case (as Kripke's own methodology suggests in even his anti-description cases, and as I think is plausible in almost any case), it follows that we have, in effect, cognitive access to a PI. >To put it simply, I don't see how something like "watery stuff" can fail >to be descriptive (and therefore adopt all the problems of descriptive >theories or reference), or (assuming it is not) how calling in something >like a "causal chain" makes the application of a PI any clearer. If this >is any objection at all then I suppose it would be sufficiently general to >make reference fixing for SI's puzzling too. My general problem is that I >can't seem to make sense of what we mean when we say "watery stuff" if it >is not shorthand for some description of the properties of that stuff. >If we want to call it a causal connection (involving some initial dubbing >ceremony), then I guess I don't know how we have access to those sorts of >connections such that we can accomplish reference fixing. Does the above help? Again, "watery stuff" is really a term of art, and isn't really meant as a perfect descriptive characterization of the PI. Using "watery" in its intuitive sense, then the description might give a very loose approximation to the PI, but there are some cases where it might come apart. In the book i think I stipulate instead that "watery stuff" is a term of art to capture the conditions involved in the PI of 'water', whatever they are. To give a full accounting of those conditions and of the PI, one can really only consider things case by case: look at scenarios, and see how the extension of 'water' should come out there. If I were to really try giving a description for the PI of "water", I suppose it would probably involve its superficial properties (though defeasibly, to allow for the epistemic possiblility of optical illusion?), and it might also involve a causal connection criterion, perhaps even a condition of causal connection to my use of the token 'water'. And any such description would likely be imperfect. But at least it gives an intuitive idea. In any case, the complexities of exactly capturing the PIs of our linguistic terms don't really matter much for our purposes in metaphysics. Beyond a certain point, the complexities of "what counts as 'water' in this situation, or in this one" become somewhat terminological. For explanation, for example, what matters is explaining the manifest properties of people such as Godel and stuff such as water, and here the terminological complexities of just who and what gets to count as "Godel" or "water" becomes irrelevant after a certain point. (Even if the PI of 'water' isn't perfectly captured by "watery stuff", it might as well be for many purposes.) It remains the case that getting things just right is very important in the philosophy of language, but it's not clear that all the fine details matter to the metaphysician or to the scientist. That's why I tended to brush aside some of the fine details in the book. >The point of this is just to express my confusion with exactly what is >meant when someone begins talking about a primary intension. "Watery >stuff" is clear enough, I suppose, but when I think about what I mean by >this, I end up thinking of a bunch of descriptive conditions. There are >two ways to run this, I guess. First, the confusion applies to PI's in >particular, not SI's. This seems strange, since it would seem very >arbitrary that "surface" properties but not the internal structure,of >something would be subject to description-theory objections (i.e., >Kripke's). Re SIs, I suppose things are a bit different since it's not part of the package that they are cognitively accessible (in a nontrivial' sense) to anyone who possesses the concept. You need substantial empirical knowledge of the actual world in order to evaluate the SI of one's concept in an arbitrary possible world (e.g., you need to know about the actual structure of water to know that the SI of 'water' picks out H2O across worlds). But I suppose there's a sense in which at least after the fact, SIs are often easier to summarize with descriptions (e.g. "H2O") than PIs are. I'm not sure that this is always so (personally, I think one might even worry about whether any H2O in any world counts as water), and its not clear that there aren't some more resources for expressing PIs (I'll talk about alternative ways of looking at them a bit later), but even if so, I think it's really an issue of relative simplicity, rather than expressing some deep metaphysical difference. (Of course, there is the epistemological difference in cognitive access.) >Second, fixing reference is equally problematic for PI's and >SI's. This doesn't seem right either, because we want to preserve the >notion that concepts like "watery stuff" apply in all possible worlds (to >water) while "Water is H20" is different in that it could have been >something else (XYZ) say, before the reference was fixed. So I suppose the >modal distinction is preserved or insulated from epistemic worries about >fixing reference, but I'm still troubled by the use and applications of >things like PI's in the discussion so far, and I'm not sure what the point >of the modal issues are if we can't make sense of PI's and SI's >epistemically. Is this all clear (or has it been clarified already?). If >so, disregard... Well, we want "water is watery stuff" to be true across all worlds considered as actual (i.e., PI true in all worlds), and "water is H2O" to be true across all worlds considered as counterfactual (i.e., SI true in all worlds). Of course one difference is that the necessity of the SI depends on empirical facts about the actual world while the necessity of the PI doesn't. But that just reflects the fact that PIs are much more closely tied to epistemic notions such a apriority than SIs are. Re epistemic access to PIs and SIs: as I said, I think one has access to a concept's PI merely by possessing the concept (you then have the ability to figure out how it refers if a given world is actual). For an SI, I think one only has conditional access to it: one doesn't have nontrivial access to it merely by possessing the concept, but one does have access given sufficient empirical knowledge. So on my view, one has a priori access to a PI, and a posteriori access to an SI, but one does have some sort of a priori access to the way that an SI depends on the way the actual world turns out. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Mon Feb 15 22:02:48 1999 Date: Mon, 15 Feb 1999 23:01:21 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Angela J Burnette Subject: Re: Epistemic and modal issues To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO all re Dave's answer to Erik's comments about PI's, the questions that follow have really already been answered but I guess I'm just not getting it through my thick scull... Dave says that primary inentions are functions from centered worlds to extensions, which is straightforward, but given that it may be impossible to specify or spell out the conditons for application, I (still, sorry) don't see how such a function can do us any good...(a function that has no specifiable conditions of application hardly seems like a function at all) I understand that the PI is supposed to "capture the conditions that an object in the real world must satisfy in order to be the referent of one's term or concept", but given that this must be done on a case by case basis and cannot be cashed out in terms of any description (even if one is available, such a descritption is not of primary importance, right?)...I still don't understand the "sense" in which anyone who has the concept can "in principle" evaluate PI's except as a matter of personal intuition...it seems to me like everyone will have their own "function" for each PI, which will also be just what they happen to have in mind...I think that Dave intends this not to be a problem....something about not mattering "after a certain point"...this is what I really need explained...why doesn't it matter after a certain point? I understand what a PI is intended to accomplish and I understand how, in principle, or in some vague sense, it is actually used, but what I'm unclear on is why it is "unimportant" that PI's be such vague and unspecified tools... angela From owner-modality@LISTSERV.ARIZONA.EDU Tue Feb 16 06:26:56 1999 Date: Tue, 16 Feb 1999 06:25:01 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: Epistemic and modal issues To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Angela asks some good questions, to which I'm not sure I have perfect answers, but I'll do my best. >Dave says that primary inentions are functions from centered worlds to >extensions, which is straightforward, but given that it may be impossible >to specify or spell out the conditons for application, I (still, sorry) >don't see how such a function can do us any good...(a function that has no >specifiable conditions of application hardly seems like a function at all) Hmm, I'd say a function is a function (i.e. a mapping from one domain to another, here a mapping from worlds to extensions), irrespective of whether there is any easy summary of a "rule" that generates the function. You're right that's its nice to have some sort of summary of a function when one can, but it's not clear to me why it should be essential. In the case of PIs, we've seen that one can get reasonably good approximations of PIs by descriptions, and better and better approximations by more and more complex descriptions. I'd like to think that that's good enough (and in fact, that even the approximability of a PI by descriptions is something of a bonus rather than being essential. But maybe I'm wrong. What do you think is the reason or purpose for which exact descriptive summaries of PIs are essential? >I understand that the PI is supposed to "capture the conditions that an >object in the real world must satisfy in order to be the referent of one's >term or concept", but given that this must be done on a case by case basis >and cannot be cashed out in terms of any description (even if one is >available, such a descritption is not of primary importance, right?)...I >still don't understand the "sense" in which anyone who has the concept can >"in principle" evaluate PI's except as a matter of personal intuition...it >seems to me like everyone will have their own "function" for each PI, >which will also be just what they happen to have in mind... Right (re your first question). Sometimes there will be a description, but even here we figure out the aptness of a description precisely by seeing whether the evaluation of a description matches our evaluation of the PI at a given world. (Witness Kripke's methodology: he argues against various descriptions by arguing that there are worlds where the description gives the wrong referent. That's to say, the evaluation of the description doesn't match our evaluation of the PI. So there is arguably some sense in which even when a description is available, the PI is prior to the description, at least epistemically. (Though it may be that at some other level, e.g. psychologically or even semantically, it's the description that fixes the PI.) You're right that "intuition" plays a central role in evaluating PIs. Again witness Kripke's methodology in the theory of reference: he considers his own intuitions about what the referent of "Godel", "Feynman", etc, will be across various situations. One might say this makes it just "personal intuition", but note that we do seem to find considerable agreement across many or most such cases (cf. the proof-stealing scenario), so it's not as if mere personal taste is at play. It seems to be a fact about us that once we know about how the actual world turns out qualitatively, we're in a position to know the reference of our terms and the truth-value of our statements (we'll be focusing directly on this "scrutability" thesis later on). This doesn't appear to be a mere vagary of personal taste, but a reasonably systematic process of reasoning and judgment. I note that if it's just "personal taste" across possible cases, by similar reasoning it would be just personal taste that determines our judgments of refrence and truth-value in the *actual* world (given empirical information about it). On the face of it, that doesn't seem too plausible: although intuition is involved in these judgments (is in almost any judgment), there's plausibly a fact of the matter in both cases about the reference of our concepts and the truth-value of our thoughts, a fact of the matter that we are plausibly capable of figuring out given sufficient qualitative empirical information and a priori reasoning. But if we can do that for qualitative information about the actual world, presumably we can also do it for information about all sorts of other ways the actual world might hypothetically turn out. Just apply the same sort of reference- and truth-figuring processes (after all, if we were all told that the actual world was just that way, we'd be able to figure out the consequences for truth and reference). So I do think something more than just personal taste is present here. Of course it may be that two users will make slightly different judgments about the truth and reference of their terms across various possible worlds (considered as actual). One way this can happen is for there to be conceptual differences between the two -- i.e. our concepts really do have slightly different PIs as an aspect of their content, as e.g. in a deferential user of "arthritis" and an expert user, or two people with slightly different concepts of "London". But that isn't a problem. Leaving that sort of case aside, I suppose another thing might happen is that we'll make slightly different judgments in borderline or fuzzy cases. I think that sort of thing is relatuvely unimportant. When it comes to borderline and indeterminate cases, all bets are off. What matters is that we have clear judgments (in principle) about the clear cases. I also note that it's not clear how having descriptions would help with this problem. If two users have different judgments about reference in a given case, presumably that would require that they have different descriptions. So if you're right that there are a lot of differences in judgment, then there will be a lot of personal descriotions, too. So the issue of intuition and judgment is arguably somewhat independent of the issue of intensions vs. descriptions. [Side note: There is a question about whether a PI should be defined in terms of a user's best a priori *judgments* about the reference and truth-value of their concepts and thoughts in given epistemic possibilities, or should be defined as what the reference and truth-value of the concepts and thoughts *would be* in those epistemic possibilities, if they turned out to be actual. I hold that these two notions always or almost always give the same results, but for some purposes it is useful to distinguish them: call them an epistemic PI and a fixing PI. Re your point above, a user's judgments and intuitions play a role in defining an epistemic PI (though I note that one is idealizing by allowing arbitrary a priori reasoning here). For fixing PIs (which are arguably the most straightforward), however, judgments and intuitions play no role in the definition. All that matters there is patterms of reference and truth. So focus on those if you want to avoid any problems with intuition and judgment. We can talk more about epistemic PIs and fixing PIs if people like. The issue will come up a fair bit anyway, re the discussion of scrutability. Basically, epistemic PIs as defined are a priori accessible, and fixing PIs as defined fix reference, but one needs epistemic PI = fixing PI in order to have something a priori accessible that fixes reference. The requisite bridge is provided by the "scrutability" thesis that reference of our concepts and truth-value of our thoughts are always knowable in principle given sufficient qualitative information about the world and sufficient a priori reasoning. That's a thesis I've already implicitly appealed to a few times, and which is consistent with e.g. the standard methodology in the theory of reference. Some might think there are exceptions to this thesis, "inscrutable truths": e.g. if the epistemic theory of vagueness is true, it might provide some. If this were so, it would raise complexities for PIs, e.g. requiring us to distinguish epistemic and fixing PIs. I hold that there are no inscrutable truths, but that's something we'll be discussing at length later.] >...I think that >Dave intends this not to be a problem....something about not mattering >"after a certain point"...this is what I really need explained...why >doesn't it matter after a certain point? I don't remember what I said didn't matter. I think small disagreement about the application of our concepts across cases. That basically suggests that our concepts have slightly different PIs, or that one of us hasn't really reasoned things through, or that we're dealing with a borderline case. But it's not clear that any of those phenomena matter much. Different people do indeed have different conceptual contents (often even associated with the same word), and a little indeterminacy around the edges of a PI doesn't matter any more in principle that indeterminacy around the edges of actual-world application. A general theme is that whatever goes for the actual world goes for possible worlds, too (e.g. re indeterminacy, epistemic access, and the like). >I understand what a PI is intended to accomplish and I understand how, in >principle, or in some vague sense, it is actually used, but what I'm >unclear on is why it is "unimportant" that PI's be such vague and >unspecified tools... Well, it depends on what you mean by vague and unspecified. "Vagueness" in the sense of indeterminacy at the edges (i.e. worlds in which a PI is borderline true/false) can happen but doesn't seem worse than actual-world vagueness. "Vagueness" in the sense of indeterminacy even on core cases I don't think exists -- for any concept and thought, there is a very wide range of scenarios on which the concept is true or false. "Unspecified" in the sense of ill-defined I'd also argue with -- PIs have a perfectly good definition (whether epistemic or fixing), associating a concept with a clear value in worlds across a range of cases. Your central problem is presumably not any of the above, but rather "unspecified" in the sense of not being given by a neat description. I'm still not sure just why this is a problem. I guess the proof has to be in the pudding, but I think that application to a range of issues shows that the notion of a PI can do a lot of explanatory work (see e.g. the content paper), and that indeed much or most of the work that descrioptions do for us is in virtue of their determining an intension rather than in virtue of their linguistic form. There's more to say on this, but I have to go to bed. I may well be missing something, and I imagine you're not the only one who is at least somewhat bothered by the issue, so you or anyone else should fire away and say just what you think we lose by having an intension instead of a description. --Dave. P.S. I'm looking forward to everyone else's posts for this week, and to the minutes of Tuesday's meeting. Note that re the "Components of Content" paper, the issues about psychological explanation and belief ascription, etc, aren't central to what we're focusing on this week, though we may come back to them. It's the general framework that's most important, and its application to the understanding of necessity, apriority, reference, and content. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 00:27:27 1999 Date: Wed, 17 Feb 1999 01:26:31 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Erik A Herman Subject: 2D To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO First off, I am in agreement with the idea of notional content. I have always considered the thinkers relation to the concept to be an integral part. I've been wondering if/how time fits into the picture. It is mentioned in a few spots in Components of Content:, "facts about the current environment...", "Tom is hungry at time t..." Take the Phosphorus/Hesperous example, if I understand correctly, these have different primary intension but fix the same referent, Venus. Can I say that there is a physical difference, namely the space and time, between Phosphorous and Hesperous? Also, I don't understand what happens in the case where someone ONLY knows about H2O's chemical properties (theoretical chemist). Is this a case of a descriptive concept where the secondary intension is simply the primary intension? My intuitions about Josh's brain in a vat, as we discussed this afternoon, are that, similar to a paradigm shift in science, water would just mean watery stuff like always, but now it just has a different explanation that the "new man" would need to learn to assign it. Erik H. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 02:42:22 1999 Date: Wed, 17 Feb 1999 02:42:10 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Erik writes: >I've been wondering if/how time fits into the picture. It is mentioned in >a few spots in Components of Content:, "facts about the current >environment...", "Tom is hungry at time t..." Take the >Phosphorus/Hesperous example, if I understand correctly, these have >different primary intension but fix the same referent, Venus. Can I say >that there is a physical difference, namely the space and time, between >Phosphorous and Hesperous? Well, time is built into the center of the world. In effect, the center consists of an individual at a specific time (or an ordered pair of an individual and a time). That's needed for concepts such as "now", which pick out the current time (i.e. the time at the center), and for a bunch of related concepts. I'm not sure exactly what you're asking in the last sentence. Basically, the idea is that the PI of "Hesperus" picks out (something like) the bright object visible in the evening sky from thje point of view of the center, and the PI of "Phosphorus" picks out the bright object visible in the evening sky from the point of view of the center. In the actual world, the evening object is the morning object, so those two intensions pick out the same object, namely Venus. So the two concepts have different PIs but the same referent (cf. Frege on sense and reference). >Also, I don't understand what happens in the case where someone ONLY knows >about H2O's chemical properties (theoretical chemist). Is this a case of >a descriptive concept where the secondary intension is simply the primary >intension? I'd say that's close to right. At a first approximation, we can say the PI and the SI both pick out H2O in all worlds, so it's like a descriptive concept. On a second appriximation, though, one can note that "hydrogen" probably isn't a descriptive concept. Its PI picks out some gaslike stuff in all worlds, and its SI picks out some particular atomic structure in all worlds. Similar for "oxygen". So it may be that the SI of "H2O" requires just that atomic structure (with just the right number of protons, neutrons, electrons, etc), while the PI of "H2O" isn't quite so constrained. (The PI of H2O is tricky, though. It comes down to tricky questions about e.g. what we'd say if hydrogen and oxygen exist but are quite unlike what we thought they were in atomic structure; but nevertheless there is still a substance with two hydrogen particles and one atomic particle, or some such. Would we count that as H2O, or would the change in chemical theory make us throw away the whole concept of "H2O" and say it doesn't refer? Maybe the former, but it's a subtle issue. Maybe in the grey area. In any case, though, it does seem that the PI will allow for at least some smallish variation in structure that the SI arguably won't.) Interesting to see people's comments re Josh's case. It would be good if people can be as explicit as possible about how their views on the referent of "water", etc, in these cases fit into the 2-D framework. (E.g. given your views on the reference of the concept and the truth of the thoughts before and after, what is the PI, and what does it pick out in the various centered worlds before and after vat-release?) A little more on Angela's note from yesterday. Maybe one can think of a PI as something like a tacit descriptive content of a concept. It may not be something that a speaker can directly articulate, and there may not even be any summary of the relevant content in words, but one could argue that in some sense the descriptive content is there at least tacitly, in guiding one's judgments about reference in various epistemically possible cases. Of course we may have to break the link between "descriptive content" and precise linguistic descriptions, though. It will be more like something analogous to descriptive content with the linguistic element removed. (Something like a property that determines reference, or some such.) That's a bit loose, but maybe it helps with the intuitive picture. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Tue Feb 16 21:59:38 1999 Date: Tue, 16 Feb 1999 21:59:21 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Josh raises a couple of interesting brain-in-vat scenarios. There are actually two sorts of question in the vicinity: (a) what to say about the beliefs in question, whether they are true or false before and after escaping from the vat, and (b) whether what one says re (a) can be accommodated within the 2-D framework, and if so, how (what do we have to say about the PIs of the relevant concepts, etc). Personally I'm not certain of the right thing to say about (a) (there are multiple intuitions), but I'd like to think that any reasonable answer to (a) can be accommodated within the 2-D framework somehow. It's a nice test case, though. Maybe I'll hold off on posting my own response to these scenarios in order to give other people a chance to post their analyses and comments on what's going on here. Fire away! --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Tue Feb 16 13:53:15 1999 Date: Tue, 16 Feb 1999 14:50:32 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Josh Cowley Subject: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Sorry to be out of the loop the last week. Its possible that in catching back up I've missed something. If something like the following has already been discussed I apologize. The following is a case that I don't know what to do with. Suppose scientists pull your brain out of the vat it has been in up until now and tell you that your whole life has been a lie. They created the world you thought you were living in all this time. Suppose you are conviced they are right. But now you are out of the vat and in a real body and can go explore the world. As you leave the laboratory and walk out into the world you see that everything is just like it was in your vat. There is grass, trees, water etc. But then one of the scientists runs up and says, "Ahh. One point we forgot to tell you. The underlying physics of the world is different from the one we presented to you in the vat. From the molecular level on down everything in your vat world was different. For example, water is XYZ." Case 2a: Same as above except that while I'm still living in my vat world, a dramatic scientific revolution occurs in which physics, from the molecular level on down is determined to be different than we though. For example, water is XYZ. I'm shown how thinking that water is H2O is wrong and actually it is XYZ. After I'm convinced of the new theory I'm pulled out of the vat and everyting runs as above except that now the real physics matches my vat physics. Now here is the puzzle. Primary intentions pick out "watery stuff" in the actual world. My intuition is that the actual world in both cases is the post-vat world. However, if I were the one pulled out of the vat in case 1 I'd say that there is no 'water' in the actual world. There is just watery stuff. But if I were pulled out of the vat in case 2, I'd say there is water in the actual world. If you share these intuitions, then there seem to be two possibilities: 1) My primary intention of water includes the chemical make up of water or 2) Primary intentions do not fix reference in the actual world but fix reference in some other way. What do you all think? Josh From owner-modality@LISTSERV.ARIZONA.EDU Tue Feb 16 23:56:08 1999 Date: Wed, 17 Feb 1999 00:55:16 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Erik J Larson Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO This is a fascinating case, but I'm not so troubled by it primarily because I don't share the feeling that water in post-vat world would be relegated to "watery stuff" if the underlying physics of the world turned out different. In a sense, the scenario is not appreciably different than one in which my "brain in the vat" life was just general ignorance of the way the world really was. For the purposes of fixing reference with a PI, you have to buy that the shift in physics from vat-world to post-vat-world will shift your PI of water, but it's not at all obvious that this would happen. A more plausible scenario (for me, anyway) would have one replacing the physics of water while retaining the PI to fix reference for whatever was watery stuff, in the vat or actual world. Here I think epistemic issues pull away from the modal concerns. Now, given that the PI of water does not change from vat to actual world, the physics of water is not part of its PI, so the first possibility can perhaps be dispensed with. The second possiblity is sort of conditional on the first, in the sense that, if one doesn't think that the PI will shift from vat to actual worlds, then the reference fixing work of the PI is intact. Of course, things get substantially trickier if one shares Josh's intuition on the vat to actual world shift in PI (from water to non-water) from the physics-of-water epistemic shift in these worlds. ErikL. to be out of the loop the last week. Its possible that in > catching back up I've missed something. If something like the > following has already been discussed I apologize. > > The following is a case that I don't know what to do with. Suppose > scientists pull your brain out of the vat it has been in up until now > and tell you that your whole life has been a lie. They created the > world you thought you were living in all this time. Suppose you are > conviced they are right. But now you are out of the vat and in a real > body and can go explore the world. As you leave the laboratory and > walk out into the world you see that everything is just like it was in > your vat. There is grass, trees, water etc. But then one of the > scientists runs up and says, "Ahh. One point we forgot to tell you. > The underlying physics of the world is different from the one we > presented to you in the vat. From the molecular level on down > everything in your vat world was different. For example, water is > XYZ." > > Case 2a: Same as above except that while I'm still living in my vat > world, a dramatic scientific revolution occurs in which physics, from > the molecular level on down is determined to be different than we > though. For example, water is XYZ. I'm shown how thinking that water > is H2O is wrong and actually it is XYZ. After I'm convinced of the > new theory I'm pulled out of the vat and everyting runs as above > except that now the real physics matches my vat physics. > > Now here is the puzzle. Primary intentions pick out "watery stuff" in > the actual world. My intuition is that the actual world in both cases > is the post-vat world. However, if I were the one pulled out of the > vat in case 1 I'd say that there is no 'water' in the actual world. > There is just watery stuff. But if I were pulled out of the vat in > case 2, I'd say there is water in the actual world. > > If you share these intuitions, then there seem to be two possibilities: > 1) My primary intention of water includes the chemical make up of > water or 2) Primary intentions do not fix reference in the actual > world but fix reference in some other way. > > What do you all think? > > Josh > "What our grammarian does is simple enough. He frames his formal reconstruction of K along the grammatically simplest lines he can, compatibly with inclusion of H, plausibility of the predicted inclusion of I, plausibility of the hypothesis of inclusion of J, and plausibility, further, of the exclusion of all sequences which ever actually do bring bizarreness reactions." -- W.V.O. Quine ---------------------- Erik J Larson erikl@U.Arizona.EDU From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 00:17:25 1999 Date: Wed, 17 Feb 1999 01:16:34 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Angela J Burnette Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO re: Josh's scenario... I guess I just don't share the intuition that I, upon exiting the vat and discovering that water was XYZ,would from that point on be of the opinion that there really was no water. Rather, I would just think that I had been deceived, or had been radically mmistaken about what water is.... I understand the temptation, especially given the amount of scientific information that is part of common knowledge, to say that H2O really is part of the primary intension of water, but, the very fact that I (and presumably others) can make sense of counterfactuals like, "Water could have been other than H2O," or "Water might not have been drinkable by humans", or "Water might not have had the property of appearing blue in daylight," etc...strongly suggests that almost nothing of what we know a posteriori of water is contained within the primary intension. best, Angela From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 00:34:32 1999 Date: Wed, 17 Feb 1999 01:32:28 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Erik A Herman Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO In response to Josh's scenario: my intuition is that the two cases are identical-- and that the new watery stuff is "water", it's just that he/we were wrong about it's makeup. Erik H. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 12:58:07 1999 Date: Wed, 17 Feb 1999 13:52:38 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Josh Cowley Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Regarding the responses to my BIV cases, I'd like to make a few clarifications. There are two ways that I see of accounting for my intuitions. The first is to claim that the chemical make up of water is part of my PI. The second is to claim that my PI is just like everyone elses BUT PIs don't map centered worlds considered as actual onto referents. I think the second of these claims is the more likely. The idea would be that PI's perhaps map centered worlds thought to be actual during concept formation onto referents. (I already see several problems with this definition, but that is the basic idea.) Since vat-Josh formed his concepts while thinking that vat-world was actual, he would then map water to clear drinkable liquid in vat-world. And in Vat-world water is H2O. Josh From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 18:10:24 1999 Date: Wed, 17 Feb 1999 18:07:56 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Here are a few thoughts on Josh's "water" case. In this message I'll give my own analysis, and in the next I'll give some thoughts on others' analyses. The first thing to get straight on is whether one thinks the judgments in question are true or false, before and after the vat. Case 1: I think "water is H2O" in the vat. I escape from vat, find a similar-seeming world, except that the watery stuff is XYZ. Case 2: I think "water is H2O" in vat, until a within-vat revolution convinces me of "water is XYZ". I escape from vat, find a similar-seeming world, and the watery stuff there is XYZ. Q1: What (if anything) do I refer to as "water" pre-escape? Q2: What (if anything) do I refer to as "water" post-escape? Q3: Is my statement "water exists" true or false pre-escape? Q4: Is my statement "water exists" true or false post-escape? There are various things to say here. It's probably simplest to first set aside the belief changes re XYZ and H2O, and think about what "water" might refer to pre- and post-vat in an ordinary BIV case. Here some distinct issues pecular to BIV cases some up. For example, many would say that while in the vat, my term "water" doesn't really refer to anything, and that my thought "water exists" is false. After all, isn't a brain in a vat a classic case of someone who is deceived about the external world? One could accommodate this line in the 2-D framework by saying e.g.: (1) the PI of "water" picks out something like "the watery stuff in the environment that the being at the center has been causally related to" (not an implausible analysis of my "water" PI), and (2) if I am a BIV, my centered world (i.e. my actual world, not merely the world-as-I-believe-it-to-be) doesn't have anything that satisfies that PI. (There's no watery stuff causally related to me in my actual environment.) So the PI will pick out nothing, and my vatted thought "water exists" will be false. On this line, what will happen when I come out of the vat? Well, arguably when I emerge for the first time into the "real world", my term "water" doesn't yet refer to actual water, because there's no causal connection yet. But after interacting with the environment for a while, the causal connection will get going, and perhaps eventually my term "water" will come to refer to the actual stuff (H2O). Note that all that will be compatible with the PI analysis above: nothing will satisfy the PI at stage 1 (centered on me in vat) at stage 2 (centered on me just after escape), but something may satisfy it at stage 3 (centered on me well after escape), as by then the center will have plenty of causal relations with some watery stuff. Another line that some philosophers take is that while in the vat, my terms like "water" refers to some chemicals or nutrients in the vat around the brain, because these are what I'm causally connected to. I don't find this very plausible myself, but if one takes this line, one could accommodate it by dropping "watery stuff" from the PI above, and by saying that the PI picks out roughly "the stuff in the environment that is causally responsible for the being at the center's use of the word 'water'". In this case, we'll have "water exists" true at stage 1 (it picks out chemicals), arguably false at stage 2 (t still picks out chemicals, but I suppose that could be true since the chemicals still exist), and true again at stage 3 (by now, the PI picks out real water). Alternatively, one could be tempted by the idea that while in the vat, one's term "water" refers to something in one's "virtual world", or something like that, so that one's vatted thoughts "water exists" is true. This has the problem of making it hard to express how the BIV is deceived, and could even end up leading to some sort of idealism or phenomenalism. But there is arguably something to the intuition that there's some sense in which the BIV's "water is H2O" is truer than "water is XYZ" (assuming the virtual world is an H2O-world). Maybe one could accommodate this by saying there's two ways to read such thoughts, a strong way in which they all come out false, and a "weak" way where some come out true, or some such. E.g., the "weak" way might involve prefixing "According to the virtual world", or might have a PI involving "seeming-objects which seem to be causally connected to me", or some such. This would be a tricky row to hoe, but maybe there are possibilities. In this case, one could say that reference of "water" gradually shifts after release, just as above. Anyway, I think one can factor away from these tricky BIV issues for considering Josh's main point. I'm not sure whether Josh's point would work equally well if the vat world were replaced by another planet in which one is brought up, and that brought to earth. That would probably give similar results at least to line 3 above, and arguably to line 2, though maybe not to line 1. But the crucial claims about how "water" refers after release may be similar each way. What to say about case 1, where I believed "water is H2O", and find out post-vat that the watery stuff is XYZ? I'd be tempted to say, as in the cases above, that immediately post-release my term "water" still refers to whatever it referred to just pre-release: either nothing, or some chemicals, or perhaps some virtual H2O in the virtual world. So my statement "water exists" then will be false, or in any case if true won't be made true by the actual watery stuff. But after a while of interacting with the real environment, my term "water" might well come to refer to the XYZ, and my claim "water exists" will be made true by the actual environment. I think that's so despite the fact that I believed in the vat that water was H2O. After all, within the vat that belief was potentially malleable (if people had told me the watery stuff was XYZ, I'd have accepted that water was XYZ), so the same ought to be the case outside the vat. In fact, I think that even if no-one outside the vat ever tells me that water is XYZ, my term "water" may eventually come to refer to the XYZ by virtue of my causally interacting with it, etc. In this case, I'll eventually have a false belief that "water is H2O" but a true belief that 'water exists". If that's how things so, I think it's compatible with the PI staying constant throughout. E.g. if the PI is "the watery stuff in the environment that the being at the center has been causally related to", this seems to give the right results: it picks out nothing at stage 1, nothing at stage 2, XYZ at stage 3. And that's so irrespective of my belief changes. I'd be tempted to say something similar about case 2. Here, I think there still needs to be a gradual reference-shift from vat-reference to actual reference. Even if one thought "water is XYZ" in the vat, and actual watery stuff is XYZ, it's not clear that one's thought (immediately post-escape) "water is XYZ" is true, or at least it's not clear that it's made true by the actual XYZ. It takes some interaction for the term "water" to pick out the actual stuff. [A side note: It may be that if the scientists tell you immediately "you were in a vat, but now you're out, and fortunately the real world is a lot like you thouight the vat world was", you might decide to make a new stipulation that all your words will refer to actual things rather than vat things, rather than keeping a lot of terms around that don't refer to anything useful. That's tricky in some cases: e.g. re "my wife", or "Ethel", say, does one really want to say that the term refers to a person in the actual world who you've never strictly speaking met before? But I suppose one could try to stipulate that at least some terms refer to their qualitative counterparts. If that's so, then "water" might immediately pick out the respective watery stuff. As for what to say if there are a few small differences between vat world and actual world (e.g. they replaced H2O by XYZ), that's awfully tricky. I guess it depends on one's stipulation. One might stipulate "in case of any differences, my term doesn't refer", or one might stipulate "in case of any differences, my term refers to whatever is superficially similar enough", or some such. In the first case, my post-escape "water exists" might at first be false; in the second; it might be true. But that all depends on just what stipulation one makes. Such a stipulation might well change one's term's PI a little bit, as any new stipulation will. Probably it's easiest to think about these things in a case where the scientists don't tell you you've escaped, so one isn't faced with the immediate wrenching discovery that forces you to reevaluate your whole conceptual scheme. Instead, maybe do things so that you don't know you've undergone such a radical shift. Even so, we can evaluate truth and reference, and we can still consider what happens if there turn out to be small changes in the environment. That way the "stipulation" aspect is removed.] Anyway, that's how I'd be tempted to analyze these cases. I think its plausible on most ways of doing things that one's PI stays constant, but its referent may gradually shift, in effect due to a change in the centered world in question (or at least, a change in the location of the center). The one case where my PI may change is the case where I know that I've shifted environments, and decide to restipulate the use of my terms; but that's a very special sort of change. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 17:24:53 1999 Date: Wed, 17 Feb 1999 18:24:00 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Timothy J Bayne Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Some thoughts on Josh's BIV case and on de se representation. . . The BIV case is a nice case, and a difficult one. I had an answer, but talked myself out of it. Here's another stab. This is in three parts, all rather convoluted. (1) Part of what is up for issue is whether the primary intension of our concepts like are egocentrically indexical, or whether they are more communal. Suppose that Josh is the only BIV, and that he thinks of water as . On this construal of the 1-intension, Josh's water concept refers to certain types of electrodes, or computer programs, or whatever it is that causes most of his watery experiences. (Why do we describe it as his "water concept"? Cos it's connected to his watery experiences, which are connected to our water concept.) This, I take it, is the Putnamian position on BIVs. On a *communal* interpretation of the 1-intension of water, it is (There are lots of indexicals in it). Now, whether this concept still refers to the electrodes or programs of whatever depends on whether Josh is the only BIV around. If the vast majority of the folks around here are also BIVs, then perhaps one could argue that even now it still refers to these electrodes. (Of course, the problem is that Josh isn't communicating with any of these - presumably. So maybe the communal condition can't get a toe-hold in giving the referent of an individual BIV's concept.) But let's suppose that the 1-intension of Josh's is communcal, and that he's the only BIV around here. Then, his thoughts about water would be mostly mistaken cos they are about water, and he isn't connected to water in the right way. Although of course his belief that water is H20 would be true if the stuff that causes most people's watery experiences is H20. Of course, it may not always be easy to tell whether one of his thoughts is true or false, because it may not be clear whether it is about water. Suppose that he thinks that he is having a shower: 'Ah, a lovely shower, this stuff sure is hot' and such like. This might be a true thought if the stuff that is causing his experiences is indeed hot. But insofar as he thinks that the stuff that is hot is the same kind of stuff that everyone else (i.e. non-BIVs) thinks is hot when they are taking a shower, his thought is false. Similar comments applying to his thought, 'It sure is good to run naked through the water of this Bondi surf.' This is true, insofar as it is good to run naked through the Bondi surf, but of course it is false in that Josh-the-BIV is not running anywhere, least of all running naked through the Bondi surf. (2) Let's push the Putnamian line for a minute and see where it goes. Take Josh-the-BIV's water concept to refer to whatever causes his watery experiences. Post-vat Josh's water concept also refers to the stuff that causes his watery experiences. These two concepts (mental representations) are similar in that they are both related to watery experiences, that is, they are both object concepts that are related to the *phenomenal concept* . But should we think that there is any sense in which VAT-Josh and post-Vat Josh have a concept, i.e. that can be tracked from the VAt to after the VAt? I wonder. Both in the Vat and after the vat Josh has a concept that is related to , but that doesn't seem sufficient to justify the claim that he has a single concept that survives the experience from VAt to actual world, that we can reidentify as the same concept. After all, both my concept of a certain chemical substance (say, glump - suppose that glup is watery) and my concept are connected to , but that doesn't give me any incentive to identify with . In order to identify the concept that we are calling Josh's pre-Vat water concept with his post-Vat water concept it needs to be the case that most (much?) of what he has learnt about water in the VAT holds true in the actual world. Does envatted Josh's knowledge of the microstructure of water play an important role here? I don't see why. On an inferential role account of concepts, beliefs about the microstructure of a substance might be just one sort of inferential node among many. If the microstructure of XYZ is such that it behaves in almost all of the same ways that H20 behaves, then we might be inclined to say that Josh has learnt that water is XYZ, or whatever it is in the actual world. (Of course, XYZ can't beahve in all of the same ways that H2O does, otherwise we would be able to discover that some stuff is H20 rather than XYZ.) To sum: supposing the individualistic spin on water, in which the primary intension is we (and Josh) might still be tempted to identify his in-Vat watery concept with his post-Vat watery concept, even thought it refers to a different stuff, because many of the inferential connections that he learnt while in the vat still hold true in the actual world. If post-Vat Josh thinks of his envatted experiences as including water thoughts, then it is clear that he must have a water concept that we can trace from the Vat to the real world. (3) Here's a similar case to Josh's that puts it in a communal setting. Suppose that we are all shifted to Twin Earth in our sleep. Twin Earth is as the travel brouchures describe it: it's the same as earth, except the predominent watery stuff in the environment is XYZ. AFter a couple of days we would discover that "this stuff in XYZ". I guess most of us would say, "Huh, I guess that water isn't H20 after all, it's XYZ." But it seems that we would all be mistaken. Water is H2O, but this ain't water. Of course, after a couple of years on twin earth, 'water' would come to refer to XYZ. So it seems. Well, I'm not sure what the point of this scenario is. Maybe it raises the question: 'Is the 1-intension of "the stuff around here that causes most of our watery experiences", or is the 1-intension of water "the stuff that is causing and has caused most of the watery experiences around here at the moment and in the past." The suggestion is that the 1-intension of natural kind terms anchors them to the past (and maybe the future). Tim University of Arizona Tucson, AZ 85721 USA Hm ph. (520) 298 1930 From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 20:31:28 1999 Date: Wed, 17 Feb 1999 20:29:40 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Some thoughts on analyses of Josh's case by Josh, Erik, Erik, Angela, and Tim. N.B. This presumes the previous message with my own analysis. Josh wrote: >Now here is the puzzle. Primary intentions pick out "watery stuff" in >the actual world. My intuition is that the actual world in both cases >is the post-vat world. However, if I were the one pulled out of the >vat in case 1 I'd say that there is no 'water' in the actual world. >There is just watery stuff. But if I were pulled out of the vat in >case 2, I'd say there is water in the actual world. > Like others, I'm not sure I share your intuitions here. Maybe there's some reason for saying that post-escape, your term "water" doesn't refer to anything in the actual world. But I think one can argue when it does begin referring to something in the actual world, it will refer just as soon in each case. After all, the only difference between the cases is that in one I believe "water is H2O" and in the other I believe "water is XYZ". This might make a difference to reference if these were "core beliefs" for me and unrevisable, or some such. But in fact it seems that these beliefs are quite revisable, a posteriori beliefs, as reflected by the fact that upon "discovering" in the vat that watery stuff is XYZ, one would say "water is XYZ" is true, not that water doesn't exist. I think the same goes outside the vat. Insofar as my word refers to stuff outside the vat at all, it's in virtue of its role as watery stuff, and it doesn't matter whether it turns out to be XYZ or H2O. >If you share these intuitions, then there seem to be two possibilities: >1) My primary intention of water includes the chemical make up of >water or 2) Primary intentions do not fix reference in the actual >world but fix reference in some other way. Hmm. I don't share the intuitions, but perhaps one can make sense of your intuitions by saying: the PI of water picks out "whatever has the same structure as the stuff I've been related to in my past environment", and going for a view (the third line in my previous post) on which in the vat environment, you're related to virtual-H2O and virtual-XYZ in case 1 and case 2 respectively. Then post-vat, out in the real XYZ-world, your "water exists in this environment" might come out true in case 2 but not in case 1 (because XYZ has the same structure as virtual-XYZ, but not as virtual-H2O). That would seem to give the results of your intuitions, though I don't know whether (a) it gets at what underlies your intuitions, or (b) it's plausible. [N.B. If things work this way, it's immediately post-escape that the changes will show up, rather than later. I.e. even before causal interaction with the new stuff, my "water exists around here" will be true in case 2 rather than case 1. With enough causal interaction, reference might still shift so that the claim will be true in both cases.] This analysis may get at something about what you're saying re "PIs don't fix reference in the actual world". I'd say that even on this analysis, they do fix reference in the actual world, but the actual world is a 4-D spacetime manifold, i.e. it includes all the facts about the past. And here, your PI will fix reference partly by appealing to facts about your past environment (i.e., the way things are in the past environment of the being at the center of the world) rather than your current environment. (It's plausible that something like that applies to many cases. E.g., if I am transported to Twin Earth without knowing, then presumably my immediate claim "there is water around here" will be false, even though there is watery stuff (in the loose sense) around there. Essentially, this is because the PI of water fixes reference to something like "the watery stuff I have causally interacted with", which involves reference to my past environment as well as my current environment. Of course if I live on Twin Earth long enough, my reference may shift enough so that "there is water around here" will be true; that's also predicted by this PI.) Looking at your (Josh's) more recent second message, maybe something like this analysis fits. >Regarding the responses to my BIV cases, I'd like to make a few >clarifications. There are two ways that I see of accounting for my >intuitions. The first is to claim that the chemical make up of water >is part of my PI. The second is to claim that my PI is just like >everyone elses BUT PIs don't map centered worlds considered as actual >onto referents. > >I think the second of these claims is the more likely. The idea would >be that PI's perhaps map centered worlds thought to be actual during >concept formation onto referents. (I already see several problems >with this definition, but that is the basic idea.) Since vat-Josh >formed his concepts while thinking that vat-world was actual, he would >then map water to clear drinkable liquid in vat-world. And in >Vat-world water is H2O. Well, one problem with what you say above is that if I grow up (in the actual world) thinking water is XYZ, then this formula will predict that my claim "water exists" is false, which doesn't seem right. And more generally it seems that it will be hard to have false beliefs while one is growing up, which seems problematic. Maybe it's better to say that something special is going on with the vat case -- these aren't just any old false beliefs about the world, but a whole virtual environment, one that is so self-consistent and central that it deserves to count in some sense as my "actual environment" while I'm in the vat. If so, then one can argue that my vat terms fix reference to things in my virtual environment, and that even after leaving the vat, they still involve the way things were in the vat environment. Note again that this doesn't involve ignoring the actual world, but it does involve concentrating on a particular aspect of it, i.e. my past environment, which in this case will be my past virtual environment (which of course is in some sense part of the actual world). On this way of doing things, the escape from vat to world is just like going from Earth to Twin Earth (in case 1) or like going from Twin Earth 1 to Twin Earth 2 (in case 2). Either way, one shifts environment, and one's initial reference is fixed to the stuff in one's original environment, so the truth of one's immediate-post-move claim "there is water around here" depends on whether there is anything around here with the same structure as the stuff in one's past environment. So true in case 2 but not in case 1. That seems to accommodate your intuitions, even with a constant PI that fixes reference via the actual world (including past environment). Anyway, note that one this way of doing things, what makes the differences between the cases isn't just that one had different beliefs in the past about one's environnment -- it's that in some sense, one actually had different environments in the two cases (albeit virtual environments, in Josh's version). To see the difference, think about what we'd say if (a) the virtual environment "really had" XYZ structure in it (i.e. if you'd looked under a virtual microscope, that's what you'd have found), but (b) you believe the whole time that there is H2O there (you never look under a microscope). I predict that in this case, Josh will say (or ought to say) that "water is H2O" is false when thought in the vat, and false after escape, while "water is XYZ" is true both times. That suggests that it's the past environment that matters, not the past beliefs. Of course one can question (a) whether a "virtual environment" is really enough of an environment to make concepts refer and to make beliefs true, and even then one could question (b) whether the "real" XYZ environment really qualifies as having the "same sort of stuff" that is present in the virtual environment (one might argue that there are deep differences between virtual XYZ and real XYZ, so that the "virtual" concept won't pick out the "real" stuff after escape). But assuming one can make a case for the appropriate answer on those two questions, then Josh's position will come out as reasonable, and the PI above seems to capture roughly what's going on. Re Erik L's, Angela's, and Erik H's messages: I think my intuitions largely agree with these, i.e. that there's no major difference between the two cases: insofar as case 2 "water" picks out the actual stuff post-vat, so does case 1 "water". (Although after going through the above, I have to say that I feel at least some of the tug of Josh's way of doing things, at least on the "virtual environment" reading.) I think one still needs to distinguish what happens immediately after release from what happens after a period of adjustment, though. I'm tempted to say that there is an element in the PI of "water" requiring a previous causal connection to the stuff, so it doesn't refer to the actual stuff immediately (in either case), but only after a while. If one doesn't agree with the causal requirement, one could go with a looser "the watery stuff around here now" PI, so that I'll come out referring in both cases immediately after escape. But I think the causal requirement is independently plausible. I do largely agree with what Angela says here: >I understand the temptation, especially given the amount of scientific >information that is part of common knowledge, to say that H2O really is >part of the primary intension of water, but, the very fact that I (and >presumably others) can make sense of counterfactuals like, "Water could >have been other than H2O," or "Water might not have been drinkable by >humans", or "Water might not have had the property of appearing blue in >daylight," etc...strongly suggests that almost nothing of what we know a >posteriori of water is contained within the primary intension. One thing to note is that "could have" and "might have" counterfactuals are often ambiguous between a PI and a SI reading, i.e. between (broad) epistemic possibility and Kripkean "metaphysical" or "subjunctive" possibility. I'm not 100% sure which (or both) you're meaning to appeal to here. Arguably one needs both to make best sense of these counterfactuals. If one accepts Kripke's subjunctive intuition, "water could have been other than H2O" will be true in the epistemic (PI) sense but not the subjunctive (SI) sense. As for "water might not have been blue", etc, this is most straightforward for me to make sense of on the subjunctive (SI) sense: one pictures the actual stuff, i.e. H2O, having a different appearance. On the epistemic (PI) sense, this is harder to make sense of, since water's appearance properties are relatively close to its PI, i.e. are "relatively a priori" of the "water" concept. But presumably one can still make sense of them by considering certain strange (broad) epistemic possibilities, e.g. on which there has been an optical illusion, or water's surface properties are very different from how one thinks they actually are. That reflects the fact that the beliefs in question aren't so "core" as to be unrevisable. As one builds in more and more core surface properties of water, though, this gets harder to do, and the epistemic "might have" claim becomes harder to make sense of. Re Tim's message: (1) Tim raises a possibility I didn't mention in the last message, that my vatted "water" concept may be deferential, fixing reference via my community. Still, I'd argue that it seems wrong to say that I defer to the people out there in the real world outside the vat, and it also seems wrong to say that I defer to other BIV's who happen to be out there. (It doesn't seem right to say that those people's reference makes a difference to my reference.) If I'm deferring, it's to people in my "virtual" world, the ones I think I "got" the term from. (I suppose it could be that those virtual people in my world are "avatars" of "real" people or of other BIVs, so I'd be deferring indirectly to those people, but I'll set that possibility aside.) If that's so, then I don't think deference makes an essential difference to the picture. If the non-deferential concept picks out nothing, or chemicals, or virtual XYZ, so will the deferential concept, on this way of looking at things. The main difference will be the precise nature of the "causal chain" connecting me to the referent, i.e. on whether it needs to go through other "virtual people", or not. (2) Tim also raises the question of in what sense I'll have the same concept inside and outside the vat. This seems most straightforward in the case where I don't know I've shifted. In that case, there seems to be the same sort of continuity that one would find if I go from one world to another without knowing. I'll deploy a lot of concepts, and they certainly won't be a whole new repertoire that I've acquired overnight -- rather, they'll be the old concepts (although a lot of my resulting beliefs may now come out false). It may be that eventually reference will shift, so that in a certain sense I'll have different concepts -- I'll at least have concepts with a different referent and SI, though not a different PI -- but that's a slightly different phenomenon. There's also the special case where I'm told I've shifted, and where I make a new stipulation about my concepts so my beliefs don't come out useless. That would plausibly involve a "different concept" in a reasonably strong sense, but that also seems to be something of a special case. Tim raises some worries about concept identity: >But should we think that there is any sense in which VAT-Josh >and post-Vat Josh have a concept, i.e. that can be tracked from >the VAt to after the VAt? I wonder. Both in the Vat and after the vat Josh >has a concept that is related to , but that doesn't seem >sufficient to justify the claim that he has a single concept that survives >the experience from VAt to actual world, that we can reidentify as the >same concept. After all, both my concept of a certain chemical substance >(say, glump - suppose that glup is watery) and my concept are >connected to , but that doesn't give me any incentive to >identify with . I presume you're talking about the vatted concept and the long-post-vat concept. In favor or "same concept" is that they have the same PI, and a continuous history of use between them. Against "same concept" is that they have a different referent and SI. What answer one gives will partly depend on whether one individuates concepts by PI, SI, referent, or history of use. I think this is somewhat terminological. I have some temptation toward "same concept", given the centrality of PIs, but then the different in referent and SI is pretty significant, too. As for water/glump, I'm not sure how to imagine the case -- do they pick out the same referent, or not? If the same referent, presumably we have same SI, different PI, so the reverse of the above. If a different referent, presumably we have different PI and SI, which is worse (presumably even if the PI of "glump" involves "watery", it will be in a different way, as it isn't a priori that water is glump). Either way, one relevant different is that the vat case involves sameness of PI and continuity of use, whereas this case doesn't. >To sum: supposing the individualistic spin on water, in which the primary >intension is we (and >Josh) might still be tempted to identify his in-Vat watery concept with >his post-Vat watery concept, even thought it refers to a different stuff, >because many of the inferential connections that he learnt while in the >vat still hold true in the actual world. If post-Vat Josh thinks of his >envatted experiences as including water thoughts, then it is clear that he >must have a water concept that we can trace from the Vat to the real >world. I think I agree with this, more or less. There are actually two issues: (1) sameness of concept pre- and post-vat, and (2) whether the concept picks out the actual stuff. That the "inferential connections still hold true in the actual world" seems to be getting at (2) rather than (1). In my terms, I'd put it by saying that as long as the actual world isn't *too* different from the vat world, actual watery stuff can come to satisfy the PI of my "water" concept (though possibly with a time lag). I'm inclined to think that (1) doesn't turn so much on the actual facts, but more on facts about the subject's psychological structure (modulo issues about individuating concepts by referent). That seems to fit with your last sentence above. (3) Tim raises the case where I'm shifted to Twin Earth in my sleep. As I said in the last message, I think Tim is right that my "water" won't initially refer to XYZ, though it may after a couple of years. I think the moral of that is that the PI of "water" anchors one to one's past environment, as Tim suggests. Basically because it picks out (at least in part) the stuff that one has causally interacted with in the past. In fact I think that for many of these cases, the present environment is mostly irrelevant (except to determining whether one's beliefs are now true or false, etc). The future environment even more so, though maybe one could come up with cases where it matters. All thoughts and reactions are welcome! --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Thu Feb 18 16:44:56 1999 Date: Thu, 18 Feb 1999 12:47:46 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Josh Cowley Subject: Minutes from Tuesday To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Here are the minutes from Tuesday. I thank Brad for filling in some details. The first thing we discussed was why we should care about the primar intentions. Roughly the line was, "if water *is* H2O then why do we care what water could have been. It was suggested that the reason is that if, via the PI we could determine that it was logical impossible for water to just be H2O, then we know that we have more to look for. This, of course, is the basic line for consciousness. The bulk of the rest of the dicussion was on proper names again. Thony argues that if the PI of a name is what picks out its referent, then two names which pick out the same object in our world could pick out different objects in another world. But it seems that as a matter of logic, if 'a=b', where 'a' and 'b' are names, then necessarily 'a=b' I suggested that in this framework proper names are not being used as labels for objects. Rather they are functions which pick out objects of a certain description. I think Thony and I now agree that the question is whether names ought to work this way. > Brad tried to provide motivation for thinking that names do in fact have > descriptive content. One such motivation comes from considering cases in > which co-referring names can't be substituted salva veritate (such as in > belief contexts). He also mentioned substitution failures in "simple > sentences" (see Saul 1997), such as "Lois Lane kissed Superman before she > kissed Clark Kent". > We also briefly discussed again the matter of what the primary intensions > of names typically are (such as in the London/Londres case discussed > before). We then dicussed my BIV cases. From an intuition point of view we were evenly split about whether or not there was really any water after you were pulled out of the vat. Brad suggested that it may just be very difficult to think of this case as anything other than a person jumping between two possible worlds. This, of course, is impossible. But it does look kind of like that. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 12:01:52 1999 Date: Wed, 17 Feb 1999 12:59:13 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Anthony T Lane Subject: possibility? To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO I think this is probably getting somewhat off the topic, but I have a general worry about making sense of secondary intensions. The 2-intension of water is H2O, given that, in this world, we take it that this chemical structure explains the various properties we associate with water. More generally, it seems that secondary intensions are different from the primary intension only when there is some sort of reductive explanation of a particular term or thing (p62). It is not, of course, directly observable that water is H2O. Rather, it is the result of theory constructed from a variety of experimental results. Consider a possible world in which water is H2O, but in which the natural laws are such that the elements behave radically different than they do in our world. Suppose, for example, that Hydrogen atoms form extremely unstable compounds when they are in the presence of other Hydrogen atoms. In this possible world, the dominant 'watery stuff' is XYZ. Thus, XYZ is the stuff picked out by the primary intension of water. XYZ behaves experimentally much like H2O behaves in our world. But XYZ is not composed of two Hydrogen atoms and an Oxygen atom in this possible world. It seems to me that the secondary intension of water cannot just be H2O. It must also include a host of other information regarding the underlying physical theory that makes this identification meaningful. If there is a possible world in which XYZ behaves for the most part like H20, only X, Y and Z are not 2 H's and an O, it seems strange that there is any value in the secondary intension of water picking out H2O, stuff that is, in this possible world, unstable. If this isn't just confused, it may be that this is a reason why you say primary intensions are particularly important to your account. Anthony From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 01:44:18 1999 Date: Fri, 19 Feb 1999 01:44:04 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: Minutes from Tuesday To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO A few comments on the minutes. >The first thing we discussed was why we should care about the primar >intentions. Roughly the line was, "if water *is* H2O then why do we >care what water could have been. It was suggested that the reason is >that if, via the PI we could determine that it was logical impossible >for water to just be H2O, then we know that we have more to look for. >This, of course, is the basic line for consciousness. Well, many of the reasons for caring about PIs are related to Frege's reasons for caring about senses. In particular, PIs seem to capture the epistemic relations between thoughts and the cognitive contents of thoughts, and are deeply related to notions such as epistemic possibility and apriority. This leads arguably to a central role for them in philosophy of language and mind (in giving an account of content that is useful for epistemic and cognitive purposes) and in epistemology. Within metaphysics, PIs are especially useful for analyzing modal issues. In particular, it seems that PIs are central in assessing conceivability claims, and the sense in which conceivability is a guide to possibility: if P is conceivable, it's generally plausible that there is a possible world satisfying the primary intension of C. And they are central to claims about a posteriori necessity. What's distinctive of a posteriori necessities seems to be that they have a necessary SI but a contingent PI. Given that for many issues in metaphysics we are concerned with questions about the relationship between conceivability and possibility, and about apriority and necessity, PIs seem pretty vital. >The bulk of the rest of the dicussion was on proper names again. >Thony argues that if the PI of a name is what picks out its referent, >then two names which pick out the same object in our world could pick >out different objects in another world. But it seems that as a matter >of logic, if 'a=b', where 'a' and 'b' are names, then necessarily 'a=b' > >I suggested that in this framework proper names are not being used as >labels for objects. Rather they are functions which pick out objects >of a certain description. > >I think Thony and I now agree that the question is whether names ought >to work this way. I'm a little concerned about this bit. Remember, one of the central points of the 2-D framework is that when S is necessary (in the usual sense), S has a necessary secondary intension. So what follows from the necessity of "a=b" is that "a=b" has a necessary SI, i.e. that 'a' and 'b' have the same SI. Nothing about PIs follows. In particular the claim that 'a' and 'b' have different PIs is completely compatible with the claim that "a=b" is necessary. (Just look at "Hesperus = Phosphorus", etc.) Once can think of the PI as being what picks out reference in the actual world (depending on how it turns out) and the SI as picking out reference in counterfactual worlds (given that the actual world is fixed). I.e., PI gives reference across worlds considered as epistemic possibilities, SI across worlds considered as subjunctive possibilities. Given this, then in a case such as the above (with different PI, same SI), 'a' and 'b' will pick out the same referent in all counterfactual worlds (all subjunctive possibilities), as Kripke et al lead us to expect. All that follows from the difference in PIs it is epistemically possible that reference comes apart, i.e. there are epistemic possibilities such that if they turned out to be actual, reference would come apart, which seems to be correct in these cases. Proper names are still being used as labels for objects in this framework, and they are still rigid designators. They pick out the same object across counterfactual worlds (considered as counterfactual). There's a sense in which actual-world reference is mediated by something a bit like a description (see previous note), but that's a far cry from saying that names here function like descriptions (which are nonrigid, etc, and have very different SIs). Witness the difference between "Hesperus" (the name) and "evening star" (the description). I had hoped that all this was clear by now, but the above suggests that maybe there is still a little confusion. Or maybe I am just misreading the above, in which case feel free to elaborate on the issue. >> Brad tried to provide motivation for thinking that names do in fact have >> descriptive content. One such motivation comes from considering cases in >> which co-referring names can't be substituted salva veritate (such as in >> belief contexts). He also mentioned substitution failures in "simple >> sentences" (see Saul 1997), such as "Lois Lane kissed Superman before she >> kissed Clark Kent". This is interesting indirect evidence, but I don't think the case for PIs relies on this sort of evidence. The mere epistemic possibility that Superman is not Clark Kent, and the way the reference of a term varies depending on which epistemic possibility turns out to be actual, is enough to make the case. Again, these things don't imply descriptive reference across counterfactual worlds, but they do arguably imply some quasi-descriptive content across epistemic possibilities. Incidentally it's probably a good idea not to get too fixated on issues specific to proper names, as many of those issues (especially those of reference via linguistic causal chains) aren't all that crucial to what comes later. Natural kind cases are probably more crucial. >We then dicussed my BIV cases. From an intuition point of view we >were evenly split about whether or not there was really any water >after you were pulled out of the vat. Brad suggested that it may just >be very difficult to think of this case as anything other than a >person jumping between two possible worlds. This, of course, is >impossible. But it does look kind of like that. Well, it may be more palatable to look at this as a person jumping between two very different parts of one world. A bit like going from one planet to another. In this case, going from a "virtual environment" to a real environment. Different environments in one sense, but still all part of the same world (it's a big world!). From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 02:05:16 1999 Date: Fri, 19 Feb 1999 02:05:05 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: possibility? To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Anthony writes: >I think this is probably getting somewhat off the topic, but I have a >general worry about making sense of secondary intensions. > >The 2-intension of water is H2O, given that, in this world, we take it >that this chemical structure explains the various properties we associate >with water. More generally, it seems that secondary intensions are >different from the primary intension only when there is some sort of >reductive explanation of a particular term or thing (p62). Well, I'd say this is often the case (especially for natural kind terms), but it needn't be the case. E.g. "one meter" has different PI and SI, but no reductive explanation in the vicinity. Similar for "Hesperus". It's true that for natural kind terms, though, we very often identify it with some explanatory microstructure across worlds. So the SI of "water" goes with H2O, of "tiger" goes with some bio-plus-DNA kind, of "heat" goes with molecular motion, etc. I don't think an actual explanation is required -- arguably the SI of "water" picked out H2O in all worlds even before we had an explanation. But it's arguable that the SI of most such terms picks out the explanatory microstructure of the stuff, whatever that happens to be in the actual world, across all worlds. >It is not, of course, directly observable that water is H2O. Rather, it is >the result of theory constructed from a variety of experimental results. >Consider a possible world in which water is H2O, but in which the natural >laws are such that the elements behave radically different than they do in >our world. Suppose, for example, that Hydrogen atoms form extremely >unstable compounds when they are in the presence of other Hydrogen atoms. >In this possible world, the dominant 'watery stuff' is XYZ. Thus, XYZ is >the stuff picked out by the primary intension of water. XYZ behaves >experimentally much like H2O behaves in our world. But XYZ is not composed >of two Hydrogen atoms and an Oxygen atom in this possible world. OK, it seems plausible that the PI of "water" picks out XYZ in this world. The Kripkean may well say that the SI of "water" picks out H2O, irrespective of its instability. >It seems to me that the secondary intension of water cannot just be H2O. >It must also include a host of other information regarding the underlying >physical theory that makes this identification meaningful. If there is a >possible world in which XYZ behaves for the most part like H20, only X, Y >and Z are not 2 H's and an O, it seems strange that there is any value in >the secondary intension of water picking out H2O, stuff that is, in this >possible world, unstable. Interesting point. The Kripkean often talks as if any H2O will be water. They will then describe this world as a world in which water molecules are unstable. It does arguably make sense to consider subjunctive possibilities such as "water molecules might have been unstable", or "if water molecules had been unstable, life on earth would not have evolved". If so, then such worlds (considered subjunctively) are worlds with water molecules in them, and thus (presumably) worlds with water in them. Still, I can see your intuition that thus might not really deserve to count as "water". One could similar consider worlds where H2O molecules make multicolored paints that people hang on walls, or are used in brick buildings, etc. Are these worlds where paintings are made of water, etc? (Even in the subjunctive SI sense?) I can at least see the counter intuition, though I go both ways. How to accommodate that intuition? Maybe (1) Insist that the SI of water still picks out H2O in all worlds, but say that that the molecules only count as H2O if they are quite a bit like our H2O molecules with current chemistry (i.e., being H2O requires more than just having 2 H's and an O).. Or (2) identify the SI of water with something like "H2O, as long as it has at least some of the other chemical or surface properties it has in the actual world". Or (3) go all the way to SI = "watery stuff", which would be a more radical move, letting in XYZ as water in some counterfactual worlds; unorthodox, but some (e.g. David Lewis) have been tempted to argue that there are worlds where water is XYZ. >If this isn't just confused, it may be that this is a reason why you say >primary intensions are particularly important to your account. Well, I do agree that there is an awful lot of slack in what counts as the SI of a given term. It depends some pretty debatble subjunctive intuitions about what it takes be water in a counterfactual worlds. There is also some slack in a PI, but arguably not so much -- we're somewhat clearer about what we'd count as water in different ways the actual world might turn out. This isn't the main reason I think PIs are more important than SIs (the central reason is that I think PIs tie much more closer to the epistemic notions that are at the heart of content, thought, and modality), but it may be one way of getting at the fact that SIs are most relevant in subjunctive and counterfactual contexts, about which our intuitive descriptions are pretty debatable. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 05:25:06 1999 Date: Fri, 19 Feb 1999 05:23:40 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Apriority, necessity, conceivability, possibility To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Recent discussion has been interesting, but we've drifted a little from the central topics for which the 2-D framework will be important for us. In particular, the framework is most useful for analyzing the relationship between apriority and necessity, the notion of the necessary a posteriori, the relationship between conceivability and possibility, and the issue of a priori access to modality. Of course all of these have are central to important issues in metaphysics, not least to analyzing epistemic and modal arguments in the philosophy of mind. So it's worth going over a few of these things. Most of what's contained in this note is very importan to the courset, so read it carefully. DEFINITION OF PRIMARY AND SECONDARY INTENSIONS I'll take a moment to officially define PIs and SIs for statements. So far we've mostly been talking about PIs and SIs for terms, but often it is whole statements that are most crucial. In effect, the PI of a statement is what you get from "composing" the PI and SI of its component terms. Note that I use "primary proposition" and "secondary proposition" for this idea in the book, but now I prefer to use "primary intension" and "secondary intension" for both. The primary intension of a statement S is a mapping from centered worlds to truth-values. The secondary intension of S is a mapping from uncentered worlds to truth-values. They're defined as follows: The primary intension of S maps a centered world W to the truth-value of S in W when W is considered as actual (i.e. considered as an epistemic possibility). The secondary intension of S maps a centered world W to the truth-value of S in W when W is considered as counterfactual (i.e. considered as a subjunctive possibility, with the actual world fixed). We can give a couple of heuristics for evaluating the PI or SI of a statement S at a world W. PI Heuristic: If W turns out to be actual, will S turn out to be true? SI Heuristic: Given that the actual world is the way it is, is S true in world W? Take W = XYZ-world, S = "water is XYZ". If the XYZ-world turns out to be actual, "water is XYZ" will turn out to be true, so the PI of S is true in W. But given that the actual world is the way it is (i.e. the H2O-world), "water is XYZ" is false in the XYZ-world, so the PI of S is false in W. (I abstract away from centering issues here.) (My favorite heuristics are actually a bit more streamlined than this, though maybe a bit less self-explanatory: PI Heuristic: If W is the case, is S the case? SI Heuristic: If W were the case, would S be the case? This ties PIs to indicative conditionals (which basically force you to evaluate the antecedent as an epistemic possibility) and ties SIs to subjunctive conditionals (which make you treat the antecedent as some sort of "fantasy"). But I'll say more on this later.) APRIORITY AND NECESSITY The 2-D framework can be understood as making the following claims concerning the necessity and apriority of statements S. NECESSITY THESIS: S is necessary iff S has a necessary secondary intension. Here, I'm taking "necessary" in the standard Kripkean sense (i.e. "subjunctive necessity"; I'll argue later that this usage is not wholly uncontestable, but set that aside for now). S is necessary in the Kripkean sense precisely if it is true in all worlds considered counterfactually, which is just to say that its secondary intension is true in all worlds. I hope this claim is fairly straightforward, given the definition of a secondary intension. APRIORITY THESIS (weak form): If S is a priori, S has a necessary primary intension. The case for this is pretty straightforward. If S is a priori, then plausibly it will be true no matter how the actual world turns out, i.e. it will be true in any world considered as actual (considered as an epistemic possibility). That is, for an arbitrary world W, if W turns out to be actual, S will turn out to be true. That is, the primary intension of S is is true in all centered worlds. APRIORITY THESIS (strong form): S is a priori iff S has a necessary primary intension. This makes the further claim that if S has a necessary PI, it is a priori. This isn't entirely straightforward to "prove". Necessary PI means that S will be true no matter how the actual world turns out, but it is still a leap from there to the epistemic claim that we can know S to be true. Maybe S could turn out to be unknowable, for example. E.g., maybe some really complex mathematical truth could have a necessary PI but not be knowable a priori? We'll come back to this issue of "scrutability" later. For now, I'd consider the weak form better established than the strong form, but I'd also say its plausible that the great majority of statements with a necessary PI seem to be knowable a priori, so that exceptions to the strong form are rare if they exist. I'm very interested to see any potential counterexamples to any of these principles (or arguments). It's not impossible that I've overlooked something here, so feel free to jig them up. A POSTERIORI NECESSITY THESIS: S is an a posteriori necessity iff S has a necessary secondary intension and a contingent primary intension. This follows directly from the necessity thesis and from the apriority thesis (strong form). I'd say that this is the central claim of the 2-D analysis of a posteriori necessity. It certainly seems to fit all the Kripke examples: "Hesperus is Phosphorus", "water is H2O", etc. Somebody might deny this thesis (presumably by denying the strong form of the apriority thesis), asserting that there is some other sort of a posteriori necessity. I think it's at least highly plausible that such a posteriori necessities would have to have their roots in considerations quite distinct from the ones Kripke raises, though. One reason all this is useful is that it lets us jump from epistemic claims about what is and isn't a priori to modal claims about the existence of various possible worlds. E.g. if S isn't a priori, there's a possible world where its PI is false; and even if S is an a posteriori necessity, its PI will be false in some world. If this is so, it's very important to the epistemology of modality, and to the role of epistemic arguments in metaphysics. What makes all this possible is a central property of the 2-D framework: that it has two dimensions of semantic evaluation (two intensions), but just one space of worlds. Centering aside, PIs and SIs are evaluated at exactly the same worlds. They just give different results because of the different ways the worlds are considered (as actual or as counterfactual). Thus the worlds considered as "epistemic possibilities" for PI purposes are perfectly respectable metaphysically possible worlds: witness the XYZ-world, for example. Just one world, but two different ways of looking at it. That means that when we have an a posteriori necessity, we have a genuine metaphysical possibility in which the PI of the statement is false. That metaphysical possibility may well have consequences for metaphysics. Again, any potential counterexamples and counterarguments are welcome. CONCEIVABILITY AND POSSIBILITY We can also put all this in terms of the relationship between conceivability and possibility. Conceivability is an epistemic notion. There are various notions in the area, as we'll see, but perhaps the simplest is the following. (N.B. I now prefer this way of doing things to the way I do things in the book.): S is conceivable iff S is not ruled out a priori (i.e., if it is not a priori that not-S). If the 2-D framework is right, then one will be able to make an inference from conceivability to possibility. If S is conceivable, not-S is not a priori, so not-S has a contingent PI (by the strong apriority thesis), so there is a metaphysically possible world in which the PI of S is true. That's not to say that S itself is "metaphysically" possible in the Kripkean sense (that sense requires a world where the SI is true), but we still have a metaphysically possible world in the vicinity. Call this the CONPOSS PRINCIPLE: If S is conceivable, there is a possible world in which the primary intension of S is true. We'll be looking at different versions of this claim later, and examining just what it takes to support the relevant versions. But note for now how it fits the Kripkean examples. People often say that the Kripke examples show that there can be no inference from conceivability to possibility. E.g., in the relevant sense of conceivability, "water is XYZ" is conceivable, as is "Hesperus is not Phosphorus", but neither of these is possible. But on the 2-D framework, there is at least a possibility in the vicinity: a possible world in which the PI of "water is XYZ" is true, and one in which the PI of "Hesperus is not Phosphorus" is true. Most importantly, when you conceive (in the relevant sense) of a scenario in which water is XYZ, or in which Hesperus isn't Phosphorus, you *do* conceive of a perfectly respectable world: the XYZ-world in the first case, and a world in which the evening object isn't the morning object in the second. These worlds don't satisfy the statements in question when they're considered as counterfactual, but they do when they're considered as actual. That's no surprise: after all, one is conceiving the scenarios in question as epistemic possibilities (i.e. considering them as actual), so one would expect that the worlds that correspond to this conceivability are worlds that satisfy S when they are considered as actual. So it's a little more indirect than a simple inference from S is conceivable to S is possible, but we still have conceivability as a guide to metaphysical possibility in some sense. MODAL RATIONALISM Note that the two-dimensional framework fits nicely with modal rationalism: the claim that we have a priori access to modality and to the space of possible worlds. Conceivability as described above is a paradigm a priori method. And when we conceive of a scenario in the relevant sense, it appears that we really are conceiving of a possible world (the XYZ-world, or the two-star world). Of course the sense in which it's an S-world is that it satisfies S's primary intension, not its secondary intension. So we have to be careful about just how we describe the worlds in question. But still, conceivability acts as a direct guide to the space of possible worlds. So the way is opened for some sort of modal rationalism. Of course all this depends on the central claims of the 2-D framework being correct. I think there are probably a few points at which an oppnent can argue (e.g. the strong apriority thesis), but its tricky, and again they'd have to go well beyond Kripke. PHILOSOPHY OF MIND All this is very relevant to epistemic arguments against materialism. One response to these arguments is to deny that there is an epistemic gap between physical and phenomenal, and assert that there is an a priori entailment. I sometimes call this "type-A materialism". It's a hard row to hoe, though, because it only seems to work if one accepts a functional analysis of phenomenal concepts, and such an analysis seems very implausible. So perhaps the most popular response is "type-B materialism": accept that there is an epistemic gap, but deny that there is an ontological gap. This doesn't have the intuitive implausibility of the type-A view, but still holds onto materialism. Typically, one will accept that there is no a priori entailment from physical to phenomenal, but will assert that there is a necessary entailment. So the entailment from physical to phenomenal is an a posteriori necessity. This view also usually goes along with the claim that conceivability doesn't imply possibility. So zombies (for example) are said to be conceivable but not possible, so materialism is not endangered. If the central claims of the 2-D framework are correct, this isn't so simple. The type-B materialist holds that "P -> Q" is an a posteriori necessity, where P is the physical truth about the world and Q is a phenomenal truth. On the 2-D framework, it follows that "P -> Q" has a contingent PI, so there is a possible world where the PI of P is true, but the PI of Q is false. Under a bit more analysis, the existence of this possible world is highly threatening to materialism. Similarly, if zombies are conceivable, it follows that there's a possible world in which at least the PI of "there are zombies" is true. From here (under a bit of analysis) one can plausibly get to a world physically identical to ours but different overall, so that materialism comes out false. The basic reason for all this, of course, is that deep down the Kripkean framework hasn't really broken the link between conceivability and possibility, it has just recast it a little. But we've seen that even Kripke's own results rely on the method of conceivability: conceive of a world, then make sure you describe it correctly. So it arguably supports rather than opposes a modal rationalism at the deep level. If so then the Kripkean framework gives much less support to the type-B materialist than is often supposed. We'll be going into the full details of how this applies to materialism etc next week, so we can save a detailed analysis of the consciousness issues for then. But its important to see how the 2-D framework might have significant metaphysical consequences, and how those rest pretty straightforwardly on the central theses of the framework. For now, I'd be interested to see some discussion of those theses, and of the relationship between apriority and necessity vis-a-vis PIs and SIs, etc. All comments are welcome. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Thu Feb 18 08:10:40 1999 Date: Thu, 18 Feb 1999 09:09:39 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Timothy J Bayne Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO > A quick comment on Dave's comment on my comment on Josh: [For Dave's comments, and some of my original comments, see below.] The point I was making about glumpy stuff was this. In Josh's scenario, I'm told (and believe) that I've just left a Vat. Presumably I will also believe that none of the stuff behind my past experiences is the same sort of stuff as the stuff behind my current expriences. In this situation, I want to suggest that we shouldn't reidentify my in-VAt <"water"> concept with my post-Vat water concept, *even if we're picking out concepts in terms of their primary intensions.* It is true that both in the vat and out of the vat, I have a concept that is closely connected with my phenomenal concept , and it's a natural kind concept too. Call these two concepts 'Water1' and 'water2' (These are just two names, I'm not yet supposing that these are two distinct concepts). When, out of the Vat, I use water2, I don't think of it as applying to the same sort of stuff that water1 applied to. And that, it seems to me, is enough to say that its primary intension is different. Here's an analogy. Suppose I have two concepts, and . I think that glumph and water are distinct natural kinds. Both act in the same (watery) way, indeed, I can't tell them apart. Almost nobody can tell them apart. (Perhaps nobody now living can tell them apart. It's just part of our mythology that there were people - the shabby pedagogues - who could tell them apart.) But I know that the stuff around here is glumph. Still, as far as I'm concerned, I interact with glumph pretty much the same way I would with water. Now, suppose that I move to an area where the watery stuff is not glumph but water. *If I know that this is the case*, I will use a different concept, my concept rather than my , to refer to the watery stuff around me. I want to say that this case is precisely analogous to Josh's BIV case in which he believes that he was a BIV. It may look like he is still using his water concept, but he's not. He's using a concept that has the same surface structure, but there's more to PI than surface structure. AT least for Natural Kinds, one has to think that one is talking about the same sort of stuff in order for "two" primary intensions to be the same. Well, maybe. > There's also the special case where I'm told I've shifted, and where I > make a new stipulation about my concepts so my beliefs don't come out > useless. That would plausibly involve a "different concept" in a > reasonably strong sense, but that also seems to be something of a > special case. > > Tim raises some worries about concept identity: > > >But should we think that there is any sense in which VAT-Josh > >and post-Vat Josh have a concept, i.e. that can be tracked from > >the VAt to after the VAt? I wonder. Both in the Vat and after the vat Josh > >has a concept that is related to , but that doesn't seem > >sufficient to justify the claim that he has a single concept that survives > >the experience from VAt to actual world, that we can reidentify as the > >same concept. After all, both my concept of a certain chemical substance > >(say, glump - suppose that glup is watery) and my concept are > >connected to , but that doesn't give me any incentive to > >identify with . > > I presume you're talking about the vatted concept and the > long-post-vat concept. In favor or "same concept" is that they have > the same PI, and a continuous history of use between them. Against > "same concept" is that they have a different referent and SI. What > answer one gives will partly depend on whether one individuates > concepts by PI, SI, referent, or history of use. I think this is > somewhat terminological. I have some temptation toward "same > concept", given the centrality of PIs, but then the different in > referent and SI is pretty significant, too. As for water/glump, I'm > not sure how to imagine the case -- do they pick out the same > referent, or not? If the same referent, presumably we have same SI, > different PI, so the reverse of the above. If a different referent, > presumably we have different PI and SI, which is worse (presumably > even if the PI of "glump" involves "watery", it will be in a different > way, as it isn't a priori that water is glump). Either way, one > relevant different is that the vat case involves sameness of PI and > continuity of use, whereas this case doesn't. > > >To sum: supposing the individualistic spin on water, in which the primary > >intension is we (and > >Josh) might still be tempted to identify his in-Vat watery concept with > >his post-Vat watery concept, even thought it refers to a different stuff, > >because many of the inferential connections that he learnt while in the > >vat still hold true in the actual world. If post-Vat Josh thinks of his > >envatted experiences as including water thoughts, then it is clear that he > >must have a water concept that we can trace from the Vat to the real > >world. > > I think I agree with this, more or less. There are actually two > issues: (1) sameness of concept pre- and post-vat, and (2) whether the > concept picks out the actual stuff. That the "inferential connections > still hold true in the actual world" seems to be getting at (2) rather > than (1). In my terms, I'd put it by saying that as long as the > actual world isn't *too* different from the vat world, actual watery > stuff can come to satisfy the PI of my "water" concept (though > possibly with a time lag). I'm inclined to think that (1) doesn't > turn so much on the actual facts, but more on facts about the > subject's psychological structure (modulo issues about individuating > concepts by referent). That seems to fit with your last sentence > above. > > > (3) Tim raises the case where I'm shifted to Twin Earth in my sleep. > As I said in the last message, I think Tim is right that my "water" > won't initially refer to XYZ, though it may after a couple of years. > I think the moral of that is that the PI of "water" anchors one to > one's past environment, as Tim suggests. Basically because it picks > out (at least in part) the stuff that one has causally interacted with > in the past. In fact I think that for many of these cases, the > present environment is mostly irrelevant (except to determining > whether one's beliefs are now true or false, etc). The future > environment even more so, though maybe one could come up with cases > where it matters. > > All thoughts and reactions are welcome! > > --Dave. > Timothy J. Bayne RM. 213 Social Science Department of Philosophy University of Arizona Tucson, AZ 85721 USA Hm ph. (520) 298 1930 From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 18:10:24 1999 Date: Wed, 17 Feb 1999 18:07:56 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Here are a few thoughts on Josh's "water" case. In this message I'll give my own analysis, and in the next I'll give some thoughts on others' analyses. The first thing to get straight on is whether one thinks the judgments in question are true or false, before and after the vat. Case 1: I think "water is H2O" in the vat. I escape from vat, find a similar-seeming world, except that the watery stuff is XYZ. Case 2: I think "water is H2O" in vat, until a within-vat revolution convinces me of "water is XYZ". I escape from vat, find a similar-seeming world, and the watery stuff there is XYZ. Q1: What (if anything) do I refer to as "water" pre-escape? Q2: What (if anything) do I refer to as "water" post-escape? Q3: Is my statement "water exists" true or false pre-escape? Q4: Is my statement "water exists" true or false post-escape? There are various things to say here. It's probably simplest to first set aside the belief changes re XYZ and H2O, and think about what "water" might refer to pre- and post-vat in an ordinary BIV case. Here some distinct issues pecular to BIV cases some up. For example, many would say that while in the vat, my term "water" doesn't really refer to anything, and that my thought "water exists" is false. After all, isn't a brain in a vat a classic case of someone who is deceived about the external world? One could accommodate this line in the 2-D framework by saying e.g.: (1) the PI of "water" picks out something like "the watery stuff in the environment that the being at the center has been causally related to" (not an implausible analysis of my "water" PI), and (2) if I am a BIV, my centered world (i.e. my actual world, not merely the world-as-I-believe-it-to-be) doesn't have anything that satisfies that PI. (There's no watery stuff causally related to me in my actual environment.) So the PI will pick out nothing, and my vatted thought "water exists" will be false. On this line, what will happen when I come out of the vat? Well, arguably when I emerge for the first time into the "real world", my term "water" doesn't yet refer to actual water, because there's no causal connection yet. But after interacting with the environment for a while, the causal connection will get going, and perhaps eventually my term "water" will come to refer to the actual stuff (H2O). Note that all that will be compatible with the PI analysis above: nothing will satisfy the PI at stage 1 (centered on me in vat) at stage 2 (centered on me just after escape), but something may satisfy it at stage 3 (centered on me well after escape), as by then the center will have plenty of causal relations with some watery stuff. Another line that some philosophers take is that while in the vat, my terms like "water" refers to some chemicals or nutrients in the vat around the brain, because these are what I'm causally connected to. I don't find this very plausible myself, but if one takes this line, one could accommodate it by dropping "watery stuff" from the PI above, and by saying that the PI picks out roughly "the stuff in the environment that is causally responsible for the being at the center's use of the word 'water'". In this case, we'll have "water exists" true at stage 1 (it picks out chemicals), arguably false at stage 2 (t still picks out chemicals, but I suppose that could be true since the chemicals still exist), and true again at stage 3 (by now, the PI picks out real water). Alternatively, one could be tempted by the idea that while in the vat, one's term "water" refers to something in one's "virtual world", or something like that, so that one's vatted thoughts "water exists" is true. This has the problem of making it hard to express how the BIV is deceived, and could even end up leading to some sort of idealism or phenomenalism. But there is arguably something to the intuition that there's some sense in which the BIV's "water is H2O" is truer than "water is XYZ" (assuming the virtual world is an H2O-world). Maybe one could accommodate this by saying there's two ways to read such thoughts, a strong way in which they all come out false, and a "weak" way where some come out true, or some such. E.g., the "weak" way might involve prefixing "According to the virtual world", or might have a PI involving "seeming-objects which seem to be causally connected to me", or some such. This would be a tricky row to hoe, but maybe there are possibilities. In this case, one could say that reference of "water" gradually shifts after release, just as above. Anyway, I think one can factor away from these tricky BIV issues for considering Josh's main point. I'm not sure whether Josh's point would work equally well if the vat world were replaced by another planet in which one is brought up, and that brought to earth. That would probably give similar results at least to line 3 above, and arguably to line 2, though maybe not to line 1. But the crucial claims about how "water" refers after release may be similar each way. What to say about case 1, where I believed "water is H2O", and find out post-vat that the watery stuff is XYZ? I'd be tempted to say, as in the cases above, that immediately post-release my term "water" still refers to whatever it referred to just pre-release: either nothing, or some chemicals, or perhaps some virtual H2O in the virtual world. So my statement "water exists" then will be false, or in any case if true won't be made true by the actual watery stuff. But after a while of interacting with the real environment, my term "water" might well come to refer to the XYZ, and my claim "water exists" will be made true by the actual environment. I think that's so despite the fact that I believed in the vat that water was H2O. After all, within the vat that belief was potentially malleable (if people had told me the watery stuff was XYZ, I'd have accepted that water was XYZ), so the same ought to be the case outside the vat. In fact, I think that even if no-one outside the vat ever tells me that water is XYZ, my term "water" may eventually come to refer to the XYZ by virtue of my causally interacting with it, etc. In this case, I'll eventually have a false belief that "water is H2O" but a true belief that 'water exists". If that's how things so, I think it's compatible with the PI staying constant throughout. E.g. if the PI is "the watery stuff in the environment that the being at the center has been causally related to", this seems to give the right results: it picks out nothing at stage 1, nothing at stage 2, XYZ at stage 3. And that's so irrespective of my belief changes. I'd be tempted to say something similar about case 2. Here, I think there still needs to be a gradual reference-shift from vat-reference to actual reference. Even if one thought "water is XYZ" in the vat, and actual watery stuff is XYZ, it's not clear that one's thought (immediately post-escape) "water is XYZ" is true, or at least it's not clear that it's made true by the actual XYZ. It takes some interaction for the term "water" to pick out the actual stuff. [A side note: It may be that if the scientists tell you immediately "you were in a vat, but now you're out, and fortunately the real world is a lot like you thouight the vat world was", you might decide to make a new stipulation that all your words will refer to actual things rather than vat things, rather than keeping a lot of terms around that don't refer to anything useful. That's tricky in some cases: e.g. re "my wife", or "Ethel", say, does one really want to say that the term refers to a person in the actual world who you've never strictly speaking met before? But I suppose one could try to stipulate that at least some terms refer to their qualitative counterparts. If that's so, then "water" might immediately pick out the respective watery stuff. As for what to say if there are a few small differences between vat world and actual world (e.g. they replaced H2O by XYZ), that's awfully tricky. I guess it depends on one's stipulation. One might stipulate "in case of any differences, my term doesn't refer", or one might stipulate "in case of any differences, my term refers to whatever is superficially similar enough", or some such. In the first case, my post-escape "water exists" might at first be false; in the second; it might be true. But that all depends on just what stipulation one makes. Such a stipulation might well change one's term's PI a little bit, as any new stipulation will. Probably it's easiest to think about these things in a case where the scientists don't tell you you've escaped, so one isn't faced with the immediate wrenching discovery that forces you to reevaluate your whole conceptual scheme. Instead, maybe do things so that you don't know you've undergone such a radical shift. Even so, we can evaluate truth and reference, and we can still consider what happens if there turn out to be small changes in the environment. That way the "stipulation" aspect is removed.] Anyway, that's how I'd be tempted to analyze these cases. I think its plausible on most ways of doing things that one's PI stays constant, but its referent may gradually shift, in effect due to a change in the centered world in question (or at least, a change in the location of the center). The one case where my PI may change is the case where I know that I've shifted environments, and decide to restipulate the use of my terms; but that's a very special sort of change. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 17:24:53 1999 Date: Wed, 17 Feb 1999 18:24:00 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Timothy J Bayne Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Some thoughts on Josh's BIV case and on de se representation. . . The BIV case is a nice case, and a difficult one. I had an answer, but talked myself out of it. Here's another stab. This is in three parts, all rather convoluted. (1) Part of what is up for issue is whether the primary intension of our concepts like are egocentrically indexical, or whether they are more communal. Suppose that Josh is the only BIV, and that he thinks of water as . On this construal of the 1-intension, Josh's water concept refers to certain types of electrodes, or computer programs, or whatever it is that causes most of his watery experiences. (Why do we describe it as his "water concept"? Cos it's connected to his watery experiences, which are connected to our water concept.) This, I take it, is the Putnamian position on BIVs. On a *communal* interpretation of the 1-intension of water, it is (There are lots of indexicals in it). Now, whether this concept still refers to the electrodes or programs of whatever depends on whether Josh is the only BIV around. If the vast majority of the folks around here are also BIVs, then perhaps one could argue that even now it still refers to these electrodes. (Of course, the problem is that Josh isn't communicating with any of these - presumably. So maybe the communal condition can't get a toe-hold in giving the referent of an individual BIV's concept.) But let's suppose that the 1-intension of Josh's is communcal, and that he's the only BIV around here. Then, his thoughts about water would be mostly mistaken cos they are about water, and he isn't connected to water in the right way. Although of course his belief that water is H20 would be true if the stuff that causes most people's watery experiences is H20. Of course, it may not always be easy to tell whether one of his thoughts is true or false, because it may not be clear whether it is about water. Suppose that he thinks that he is having a shower: 'Ah, a lovely shower, this stuff sure is hot' and such like. This might be a true thought if the stuff that is causing his experiences is indeed hot. But insofar as he thinks that the stuff that is hot is the same kind of stuff that everyone else (i.e. non-BIVs) thinks is hot when they are taking a shower, his thought is false. Similar comments applying to his thought, 'It sure is good to run naked through the water of this Bondi surf.' This is true, insofar as it is good to run naked through the Bondi surf, but of course it is false in that Josh-the-BIV is not running anywhere, least of all running naked through the Bondi surf. (2) Let's push the Putnamian line for a minute and see where it goes. Take Josh-the-BIV's water concept to refer to whatever causes his watery experiences. Post-vat Josh's water concept also refers to the stuff that causes his watery experiences. These two concepts (mental representations) are similar in that they are both related to watery experiences, that is, they are both object concepts that are related to the *phenomenal concept* . But should we think that there is any sense in which VAT-Josh and post-Vat Josh have a concept, i.e. that can be tracked from the VAt to after the VAt? I wonder. Both in the Vat and after the vat Josh has a concept that is related to , but that doesn't seem sufficient to justify the claim that he has a single concept that survives the experience from VAt to actual world, that we can reidentify as the same concept. After all, both my concept of a certain chemical substance (say, glump - suppose that glup is watery) and my concept are connected to , but that doesn't give me any incentive to identify with . In order to identify the concept that we are calling Josh's pre-Vat water concept with his post-Vat water concept it needs to be the case that most (much?) of what he has learnt about water in the VAT holds true in the actual world. Does envatted Josh's knowledge of the microstructure of water play an important role here? I don't see why. On an inferential role account of concepts, beliefs about the microstructure of a substance might be just one sort of inferential node among many. If the microstructure of XYZ is such that it behaves in almost all of the same ways that H20 behaves, then we might be inclined to say that Josh has learnt that water is XYZ, or whatever it is in the actual world. (Of course, XYZ can't beahve in all of the same ways that H2O does, otherwise we would be able to discover that some stuff is H20 rather than XYZ.) To sum: supposing the individualistic spin on water, in which the primary intension is we (and Josh) might still be tempted to identify his in-Vat watery concept with his post-Vat watery concept, even thought it refers to a different stuff, because many of the inferential connections that he learnt while in the vat still hold true in the actual world. If post-Vat Josh thinks of his envatted experiences as including water thoughts, then it is clear that he must have a water concept that we can trace from the Vat to the real world. (3) Here's a similar case to Josh's that puts it in a communal setting. Suppose that we are all shifted to Twin Earth in our sleep. Twin Earth is as the travel brouchures describe it: it's the same as earth, except the predominent watery stuff in the environment is XYZ. AFter a couple of days we would discover that "this stuff in XYZ". I guess most of us would say, "Huh, I guess that water isn't H20 after all, it's XYZ." But it seems that we would all be mistaken. Water is H2O, but this ain't water. Of course, after a couple of years on twin earth, 'water' would come to refer to XYZ. So it seems. Well, I'm not sure what the point of this scenario is. Maybe it raises the question: 'Is the 1-intension of "the stuff around here that causes most of our watery experiences", or is the 1-intension of water "the stuff that is causing and has caused most of the watery experiences around here at the moment and in the past." The suggestion is that the 1-intension of natural kind terms anchors them to the past (and maybe the future). Tim University of Arizona Tucson, AZ 85721 USA Hm ph. (520) 298 1930 From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 20:31:28 1999 Date: Wed, 17 Feb 1999 20:29:40 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Some thoughts on analyses of Josh's case by Josh, Erik, Erik, Angela, and Tim. N.B. This presumes the previous message with my own analysis. Josh wrote: >Now here is the puzzle. Primary intentions pick out "watery stuff" in >the actual world. My intuition is that the actual world in both cases >is the post-vat world. However, if I were the one pulled out of the >vat in case 1 I'd say that there is no 'water' in the actual world. >There is just watery stuff. But if I were pulled out of the vat in >case 2, I'd say there is water in the actual world. > Like others, I'm not sure I share your intuitions here. Maybe there's some reason for saying that post-escape, your term "water" doesn't refer to anything in the actual world. But I think one can argue when it does begin referring to something in the actual world, it will refer just as soon in each case. After all, the only difference between the cases is that in one I believe "water is H2O" and in the other I believe "water is XYZ". This might make a difference to reference if these were "core beliefs" for me and unrevisable, or some such. But in fact it seems that these beliefs are quite revisable, a posteriori beliefs, as reflected by the fact that upon "discovering" in the vat that watery stuff is XYZ, one would say "water is XYZ" is true, not that water doesn't exist. I think the same goes outside the vat. Insofar as my word refers to stuff outside the vat at all, it's in virtue of its role as watery stuff, and it doesn't matter whether it turns out to be XYZ or H2O. >If you share these intuitions, then there seem to be two possibilities: >1) My primary intention of water includes the chemical make up of >water or 2) Primary intentions do not fix reference in the actual >world but fix reference in some other way. Hmm. I don't share the intuitions, but perhaps one can make sense of your intuitions by saying: the PI of water picks out "whatever has the same structure as the stuff I've been related to in my past environment", and going for a view (the third line in my previous post) on which in the vat environment, you're related to virtual-H2O and virtual-XYZ in case 1 and case 2 respectively. Then post-vat, out in the real XYZ-world, your "water exists in this environment" might come out true in case 2 but not in case 1 (because XYZ has the same structure as virtual-XYZ, but not as virtual-H2O). That would seem to give the results of your intuitions, though I don't know whether (a) it gets at what underlies your intuitions, or (b) it's plausible. [N.B. If things work this way, it's immediately post-escape that the changes will show up, rather than later. I.e. even before causal interaction with the new stuff, my "water exists around here" will be true in case 2 rather than case 1. With enough causal interaction, reference might still shift so that the claim will be true in both cases.] This analysis may get at something about what you're saying re "PIs don't fix reference in the actual world". I'd say that even on this analysis, they do fix reference in the actual world, but the actual world is a 4-D spacetime manifold, i.e. it includes all the facts about the past. And here, your PI will fix reference partly by appealing to facts about your past environment (i.e., the way things are in the past environment of the being at the center of the world) rather than your current environment. (It's plausible that something like that applies to many cases. E.g., if I am transported to Twin Earth without knowing, then presumably my immediate claim "there is water around here" will be false, even though there is watery stuff (in the loose sense) around there. Essentially, this is because the PI of water fixes reference to something like "the watery stuff I have causally interacted with", which involves reference to my past environment as well as my current environment. Of course if I live on Twin Earth long enough, my reference may shift enough so that "there is water around here" will be true; that's also predicted by this PI.) Looking at your (Josh's) more recent second message, maybe something like this analysis fits. >Regarding the responses to my BIV cases, I'd like to make a few >clarifications. There are two ways that I see of accounting for my >intuitions. The first is to claim that the chemical make up of water >is part of my PI. The second is to claim that my PI is just like >everyone elses BUT PIs don't map centered worlds considered as actual >onto referents. > >I think the second of these claims is the more likely. The idea would >be that PI's perhaps map centered worlds thought to be actual during >concept formation onto referents. (I already see several problems >with this definition, but that is the basic idea.) Since vat-Josh >formed his concepts while thinking that vat-world was actual, he would >then map water to clear drinkable liquid in vat-world. And in >Vat-world water is H2O. Well, one problem with what you say above is that if I grow up (in the actual world) thinking water is XYZ, then this formula will predict that my claim "water exists" is false, which doesn't seem right. And more generally it seems that it will be hard to have false beliefs while one is growing up, which seems problematic. Maybe it's better to say that something special is going on with the vat case -- these aren't just any old false beliefs about the world, but a whole virtual environment, one that is so self-consistent and central that it deserves to count in some sense as my "actual environment" while I'm in the vat. If so, then one can argue that my vat terms fix reference to things in my virtual environment, and that even after leaving the vat, they still involve the way things were in the vat environment. Note again that this doesn't involve ignoring the actual world, but it does involve concentrating on a particular aspect of it, i.e. my past environment, which in this case will be my past virtual environment (which of course is in some sense part of the actual world). On this way of doing things, the escape from vat to world is just like going from Earth to Twin Earth (in case 1) or like going from Twin Earth 1 to Twin Earth 2 (in case 2). Either way, one shifts environment, and one's initial reference is fixed to the stuff in one's original environment, so the truth of one's immediate-post-move claim "there is water around here" depends on whether there is anything around here with the same structure as the stuff in one's past environment. So true in case 2 but not in case 1. That seems to accommodate your intuitions, even with a constant PI that fixes reference via the actual world (including past environment). Anyway, note that one this way of doing things, what makes the differences between the cases isn't just that one had different beliefs in the past about one's environnment -- it's that in some sense, one actually had different environments in the two cases (albeit virtual environments, in Josh's version). To see the difference, think about what we'd say if (a) the virtual environment "really had" XYZ structure in it (i.e. if you'd looked under a virtual microscope, that's what you'd have found), but (b) you believe the whole time that there is H2O there (you never look under a microscope). I predict that in this case, Josh will say (or ought to say) that "water is H2O" is false when thought in the vat, and false after escape, while "water is XYZ" is true both times. That suggests that it's the past environment that matters, not the past beliefs. Of course one can question (a) whether a "virtual environment" is really enough of an environment to make concepts refer and to make beliefs true, and even then one could question (b) whether the "real" XYZ environment really qualifies as having the "same sort of stuff" that is present in the virtual environment (one might argue that there are deep differences between virtual XYZ and real XYZ, so that the "virtual" concept won't pick out the "real" stuff after escape). But assuming one can make a case for the appropriate answer on those two questions, then Josh's position will come out as reasonable, and the PI above seems to capture roughly what's going on. Re Erik L's, Angela's, and Erik H's messages: I think my intuitions largely agree with these, i.e. that there's no major difference between the two cases: insofar as case 2 "water" picks out the actual stuff post-vat, so does case 1 "water". (Although after going through the above, I have to say that I feel at least some of the tug of Josh's way of doing things, at least on the "virtual environment" reading.) I think one still needs to distinguish what happens immediately after release from what happens after a period of adjustment, though. I'm tempted to say that there is an element in the PI of "water" requiring a previous causal connection to the stuff, so it doesn't refer to the actual stuff immediately (in either case), but only after a while. If one doesn't agree with the causal requirement, one could go with a looser "the watery stuff around here now" PI, so that I'll come out referring in both cases immediately after escape. But I think the causal requirement is independently plausible. I do largely agree with what Angela says here: >I understand the temptation, especially given the amount of scientific >information that is part of common knowledge, to say that H2O really is >part of the primary intension of water, but, the very fact that I (and >presumably others) can make sense of counterfactuals like, "Water could >have been other than H2O," or "Water might not have been drinkable by >humans", or "Water might not have had the property of appearing blue in >daylight," etc...strongly suggests that almost nothing of what we know a >posteriori of water is contained within the primary intension. One thing to note is that "could have" and "might have" counterfactuals are often ambiguous between a PI and a SI reading, i.e. between (broad) epistemic possibility and Kripkean "metaphysical" or "subjunctive" possibility. I'm not 100% sure which (or both) you're meaning to appeal to here. Arguably one needs both to make best sense of these counterfactuals. If one accepts Kripke's subjunctive intuition, "water could have been other than H2O" will be true in the epistemic (PI) sense but not the subjunctive (SI) sense. As for "water might not have been blue", etc, this is most straightforward for me to make sense of on the subjunctive (SI) sense: one pictures the actual stuff, i.e. H2O, having a different appearance. On the epistemic (PI) sense, this is harder to make sense of, since water's appearance properties are relatively close to its PI, i.e. are "relatively a priori" of the "water" concept. But presumably one can still make sense of them by considering certain strange (broad) epistemic possibilities, e.g. on which there has been an optical illusion, or water's surface properties are very different from how one thinks they actually are. That reflects the fact that the beliefs in question aren't so "core" as to be unrevisable. As one builds in more and more core surface properties of water, though, this gets harder to do, and the epistemic "might have" claim becomes harder to make sense of. Re Tim's message: (1) Tim raises a possibility I didn't mention in the last message, that my vatted "water" concept may be deferential, fixing reference via my community. Still, I'd argue that it seems wrong to say that I defer to the people out there in the real world outside the vat, and it also seems wrong to say that I defer to other BIV's who happen to be out there. (It doesn't seem right to say that those people's reference makes a difference to my reference.) If I'm deferring, it's to people in my "virtual" world, the ones I think I "got" the term from. (I suppose it could be that those virtual people in my world are "avatars" of "real" people or of other BIVs, so I'd be deferring indirectly to those people, but I'll set that possibility aside.) If that's so, then I don't think deference makes an essential difference to the picture. If the non-deferential concept picks out nothing, or chemicals, or virtual XYZ, so will the deferential concept, on this way of looking at things. The main difference will be the precise nature of the "causal chain" connecting me to the referent, i.e. on whether it needs to go through other "virtual people", or not. (2) Tim also raises the question of in what sense I'll have the same concept inside and outside the vat. This seems most straightforward in the case where I don't know I've shifted. In that case, there seems to be the same sort of continuity that one would find if I go from one world to another without knowing. I'll deploy a lot of concepts, and they certainly won't be a whole new repertoire that I've acquired overnight -- rather, they'll be the old concepts (although a lot of my resulting beliefs may now come out false). It may be that eventually reference will shift, so that in a certain sense I'll have different concepts -- I'll at least have concepts with a different referent and SI, though not a different PI -- but that's a slightly different phenomenon. There's also the special case where I'm told I've shifted, and where I make a new stipulation about my concepts so my beliefs don't come out useless. That would plausibly involve a "different concept" in a reasonably strong sense, but that also seems to be something of a special case. Tim raises some worries about concept identity: >But should we think that there is any sense in which VAT-Josh >and post-Vat Josh have a concept, i.e. that can be tracked from >the VAt to after the VAt? I wonder. Both in the Vat and after the vat Josh >has a concept that is related to , but that doesn't seem >sufficient to justify the claim that he has a single concept that survives >the experience from VAt to actual world, that we can reidentify as the >same concept. After all, both my concept of a certain chemical substance >(say, glump - suppose that glup is watery) and my concept are >connected to , but that doesn't give me any incentive to >identify with . I presume you're talking about the vatted concept and the long-post-vat concept. In favor or "same concept" is that they have the same PI, and a continuous history of use between them. Against "same concept" is that they have a different referent and SI. What answer one gives will partly depend on whether one individuates concepts by PI, SI, referent, or history of use. I think this is somewhat terminological. I have some temptation toward "same concept", given the centrality of PIs, but then the different in referent and SI is pretty significant, too. As for water/glump, I'm not sure how to imagine the case -- do they pick out the same referent, or not? If the same referent, presumably we have same SI, different PI, so the reverse of the above. If a different referent, presumably we have different PI and SI, which is worse (presumably even if the PI of "glump" involves "watery", it will be in a different way, as it isn't a priori that water is glump). Either way, one relevant different is that the vat case involves sameness of PI and continuity of use, whereas this case doesn't. >To sum: supposing the individualistic spin on water, in which the primary >intension is we (and >Josh) might still be tempted to identify his in-Vat watery concept with >his post-Vat watery concept, even thought it refers to a different stuff, >because many of the inferential connections that he learnt while in the >vat still hold true in the actual world. If post-Vat Josh thinks of his >envatted experiences as including water thoughts, then it is clear that he >must have a water concept that we can trace from the Vat to the real >world. I think I agree with this, more or less. There are actually two issues: (1) sameness of concept pre- and post-vat, and (2) whether the concept picks out the actual stuff. That the "inferential connections still hold true in the actual world" seems to be getting at (2) rather than (1). In my terms, I'd put it by saying that as long as the actual world isn't *too* different from the vat world, actual watery stuff can come to satisfy the PI of my "water" concept (though possibly with a time lag). I'm inclined to think that (1) doesn't turn so much on the actual facts, but more on facts about the subject's psychological structure (modulo issues about individuating concepts by referent). That seems to fit with your last sentence above. (3) Tim raises the case where I'm shifted to Twin Earth in my sleep. As I said in the last message, I think Tim is right that my "water" won't initially refer to XYZ, though it may after a couple of years. I think the moral of that is that the PI of "water" anchors one to one's past environment, as Tim suggests. Basically because it picks out (at least in part) the stuff that one has causally interacted with in the past. In fact I think that for many of these cases, the present environment is mostly irrelevant (except to determining whether one's beliefs are now true or false, etc). The future environment even more so, though maybe one could come up with cases where it matters. All thoughts and reactions are welcome! --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Thu Feb 18 16:44:56 1999 Date: Thu, 18 Feb 1999 12:47:46 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Josh Cowley Subject: Minutes from Tuesday To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Here are the minutes from Tuesday. I thank Brad for filling in some details. The first thing we discussed was why we should care about the primar intentions. Roughly the line was, "if water *is* H2O then why do we care what water could have been. It was suggested that the reason is that if, via the PI we could determine that it was logical impossible for water to just be H2O, then we know that we have more to look for. This, of course, is the basic line for consciousness. The bulk of the rest of the dicussion was on proper names again. Thony argues that if the PI of a name is what picks out its referent, then two names which pick out the same object in our world could pick out different objects in another world. But it seems that as a matter of logic, if 'a=b', where 'a' and 'b' are names, then necessarily 'a=b' I suggested that in this framework proper names are not being used as labels for objects. Rather they are functions which pick out objects of a certain description. I think Thony and I now agree that the question is whether names ought to work this way. > Brad tried to provide motivation for thinking that names do in fact have > descriptive content. One such motivation comes from considering cases in > which co-referring names can't be substituted salva veritate (such as in > belief contexts). He also mentioned substitution failures in "simple > sentences" (see Saul 1997), such as "Lois Lane kissed Superman before she > kissed Clark Kent". > We also briefly discussed again the matter of what the primary intensions > of names typically are (such as in the London/Londres case discussed > before). We then dicussed my BIV cases. From an intuition point of view we were evenly split about whether or not there was really any water after you were pulled out of the vat. Brad suggested that it may just be very difficult to think of this case as anything other than a person jumping between two possible worlds. This, of course, is impossible. But it does look kind of like that. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 17 12:01:52 1999 Date: Wed, 17 Feb 1999 12:59:13 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Anthony T Lane Subject: possibility? To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO I think this is probably getting somewhat off the topic, but I have a general worry about making sense of secondary intensions. The 2-intension of water is H2O, given that, in this world, we take it that this chemical structure explains the various properties we associate with water. More generally, it seems that secondary intensions are different from the primary intension only when there is some sort of reductive explanation of a particular term or thing (p62). It is not, of course, directly observable that water is H2O. Rather, it is the result of theory constructed from a variety of experimental results. Consider a possible world in which water is H2O, but in which the natural laws are such that the elements behave radically different than they do in our world. Suppose, for example, that Hydrogen atoms form extremely unstable compounds when they are in the presence of other Hydrogen atoms. In this possible world, the dominant 'watery stuff' is XYZ. Thus, XYZ is the stuff picked out by the primary intension of water. XYZ behaves experimentally much like H2O behaves in our world. But XYZ is not composed of two Hydrogen atoms and an Oxygen atom in this possible world. It seems to me that the secondary intension of water cannot just be H2O. It must also include a host of other information regarding the underlying physical theory that makes this identification meaningful. If there is a possible world in which XYZ behaves for the most part like H20, only X, Y and Z are not 2 H's and an O, it seems strange that there is any value in the secondary intension of water picking out H2O, stuff that is, in this possible world, unstable. If this isn't just confused, it may be that this is a reason why you say primary intensions are particularly important to your account. Anthony From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 01:44:18 1999 Date: Fri, 19 Feb 1999 01:44:04 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: Minutes from Tuesday To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO A few comments on the minutes. >The first thing we discussed was why we should care about the primar >intentions. Roughly the line was, "if water *is* H2O then why do we >care what water could have been. It was suggested that the reason is >that if, via the PI we could determine that it was logical impossible >for water to just be H2O, then we know that we have more to look for. >This, of course, is the basic line for consciousness. Well, many of the reasons for caring about PIs are related to Frege's reasons for caring about senses. In particular, PIs seem to capture the epistemic relations between thoughts and the cognitive contents of thoughts, and are deeply related to notions such as epistemic possibility and apriority. This leads arguably to a central role for them in philosophy of language and mind (in giving an account of content that is useful for epistemic and cognitive purposes) and in epistemology. Within metaphysics, PIs are especially useful for analyzing modal issues. In particular, it seems that PIs are central in assessing conceivability claims, and the sense in which conceivability is a guide to possibility: if P is conceivable, it's generally plausible that there is a possible world satisfying the primary intension of C. And they are central to claims about a posteriori necessity. What's distinctive of a posteriori necessities seems to be that they have a necessary SI but a contingent PI. Given that for many issues in metaphysics we are concerned with questions about the relationship between conceivability and possibility, and about apriority and necessity, PIs seem pretty vital. >The bulk of the rest of the dicussion was on proper names again. >Thony argues that if the PI of a name is what picks out its referent, >then two names which pick out the same object in our world could pick >out different objects in another world. But it seems that as a matter >of logic, if 'a=b', where 'a' and 'b' are names, then necessarily 'a=b' > >I suggested that in this framework proper names are not being used as >labels for objects. Rather they are functions which pick out objects >of a certain description. > >I think Thony and I now agree that the question is whether names ought >to work this way. I'm a little concerned about this bit. Remember, one of the central points of the 2-D framework is that when S is necessary (in the usual sense), S has a necessary secondary intension. So what follows from the necessity of "a=b" is that "a=b" has a necessary SI, i.e. that 'a' and 'b' have the same SI. Nothing about PIs follows. In particular the claim that 'a' and 'b' have different PIs is completely compatible with the claim that "a=b" is necessary. (Just look at "Hesperus = Phosphorus", etc.) Once can think of the PI as being what picks out reference in the actual world (depending on how it turns out) and the SI as picking out reference in counterfactual worlds (given that the actual world is fixed). I.e., PI gives reference across worlds considered as epistemic possibilities, SI across worlds considered as subjunctive possibilities. Given this, then in a case such as the above (with different PI, same SI), 'a' and 'b' will pick out the same referent in all counterfactual worlds (all subjunctive possibilities), as Kripke et al lead us to expect. All that follows from the difference in PIs it is epistemically possible that reference comes apart, i.e. there are epistemic possibilities such that if they turned out to be actual, reference would come apart, which seems to be correct in these cases. Proper names are still being used as labels for objects in this framework, and they are still rigid designators. They pick out the same object across counterfactual worlds (considered as counterfactual). There's a sense in which actual-world reference is mediated by something a bit like a description (see previous note), but that's a far cry from saying that names here function like descriptions (which are nonrigid, etc, and have very different SIs). Witness the difference between "Hesperus" (the name) and "evening star" (the description). I had hoped that all this was clear by now, but the above suggests that maybe there is still a little confusion. Or maybe I am just misreading the above, in which case feel free to elaborate on the issue. >> Brad tried to provide motivation for thinking that names do in fact have >> descriptive content. One such motivation comes from considering cases in >> which co-referring names can't be substituted salva veritate (such as in >> belief contexts). He also mentioned substitution failures in "simple >> sentences" (see Saul 1997), such as "Lois Lane kissed Superman before she >> kissed Clark Kent". This is interesting indirect evidence, but I don't think the case for PIs relies on this sort of evidence. The mere epistemic possibility that Superman is not Clark Kent, and the way the reference of a term varies depending on which epistemic possibility turns out to be actual, is enough to make the case. Again, these things don't imply descriptive reference across counterfactual worlds, but they do arguably imply some quasi-descriptive content across epistemic possibilities. Incidentally it's probably a good idea not to get too fixated on issues specific to proper names, as many of those issues (especially those of reference via linguistic causal chains) aren't all that crucial to what comes later. Natural kind cases are probably more crucial. >We then dicussed my BIV cases. From an intuition point of view we >were evenly split about whether or not there was really any water >after you were pulled out of the vat. Brad suggested that it may just >be very difficult to think of this case as anything other than a >person jumping between two possible worlds. This, of course, is >impossible. But it does look kind of like that. Well, it may be more palatable to look at this as a person jumping between two very different parts of one world. A bit like going from one planet to another. In this case, going from a "virtual environment" to a real environment. Different environments in one sense, but still all part of the same world (it's a big world!). From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 02:05:16 1999 Date: Fri, 19 Feb 1999 02:05:05 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: possibility? To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Anthony writes: >I think this is probably getting somewhat off the topic, but I have a >general worry about making sense of secondary intensions. > >The 2-intension of water is H2O, given that, in this world, we take it >that this chemical structure explains the various properties we associate >with water. More generally, it seems that secondary intensions are >different from the primary intension only when there is some sort of >reductive explanation of a particular term or thing (p62). Well, I'd say this is often the case (especially for natural kind terms), but it needn't be the case. E.g. "one meter" has different PI and SI, but no reductive explanation in the vicinity. Similar for "Hesperus". It's true that for natural kind terms, though, we very often identify it with some explanatory microstructure across worlds. So the SI of "water" goes with H2O, of "tiger" goes with some bio-plus-DNA kind, of "heat" goes with molecular motion, etc. I don't think an actual explanation is required -- arguably the SI of "water" picked out H2O in all worlds even before we had an explanation. But it's arguable that the SI of most such terms picks out the explanatory microstructure of the stuff, whatever that happens to be in the actual world, across all worlds. >It is not, of course, directly observable that water is H2O. Rather, it is >the result of theory constructed from a variety of experimental results. >Consider a possible world in which water is H2O, but in which the natural >laws are such that the elements behave radically different than they do in >our world. Suppose, for example, that Hydrogen atoms form extremely >unstable compounds when they are in the presence of other Hydrogen atoms. >In this possible world, the dominant 'watery stuff' is XYZ. Thus, XYZ is >the stuff picked out by the primary intension of water. XYZ behaves >experimentally much like H2O behaves in our world. But XYZ is not composed >of two Hydrogen atoms and an Oxygen atom in this possible world. OK, it seems plausible that the PI of "water" picks out XYZ in this world. The Kripkean may well say that the SI of "water" picks out H2O, irrespective of its instability. >It seems to me that the secondary intension of water cannot just be H2O. >It must also include a host of other information regarding the underlying >physical theory that makes this identification meaningful. If there is a >possible world in which XYZ behaves for the most part like H20, only X, Y >and Z are not 2 H's and an O, it seems strange that there is any value in >the secondary intension of water picking out H2O, stuff that is, in this >possible world, unstable. Interesting point. The Kripkean often talks as if any H2O will be water. They will then describe this world as a world in which water molecules are unstable. It does arguably make sense to consider subjunctive possibilities such as "water molecules might have been unstable", or "if water molecules had been unstable, life on earth would not have evolved". If so, then such worlds (considered subjunctively) are worlds with water molecules in them, and thus (presumably) worlds with water in them. Still, I can see your intuition that thus might not really deserve to count as "water". One could similar consider worlds where H2O molecules make multicolored paints that people hang on walls, or are used in brick buildings, etc. Are these worlds where paintings are made of water, etc? (Even in the subjunctive SI sense?) I can at least see the counter intuition, though I go both ways. How to accommodate that intuition? Maybe (1) Insist that the SI of water still picks out H2O in all worlds, but say that that the molecules only count as H2O if they are quite a bit like our H2O molecules with current chemistry (i.e., being H2O requires more than just having 2 H's and an O).. Or (2) identify the SI of water with something like "H2O, as long as it has at least some of the other chemical or surface properties it has in the actual world". Or (3) go all the way to SI = "watery stuff", which would be a more radical move, letting in XYZ as water in some counterfactual worlds; unorthodox, but some (e.g. David Lewis) have been tempted to argue that there are worlds where water is XYZ. >If this isn't just confused, it may be that this is a reason why you say >primary intensions are particularly important to your account. Well, I do agree that there is an awful lot of slack in what counts as the SI of a given term. It depends some pretty debatble subjunctive intuitions about what it takes be water in a counterfactual worlds. There is also some slack in a PI, but arguably not so much -- we're somewhat clearer about what we'd count as water in different ways the actual world might turn out. This isn't the main reason I think PIs are more important than SIs (the central reason is that I think PIs tie much more closer to the epistemic notions that are at the heart of content, thought, and modality), but it may be one way of getting at the fact that SIs are most relevant in subjunctive and counterfactual contexts, about which our intuitive descriptions are pretty debatable. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 05:25:06 1999 Date: Fri, 19 Feb 1999 05:23:40 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Apriority, necessity, conceivability, possibility To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Recent discussion has been interesting, but we've drifted a little from the central topics for which the 2-D framework will be important for us. In particular, the framework is most useful for analyzing the relationship between apriority and necessity, the notion of the necessary a posteriori, the relationship between conceivability and possibility, and the issue of a priori access to modality. Of course all of these have are central to important issues in metaphysics, not least to analyzing epistemic and modal arguments in the philosophy of mind. So it's worth going over a few of these things. Most of what's contained in this note is very importan to the courset, so read it carefully. DEFINITION OF PRIMARY AND SECONDARY INTENSIONS I'll take a moment to officially define PIs and SIs for statements. So far we've mostly been talking about PIs and SIs for terms, but often it is whole statements that are most crucial. In effect, the PI of a statement is what you get from "composing" the PI and SI of its component terms. Note that I use "primary proposition" and "secondary proposition" for this idea in the book, but now I prefer to use "primary intension" and "secondary intension" for both. The primary intension of a statement S is a mapping from centered worlds to truth-values. The secondary intension of S is a mapping from uncentered worlds to truth-values. They're defined as follows: The primary intension of S maps a centered world W to the truth-value of S in W when W is considered as actual (i.e. considered as an epistemic possibility). The secondary intension of S maps a centered world W to the truth-value of S in W when W is considered as counterfactual (i.e. considered as a subjunctive possibility, with the actual world fixed). We can give a couple of heuristics for evaluating the PI or SI of a statement S at a world W. PI Heuristic: If W turns out to be actual, will S turn out to be true? SI Heuristic: Given that the actual world is the way it is, is S true in world W? Take W = XYZ-world, S = "water is XYZ". If the XYZ-world turns out to be actual, "water is XYZ" will turn out to be true, so the PI of S is true in W. But given that the actual world is the way it is (i.e. the H2O-world), "water is XYZ" is false in the XYZ-world, so the PI of S is false in W. (I abstract away from centering issues here.) (My favorite heuristics are actually a bit more streamlined than this, though maybe a bit less self-explanatory: PI Heuristic: If W is the case, is S the case? SI Heuristic: If W were the case, would S be the case? This ties PIs to indicative conditionals (which basically force you to evaluate the antecedent as an epistemic possibility) and ties SIs to subjunctive conditionals (which make you treat the antecedent as some sort of "fantasy"). But I'll say more on this later.) APRIORITY AND NECESSITY The 2-D framework can be understood as making the following claims concerning the necessity and apriority of statements S. NECESSITY THESIS: S is necessary iff S has a necessary secondary intension. Here, I'm taking "necessary" in the standard Kripkean sense (i.e. "subjunctive necessity"; I'll argue later that this usage is not wholly uncontestable, but set that aside for now). S is necessary in the Kripkean sense precisely if it is true in all worlds considered counterfactually, which is just to say that its secondary intension is true in all worlds. I hope this claim is fairly straightforward, given the definition of a secondary intension. APRIORITY THESIS (weak form): If S is a priori, S has a necessary primary intension. The case for this is pretty straightforward. If S is a priori, then plausibly it will be true no matter how the actual world turns out, i.e. it will be true in any world considered as actual (considered as an epistemic possibility). That is, for an arbitrary world W, if W turns out to be actual, S will turn out to be true. That is, the primary intension of S is is true in all centered worlds. APRIORITY THESIS (strong form): S is a priori iff S has a necessary primary intension. This makes the further claim that if S has a necessary PI, it is a priori. This isn't entirely straightforward to "prove". Necessary PI means that S will be true no matter how the actual world turns out, but it is still a leap from there to the epistemic claim that we can know S to be true. Maybe S could turn out to be unknowable, for example. E.g., maybe some really complex mathematical truth could have a necessary PI but not be knowable a priori? We'll come back to this issue of "scrutability" later. For now, I'd consider the weak form better established than the strong form, but I'd also say its plausible that the great majority of statements with a necessary PI seem to be knowable a priori, so that exceptions to the strong form are rare if they exist. I'm very interested to see any potential counterexamples to any of these principles (or arguments). It's not impossible that I've overlooked something here, so feel free to jig them up. A POSTERIORI NECESSITY THESIS: S is an a posteriori necessity iff S has a necessary secondary intension and a contingent primary intension. This follows directly from the necessity thesis and from the apriority thesis (strong form). I'd say that this is the central claim of the 2-D analysis of a posteriori necessity. It certainly seems to fit all the Kripke examples: "Hesperus is Phosphorus", "water is H2O", etc. Somebody might deny this thesis (presumably by denying the strong form of the apriority thesis), asserting that there is some other sort of a posteriori necessity. I think it's at least highly plausible that such a posteriori necessities would have to have their roots in considerations quite distinct from the ones Kripke raises, though. One reason all this is useful is that it lets us jump from epistemic claims about what is and isn't a priori to modal claims about the existence of various possible worlds. E.g. if S isn't a priori, there's a possible world where its PI is false; and even if S is an a posteriori necessity, its PI will be false in some world. If this is so, it's very important to the epistemology of modality, and to the role of epistemic arguments in metaphysics. What makes all this possible is a central property of the 2-D framework: that it has two dimensions of semantic evaluation (two intensions), but just one space of worlds. Centering aside, PIs and SIs are evaluated at exactly the same worlds. They just give different results because of the different ways the worlds are considered (as actual or as counterfactual). Thus the worlds considered as "epistemic possibilities" for PI purposes are perfectly respectable metaphysically possible worlds: witness the XYZ-world, for example. Just one world, but two different ways of looking at it. That means that when we have an a posteriori necessity, we have a genuine metaphysical possibility in which the PI of the statement is false. That metaphysical possibility may well have consequences for metaphysics. Again, any potential counterexamples and counterarguments are welcome. CONCEIVABILITY AND POSSIBILITY We can also put all this in terms of the relationship between conceivability and possibility. Conceivability is an epistemic notion. There are various notions in the area, as we'll see, but perhaps the simplest is the following. (N.B. I now prefer this way of doing things to the way I do things in the book.): S is conceivable iff S is not ruled out a priori (i.e., if it is not a priori that not-S). If the 2-D framework is right, then one will be able to make an inference from conceivability to possibility. If S is conceivable, not-S is not a priori, so not-S has a contingent PI (by the strong apriority thesis), so there is a metaphysically possible world in which the PI of S is true. That's not to say that S itself is "metaphysically" possible in the Kripkean sense (that sense requires a world where the SI is true), but we still have a metaphysically possible world in the vicinity. Call this the CONPOSS PRINCIPLE: If S is conceivable, there is a possible world in which the primary intension of S is true. We'll be looking at different versions of this claim later, and examining just what it takes to support the relevant versions. But note for now how it fits the Kripkean examples. People often say that the Kripke examples show that there can be no inference from conceivability to possibility. E.g., in the relevant sense of conceivability, "water is XYZ" is conceivable, as is "Hesperus is not Phosphorus", but neither of these is possible. But on the 2-D framework, there is at least a possibility in the vicinity: a possible world in which the PI of "water is XYZ" is true, and one in which the PI of "Hesperus is not Phosphorus" is true. Most importantly, when you conceive (in the relevant sense) of a scenario in which water is XYZ, or in which Hesperus isn't Phosphorus, you *do* conceive of a perfectly respectable world: the XYZ-world in the first case, and a world in which the evening object isn't the morning object in the second. These worlds don't satisfy the statements in question when they're considered as counterfactual, but they do when they're considered as actual. That's no surprise: after all, one is conceiving the scenarios in question as epistemic possibilities (i.e. considering them as actual), so one would expect that the worlds that correspond to this conceivability are worlds that satisfy S when they are considered as actual. So it's a little more indirect than a simple inference from S is conceivable to S is possible, but we still have conceivability as a guide to metaphysical possibility in some sense. MODAL RATIONALISM Note that the two-dimensional framework fits nicely with modal rationalism: the claim that we have a priori access to modality and to the space of possible worlds. Conceivability as described above is a paradigm a priori method. And when we conceive of a scenario in the relevant sense, it appears that we really are conceiving of a possible world (the XYZ-world, or the two-star world). Of course the sense in which it's an S-world is that it satisfies S's primary intension, not its secondary intension. So we have to be careful about just how we describe the worlds in question. But still, conceivability acts as a direct guide to the space of possible worlds. So the way is opened for some sort of modal rationalism. Of course all this depends on the central claims of the 2-D framework being correct. I think there are probably a few points at which an oppnent can argue (e.g. the strong apriority thesis), but its tricky, and again they'd have to go well beyond Kripke. PHILOSOPHY OF MIND All this is very relevant to epistemic arguments against materialism. One response to these arguments is to deny that there is an epistemic gap between physical and phenomenal, and assert that there is an a priori entailment. I sometimes call this "type-A materialism". It's a hard row to hoe, though, because it only seems to work if one accepts a functional analysis of phenomenal concepts, and such an analysis seems very implausible. So perhaps the most popular response is "type-B materialism": accept that there is an epistemic gap, but deny that there is an ontological gap. This doesn't have the intuitive implausibility of the type-A view, but still holds onto materialism. Typically, one will accept that there is no a priori entailment from physical to phenomenal, but will assert that there is a necessary entailment. So the entailment from physical to phenomenal is an a posteriori necessity. This view also usually goes along with the claim that conceivability doesn't imply possibility. So zombies (for example) are said to be conceivable but not possible, so materialism is not endangered. If the central claims of the 2-D framework are correct, this isn't so simple. The type-B materialist holds that "P -> Q" is an a posteriori necessity, where P is the physical truth about the world and Q is a phenomenal truth. On the 2-D framework, it follows that "P -> Q" has a contingent PI, so there is a possible world where the PI of P is true, but the PI of Q is false. Under a bit more analysis, the existence of this possible world is highly threatening to materialism. Similarly, if zombies are conceivable, it follows that there's a possible world in which at least the PI of "there are zombies" is true. From here (under a bit of analysis) one can plausibly get to a world physically identical to ours but different overall, so that materialism comes out false. The basic reason for all this, of course, is that deep down the Kripkean framework hasn't really broken the link between conceivability and possibility, it has just recast it a little. But we've seen that even Kripke's own results rely on the method of conceivability: conceive of a world, then make sure you describe it correctly. So it arguably supports rather than opposes a modal rationalism at the deep level. If so then the Kripkean framework gives much less support to the type-B materialist than is often supposed. We'll be going into the full details of how this applies to materialism etc next week, so we can save a detailed analysis of the consciousness issues for then. But its important to see how the 2-D framework might have significant metaphysical consequences, and how those rest pretty straightforwardly on the central theses of the framework. For now, I'd be interested to see some discussion of those theses, and of the relationship between apriority and necessity vis-a-vis PIs and SIs, etc. All comments are welcome. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 06:21:34 1999 Date: Fri, 19 Feb 1999 06:20:07 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Next week To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO My visa still isn't through, and although there are indications that it shouldn't be too long now, I won't be there by next Tuesday. This means I have to rearrange things a little. Week 5 was originally supposed to be on the "tyranny of the subjunctive", on relationships between the 2-D framework and indicative and subjunctive conditionals, and on implications for the analysis of necessity. But that isn't really written up (except very briefly in the Princeton outline); I'd been planning to mostly talk about it in class. So as things stand, we will skip straight to the following topic (the original week 6) instead. This topic involves applying the 2-D framework to epistemic and modal arguments against materialism. The readings are TCM, Chapter 4 -- mostly pp. 131-149. N&N, pp. 144-55. Mind and Modality, Lecture 1 (esp. sections 5-7). The Mind and Modality notes (on the web) have my currently preferred formulation of the 2-D argument against materialism. It's a bit crisper and tighter than the book version, though it comes to much the same thing in the end. The book version has more discursive detail, so you'll probably want to look at the two of them together (the real core of the book argument is pp. 131-36, but pp. 147-49 also has relevant material). It might also be interesting to look at Kripke's anti-materialist argument in N&N, which is similar in spirit, and compare and contrast. (N.B. The Princeton notes have some contrastive analysis that goes a bit beyond what is in the book.) Anyway, consider these the readings for next week's discussion. I'll look forward to seeing your thoughts on them early next week. As promised, there's also an assignment. I'd like you to take two terms that Kripke discusses, and translate his discussion into the 2-D framework. One is "yard", as discussed on p. 76; the other is "cat", as discussed on p. 122 and pp. 125-6. I'd like you to take every significant part of Kripke's discussion here and translate it into the 2-D framework. Characterize roughly what the PI and the SI of these terms look like, at least given Kripke's intuitions. In places where Kripke considers a hypothetical scenario, you should roughly specify the world, say whether it is being considered as actual or as counterfactual, note what the referent of the relevant term or statement is in that world (considered the relevant way, and according to Kripke's stated intuitions), and say what the upshot is the for relevant PI or SI. Try to translate Kripke's more general points in these passages into the framework too, if you can. If you're so inclined, you can say whether you agree or disagree with Kripke's specific and general analyses and why, though that isn't compulsory. This shouldn't be much work (the passages are pretty short). This will be due by next Tuesday at noon, Arizona time. Don't be late. E-mail it directly to me (not to the mailing list). It should be your own work. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Thu Feb 18 08:10:40 1999 Date: Thu, 18 Feb 1999 09:09:39 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Timothy J Bayne Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO > A quick comment on Dave's comment on my comment on Josh: [For Dave's comments, and some of my original comments, see below.] The point I was making about glumpy stuff was this. In Josh's scenario, I'm told (and believe) that I've just left a Vat. Presumably I will also believe that none of the stuff behind my past experiences is the same sort of stuff as the stuff behind my current expriences. In this situation, I want to suggest that we shouldn't reidentify my in-VAt <"water"> concept with my post-Vat water concept, *even if we're picking out concepts in terms of their primary intensions.* It is true that both in the vat and out of the vat, I have a concept that is closely connected with my phenomenal concept , and it's a natural kind concept too. Call these two concepts 'Water1' and 'water2' (These are just two names, I'm not yet supposing that these are two distinct concepts). When, out of the Vat, I use water2, I don't think of it as applying to the same sort of stuff that water1 applied to. And that, it seems to me, is enough to say that its primary intension is different. Here's an analogy. Suppose I have two concepts, and . I think that glumph and water are distinct natural kinds. Both act in the same (watery) way, indeed, I can't tell them apart. Almost nobody can tell them apart. (Perhaps nobody now living can tell them apart. It's just part of our mythology that there were people - the shabby pedagogues - who could tell them apart.) But I know that the stuff around here is glumph. Still, as far as I'm concerned, I interact with glumph pretty much the same way I would with water. Now, suppose that I move to an area where the watery stuff is not glumph but water. *If I know that this is the case*, I will use a different concept, my concept rather than my , to refer to the watery stuff around me. I want to say that this case is precisely analogous to Josh's BIV case in which he believes that he was a BIV. It may look like he is still using his water concept, but he's not. He's using a concept that has the same surface structure, but there's more to PI than surface structure. AT least for Natural Kinds, one has to think that one is talking about the same sort of stuff in order for "two" primary intensions to be the same. Well, maybe. > There's also the special case where I'm told I've shifted, and where I > make a new stipulation about my concepts so my beliefs don't come out > useless. That would plausibly involve a "different concept" in a > reasonably strong sense, but that also seems to be something of a > special case. > > Tim raises some worries about concept identity: > > >But should we think that there is any sense in which VAT-Josh > >and post-Vat Josh have a concept, i.e. that can be tracked from > >the VAt to after the VAt? I wonder. Both in the Vat and after the vat Josh > >has a concept that is related to , but that doesn't seem > >sufficient to justify the claim that he has a single concept that survives > >the experience from VAt to actual world, that we can reidentify as the > >same concept. After all, both my concept of a certain chemical substance > >(say, glump - suppose that glup is watery) and my concept are > >connected to , but that doesn't give me any incentive to > >identify with . > > I presume you're talking about the vatted concept and the > long-post-vat concept. In favor or "same concept" is that they have > the same PI, and a continuous history of use between them. Against > "same concept" is that they have a different referent and SI. What > answer one gives will partly depend on whether one individuates > concepts by PI, SI, referent, or history of use. I think this is > somewhat terminological. I have some temptation toward "same > concept", given the centrality of PIs, but then the different in > referent and SI is pretty significant, too. As for water/glump, I'm > not sure how to imagine the case -- do they pick out the same > referent, or not? If the same referent, presumably we have same SI, > different PI, so the reverse of the above. If a different referent, > presumably we have different PI and SI, which is worse (presumably > even if the PI of "glump" involves "watery", it will be in a different > way, as it isn't a priori that water is glump). Either way, one > relevant different is that the vat case involves sameness of PI and > continuity of use, whereas this case doesn't. > > >To sum: supposing the individualistic spin on water, in which the primary > >intension is we (and > >Josh) might still be tempted to identify his in-Vat watery concept with > >his post-Vat watery concept, even thought it refers to a different stuff, > >because many of the inferential connections that he learnt while in the > >vat still hold true in the actual world. If post-Vat Josh thinks of his > >envatted experiences as including water thoughts, then it is clear that he > >must have a water concept that we can trace from the Vat to the real > >world. > > I think I agree with this, more or less. There are actually two > issues: (1) sameness of concept pre- and post-vat, and (2) whether the > concept picks out the actual stuff. That the "inferential connections > still hold true in the actual world" seems to be getting at (2) rather > than (1). In my terms, I'd put it by saying that as long as the > actual world isn't *too* different from the vat world, actual watery > stuff can come to satisfy the PI of my "water" concept (though > possibly with a time lag). I'm inclined to think that (1) doesn't > turn so much on the actual facts, but more on facts about the > subject's psychological structure (modulo issues about individuating > concepts by referent). That seems to fit with your last sentence > above. > > > (3) Tim raises the case where I'm shifted to Twin Earth in my sleep. > As I said in the last message, I think Tim is right that my "water" > won't initially refer to XYZ, though it may after a couple of years. > I think the moral of that is that the PI of "water" anchors one to > one's past environment, as Tim suggests. Basically because it picks > out (at least in part) the stuff that one has causally interacted with > in the past. In fact I think that for many of these cases, the > present environment is mostly irrelevant (except to determining > whether one's beliefs are now true or false, etc). The future > environment even more so, though maybe one could come up with cases > where it matters. > > All thoughts and reactions are welcome! > > --Dave. > Timothy J. Bayne RM. 213 Social Science Department of Philosophy University of Arizona Tucson, AZ 85721 USA Hm ph. (520) 298 1930 From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 19:38:16 1999 Date: Fri, 19 Feb 1999 19:38:01 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Tim suggests that maybe the BIV pre- and post-emergence should be regarded as having two sets of concepts, stemming from the fact that he knows he has changed environments. I think this isn't implausible. It's related to what I said earlier about the BIV being able to make a new "stipulation" about its concepts once it knows it's environment has changed so much (or once it knows it was previously so decevied, depending on how you do things). In the case where the ex-BIV doesn't know about the change, presumably this won't work -- it will keep using the old concepts, and for a good while, at least, it will make a lot of claims that are strictly false (e.g., "there's water around here"). Even if the ex-BIV knows about the change, presumably it doesn't need to make the stipulation -- it might correctly say "there's no water around here, just stuff which looks a lot like the stuff I knew as water". But that would probably lead to inconvenience and communication problems for the ex-BIV, who might prefer truth and communication, so it might decide to adopt the conceptual system of his new community wholesale. Of course the new community will be using the same words (or same sounds) as the old community, but that's fairly incidental -- he's really deciding to use new concepts. So maybe his old "water" concept had as PI "the watery-looking stuff I was causally connected to", and his new one has PI "the watery stuff that's around me recently", or "what people call 'water' around here", or something like that. That does seem to be a special case, stemming from (a) the fact that he knows about the change, and (b) the fact that due to some sort of coincidence and conspiracy, his new environment is enough like his old environment that his old words "sort of" work. If one had (b) without (a), one presumably wouldn't get this sort of new concept. And if one had (a) without (b), maybe he'd adopt a new set of concepts, but presumably they wouldn't have such a close relationship to the old concepts (quite different semantics, probably different sounds). But given the special circumstances of (a) and (b), I think Tim's analysis is pretty reasonable. >When, out of >the Vat, I use water2, I don't think of it as applying to the same sort of >stuff that water1 applied to. And that, it seems to me, is enough to say >that its primary intension is different. That's a nice point. The fact that a person knows that a new concept doesn't apply to what the old concept applies to is a good indication that it's a different concept. It's a bit like the a posteriority of "a=b" implying that 'a' and 'b' have different PIs. Here, he can still use his old concept if he wants, and then he'll hold that "water1 = water 2" is false, and so certainly not a priori. So one has different concepts. Unlike the case where he doesn't know about the change, and can't entertain the false identity. The glumph/water case is a nice way of making the point too. The main disanalogy is that here one had both concepts beforehand, but otherwise things seem similar. Again, the fact that "water = glumph" can be false, or even its non-apriority, seems to be a sign of two different PIs and two different concepts. >At least for Natural Kinds, one has to think that one is talking about the >same sort of stuff in order for "two" primary intensions to be the same. That has a certain plausibility. Certainly if one has two concepts 'a' and 'b' and holds that "a=b" is false, that suggests two different PIs for the two concepts. A slight complication, though: There may be cases where I don't so much have two concepts, but hold that "what I now refer to as 'a' is different from what I used to refer to as 'a'" -- e.g., take 'a' = 'here', or 'the president'. Presumably this is compatible with 'a' having the same PI, just a different referent due to the change in location. That's due to a special property of indexicals -- they're not supposed to have a stable referent. Can this happen for natural kind terms? Well, I suppose it could happen if one goes to twin earth, keeps using one's old "water" concept, the reference gradually shifts to XYZ, and only then does one find out about the change. That might be a case where the PI has stayed constant. So maybe one has to qualify your thesis above. What exactly is the different between the ex-BIV shift and these two cases? Something to do with the fact that natural kind referents can't shift *quickly*, given just the one concept. In the ex-BIV case, the old PI should pick out the old stuff just after moving, not the new stuff. So if one has the possibility of a different referent so quickly, that indicates a new PI and a new concept. Of course that constraint doesn't imply to standard indexicals, whose reference can shift in a flash. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Sun Feb 21 19:55:57 1999 Date: Sun, 21 Feb 1999 20:55:04 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Erik J Larson Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Angela brought to my attention a case we were discussing a while back involving quantum mechanics that may have relevance for evaulating the strong apriority thesis; specifically the conditional from "S has a necessary primary intension" to "S is a priori". Consider some quantum mechanical superposed state (of some arbitrary photon impinging on a half-silvered mirror, say). The state is described by the mathematics of Shrodinger's "U" equation, or the linear complex number equations governing the evolution of the photon as it superposes into different possible states. Now, since this is entirely mathematical it ought to have the same status as any "complex mathematical truth"; hence, if the latter has a necessary primary intension so will the former. The actual position of the particle, however (which is what the equations is tracking) is not knowable apriori. After the "reduction" or collapse of the wave function, some distribution of probabilities will tell us where (in what final state) the photon will be, but even now there is no apriori truth--accessible by us anyway--of what the final state of the proton will be in. Worse, there is no way of knowing even while the unitary evolution is underway, although "the position of this particle" is presumbably the primary intension of the equation. If we change this to "the possible positions of this particle" then among most construals of the U-evolution (which is supposed to be entirely deterministic prior to collapse) we are not getting the right primary intension. So it seems in this case (although I admit this is the subject of much debate) that, if an equation of this sort has a primary intension (and why shouldn't it, since it is just some mathematical statement), then the real content of it is not knowable apriori. There is a big debate over whether to treat the collapse or state vector reduction as real or imaginary, and also whether to consider the linear superposition of subatomic phenomena as descriptions of real superpositions or just incomplete descriptions of particle positions (I suppose these are slightly different issues), but the point is that in as much as something like a Schrodinger equation has a primary intension, it is about the positions of particles, and this is not knowable apriori in a very clear cut way, in this case. One other one, take as a formal system the smartest mathematician, construct his Godel number. There is a truth not knowable to any of us but knowable in principle apriori. Erik L. "What our grammarian does is simple enough. He frames his formal reconstruction of K along the grammatically simplest lines he can, compatibly with inclusion of H, plausibility of the predicted inclusion of I, plausibility of the hypothesis of inclusion of J, and plausibility, further, of the exclusion of all sequences which ever actually do bring bizarreness reactions." -- W.V.O. Quine ---------------------- Erik J Larson erikl@U.Arizona.EDU From owner-modality@LISTSERV.ARIZONA.EDU Mon Feb 22 03:52:31 1999 Date: Mon, 22 Feb 1999 03:51:31 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Erik raises a couple of intriguing potential counterexamples to the strong apriority thesis (the claim that S has a necessary PI iff S is a priori). I'll take the second first, as I think I understand it a bit better. >One other one, take as a formal system the smartest mathematician, >construct his Godel number. There is a truth not knowable to any of us >but knowable in principle apriori. Right. On the assumption that each of us is a finite computational system, it follows by a natural chain of reasoning from Godel's theorem that there are mathematical truths that none of us can know to be true. If G is such a statement for the smartest of us (or for the whole community), then presumably G is a necessary truth, and has a necessary PI (there's no difference between PI and SI for mathematical statements). But none of us can know it a priori. Counterexample? This is one of the trickiest cases for the strong apriority thesis to handle. But I think it can be handled by noting (as you note in your last sentence implicitly) that we are using an idealized notion of the a priori. S is a priori if it is knowable a priori, and in considering what is knowable a priori we abstract away from our contingent cognitive limitations. On this way of doing things, it's plausible that G comes out to be a priori. It's true that none of us can know it, but that's just because of our cognitive limitations. There are presumably smarter beings that can come to know it a priori. So G is a priori. [Semi-technical note (don't worry if you don't follow this): To justify this, one can note a result of Kleene's (I believe, though I may be misrecalling the details) to the effect that any arithmetical statement that is undecidable within Peano arithmetic (PA) is decidable in some extension of PA, where the extensions are obtained by repeated "Godelizing" (adding a Godel sentence), onward through the ordinal hierarchy. The trouble with us is that we can't Godelize forever -- being finite systems, we eventually lose track of ordinal counting (which is itself not recursively systematizable). But for any undecidable statement of arithmetic, *some* amount of repeated Godelization will decide it. So all we need is a being smarter than us who can count further through the ordinals. Presumably there's no obstacle to the possibility of such a being; our own specific limits here seem to be contingent cognitive limitations. (Shaughan, feel free to correct or expand on this!)] This is some heavy machinery to bring out, and arguably there are other strategies. Sometimes I'm even tempted to appeal to the possibility of a being who can go through all the integers at once to determine whether "for all n, P(n)" is true, on the theory that our inability to do this is a contingent cognitive limitation, but that might make one a little queasy. There are also questions about what to say for higher set theory and the like. We'll be coming back to these matters a bit later on, but the important thing to note is that for the strong apriority thesis to be plausible, we have to appeal to a notion of the a priori that idealizes away from our contingent limitations. Incidentally I discuss the relevance of this sort of analogy to the mind-body discussion in the book briefly around pp. 138-40. The way I prefer to do things now, in terms of positive and negative conceivability (coming up in two or three weeks), it turns out fortunately that one doesn't need to refute this sort of example at least to make the case re consciousness. A thesis slightly weaker than the strong apriority thesis can do the job for the anti-materialist, and this sort of consideration gets factored out. It still raises independently important issues for modal rationalism, though. Feel free to follow up here, as there are plenty of interesting issues in the vicinity. >Consider some quantum >mechanical superposed state (of some arbitrary photon impinging on a >half-silvered mirror, say). The state is described by the mathematics of >Shrodinger's "U" equation, or the linear complex number equations >governing the evolution of the photon as it superposes into different >possible states. Now, since this is entirely mathematical it ought to >have the same status as any "complex mathematical truth"; hence, if the >latter has a necessary primary intension so will the former. I'm not sure exactly what the sentence is whose PI we are talking about. But in any case I think I might get off the bus here. I don't think the Schrodinger equation has quite the same status as a mathematical truth. As it's being used here, it's a *physical* truth, and as such is both a posteriori and has a contingent PI. Of course the mathematical truth that a certain differential equation has such-and-such solutions is a priori (and has a necessary PI); but here we are concerned with the claim that these solutions correspond to the way things are in physical reality, and that's a different claim altogether. >The actual >position of the particle, however (which is what the equations is >tracking) is not knowable apriori. After the "reduction" or collapse of >the wave function, some distribution of probabilities will tell us where >(in what final state) the photon will be, but even now there is no apriori >truth--accessible by us anyway--of what the final state of the proton will >be in. Worse, there is no way of knowing even while the unitary evolution >is underway, although "the position of this particle" is presumbably the >primary intension of the equation. If we change this to "the possible >positions of this particle" then among most construals of the U-evolution >(which is supposed to be entirely deterministic prior to collapse) we are >not getting the right primary intension. So it seems in this case >(although I admit this is the subject of much debate) that, if an equation >of this sort has a primary intension (and why shouldn't it, since it is >just some mathematical statement), then the real content of it is not >knowable apriori. Well, let's separate a few things. (1) The mathematical fact that the Schrodinger equation has certain solutions. That's a priori and has a necessary PI. (2) The physical fact that the Schrodinger equation describes physical reality. That's a posteriori and has a contingent PI. (3) The physical fact that a certain particle has a certain position. That's highly a posteriori and has a contingent PI. (More technicalities coming up.) I think what you may really be talking about, though, is the conditional from initial state of the wave function plus Schrodinger equation (as a physical law) to final state of the particle. The status of this depends on just which interpretation of QM we are using, as the claim the Schordinger equation is taken to make about physical reality will vary with the interpretation. On the Everett interpretation, I suppose the conditional will be a priori, as the equation is the only dynamics; and indeed, given knowledge of an initial state and of the truth of the interpretation, one can know the (superposed) final state just fine. Something similar arguably goes for the Bohm interpretation. As long as we are given full enough information about wave plus particles in the antecedent, then we can come to know the determinate final state no problem. In a "collapse" interpretation, on the other hand, the conditional will never be a priori, as such interpretations have a further nondeterministic part of the dynamics, so we'll at best be able to figure out a probability distribution. Anyway, it seems to me that in each of these cases, a priority and necessity of PI will line up. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Mon Feb 22 10:16:12 1999 Date: Mon, 22 Feb 1999 11:14:44 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Erik J Larson Subject: Re: 2D questions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Dave, good point, I was using the "2-D" interpretation of QM with the U equation governing particle positions prior to "collapse" and the probability distributions after the collapse. On this view, the conditional wouldn't be apriori. One further comment, the superbeing you mentioned that can run through all integers to determine "for all n, P(n)" could not be formalizable, by Godel's result. If it was formalizable (and consistent), then it would get stuck too. If such a being could exist, it would be an "oracle" about which we could have no computational knowledge (even in principle). Erik On Mon, 22 Feb 1999, David Chalmers wrote: > Erik raises a couple of intriguing potential counterexamples to the > strong apriority thesis (the claim that S has a necessary PI iff S is > a priori). I'll take the second first, as I think I understand it a > bit better. > > >One other one, take as a formal system the smartest mathematician, > >construct his Godel number. There is a truth not knowable to any of us > >but knowable in principle apriori. > > Right. On the assumption that each of us is a finite computational > system, it follows by a natural chain of reasoning from Godel's > theorem that there are mathematical truths that none of us can know to > be true. If G is such a statement for the smartest of us (or for the > whole community), then presumably G is a necessary truth, and has a > necessary PI (there's no difference between PI and SI for mathematical > statements). But none of us can know it a priori. Counterexample? > > This is one of the trickiest cases for the strong apriority thesis to > handle. But I think it can be handled by noting (as you note in your > last sentence implicitly) that we are using an idealized notion of the > a priori. S is a priori if it is knowable a priori, and in > considering what is knowable a priori we abstract away from our > contingent cognitive limitations. On this way of doing things, it's > plausible that G comes out to be a priori. It's true that none of us > can know it, but that's just because of our cognitive limitations. > There are presumably smarter beings that can come to know it a priori. > So G is a priori. > > [Semi-technical note (don't worry if you don't follow this): To > justify this, one can note a result of Kleene's (I believe, though I > may be misrecalling the details) to the effect that any arithmetical > statement that is undecidable within Peano arithmetic (PA) is > decidable in some extension of PA, where the extensions are obtained > by repeated "Godelizing" (adding a Godel sentence), onward through the > ordinal hierarchy. The trouble with us is that we can't Godelize > forever -- being finite systems, we eventually lose track of ordinal > counting (which is itself not recursively systematizable). But for > any undecidable statement of arithmetic, *some* amount of repeated > Godelization will decide it. So all we need is a being smarter than > us who can count further through the ordinals. Presumably there's no > obstacle to the possibility of such a being; our own specific limits > here seem to be contingent cognitive limitations. (Shaughan, feel > free to correct or expand on this!)] > > This is some heavy machinery to bring out, and arguably there are > other strategies. Sometimes I'm even tempted to appeal to the > possibility of a being who can go through all the integers at once to > determine whether "for all n, P(n)" is true, on the theory that our > inability to do this is a contingent cognitive limitation, but that > might make one a little queasy. There are also questions about what > to say for higher set theory and the like. We'll be coming back to > these matters a bit later on, but the important thing to note is that > for the strong apriority thesis to be plausible, we have to appeal to > a notion of the a priori that idealizes away from our contingent > limitations. > > Incidentally I discuss the relevance of this sort of analogy to the > mind-body discussion in the book briefly around pp. 138-40. The way I > prefer to do things now, in terms of positive and negative > conceivability (coming up in two or three weeks), it turns out > fortunately that one doesn't need to refute this sort of example at > least to make the case re consciousness. A thesis slightly weaker > than the strong apriority thesis can do the job for the > anti-materialist, and this sort of consideration gets factored out. > It still raises independently important issues for modal rationalism, > though. > > Feel free to follow up here, as there are plenty of interesting issues > in the vicinity. > > >Consider some quantum > >mechanical superposed state (of some arbitrary photon impinging on a > >half-silvered mirror, say). The state is described by the mathematics of > >Shrodinger's "U" equation, or the linear complex number equations > >governing the evolution of the photon as it superposes into different > >possible states. Now, since this is entirely mathematical it ought to > >have the same status as any "complex mathematical truth"; hence, if the > >latter has a necessary primary intension so will the former. > > I'm not sure exactly what the sentence is whose PI we are talking > about. But in any case I think I might get off the bus here. I don't > think the Schrodinger equation has quite the same status as a > mathematical truth. As it's being used here, it's a *physical* truth, > and as such is both a posteriori and has a contingent PI. Of course > the mathematical truth that a certain differential equation has > such-and-such solutions is a priori (and has a necessary PI); but here > we are concerned with the claim that these solutions correspond to the > way things are in physical reality, and that's a different claim > altogether. > > >The actual > >position of the particle, however (which is what the equations is > >tracking) is not knowable apriori. After the "reduction" or collapse of > >the wave function, some distribution of probabilities will tell us where > >(in what final state) the photon will be, but even now there is no apriori > >truth--accessible by us anyway--of what the final state of the proton will > >be in. Worse, there is no way of knowing even while the unitary evolution > >is underway, although "the position of this particle" is presumbably the > >primary intension of the equation. If we change this to "the possible > >positions of this particle" then among most construals of the U-evolution > >(which is supposed to be entirely deterministic prior to collapse) we are > >not getting the right primary intension. So it seems in this case > >(although I admit this is the subject of much debate) that, if an equation > >of this sort has a primary intension (and why shouldn't it, since it is > >just some mathematical statement), then the real content of it is not > >knowable apriori. > > Well, let's separate a few things. (1) The mathematical fact that the > Schrodinger equation has certain solutions. That's a priori and has a > necessary PI. (2) The physical fact that the Schrodinger equation > describes physical reality. That's a posteriori and has a contingent > PI. (3) The physical fact that a certain particle has a certain > position. That's highly a posteriori and has a contingent PI. > > (More technicalities coming up.) > > I think what you may really be talking about, though, is the > conditional from initial state of the wave function plus Schrodinger > equation (as a physical law) to final state of the particle. The > status of this depends on just which interpretation of QM we are > using, as the claim the Schordinger equation is taken to make about > physical reality will vary with the interpretation. On the Everett > interpretation, I suppose the conditional will be a priori, as the > equation is the only dynamics; and indeed, given knowledge of an > initial state and of the truth of the interpretation, one can know the > (superposed) final state just fine. Something similar arguably goes > for the Bohm interpretation. As long as we are given full enough > information about wave plus particles in the antecedent, then we can > come to know the determinate final state no problem. In a "collapse" > interpretation, on the other hand, the conditional will never be a > priori, as such interpretations have a further nondeterministic part > of the dynamics, so we'll at best be able to figure out a probability > distribution. Anyway, it seems to me that in each of these cases, a > priority and necessity of PI will line up. > > --Dave. > "What our grammarian does is simple enough. He frames his formal reconstruction of K along the grammatically simplest lines he can, compatibly with inclusion of H, plausibility of the predicted inclusion of I, plausibility of the hypothesis of inclusion of J, and plausibility, further, of the exclusion of all sequences which ever actually do bring bizarreness reactions." -- W.V.O. Quine ---------------------- Erik J Larson erikl@U.Arizona.EDU From owner-modality@LISTSERV.ARIZONA.EDU Fri Feb 19 06:21:34 1999 Date: Fri, 19 Feb 1999 06:20:07 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Next week To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO My visa still isn't through, and although there are indications that it shouldn't be too long now, I won't be there by next Tuesday. This means I have to rearrange things a little. Week 5 was originally supposed to be on the "tyranny of the subjunctive", on relationships between the 2-D framework and indicative and subjunctive conditionals, and on implications for the analysis of necessity. But that isn't really written up (except very briefly in the Princeton outline); I'd been planning to mostly talk about it in class. So as things stand, we will skip straight to the following topic (the original week 6) instead. This topic involves applying the 2-D framework to epistemic and modal arguments against materialism. The readings are TCM, Chapter 4 -- mostly pp. 131-149. N&N, pp. 144-55. Mind and Modality, Lecture 1 (esp. sections 5-7). The Mind and Modality notes (on the web) have my currently preferred formulation of the 2-D argument against materialism. It's a bit crisper and tighter than the book version, though it comes to much the same thing in the end. The book version has more discursive detail, so you'll probably want to look at the two of them together (the real core of the book argument is pp. 131-36, but pp. 147-49 also has relevant material). It might also be interesting to look at Kripke's anti-materialist argument in N&N, which is similar in spirit, and compare and contrast. (N.B. The Princeton notes have some contrastive analysis that goes a bit beyond what is in the book.) Anyway, consider these the readings for next week's discussion. I'll look forward to seeing your thoughts on them early next week. As promised, there's also an assignment. I'd like you to take two terms that Kripke discusses, and translate his discussion into the 2-D framework. One is "yard", as discussed on p. 76; the other is "cat", as discussed on p. 122 and pp. 125-6. I'd like you to take every significant part of Kripke's discussion here and translate it into the 2-D framework. Characterize roughly what the PI and the SI of these terms look like, at least given Kripke's intuitions. In places where Kripke considers a hypothetical scenario, you should roughly specify the world, say whether it is being considered as actual or as counterfactual, note what the referent of the relevant term or statement is in that world (considered the relevant way, and according to Kripke's stated intuitions), and say what the upshot is the for relevant PI or SI. Try to translate Kripke's more general points in these passages into the framework too, if you can. If you're so inclined, you can say whether you agree or disagree with Kripke's specific and general analyses and why, though that isn't compulsory. This shouldn't be much work (the passages are pretty short). This will be due by next Tuesday at noon, Arizona time. Don't be late. E-mail it directly to me (not to the mailing list). It should be your own work. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Wed Feb 24 01:33:12 1999 Date: Wed, 24 Feb 1999 01:32:52 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: assignments To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO I've been over the assignments and made comments on them. I'll send them all back individually. But here are some general comments. First I'll give something like the answers I would have given. -------------------------------------------- "yard": PI: picks out the distance from Henry's nose to finger in a given world. SI: picks out this specific length (the actual Henry length) in all worlds. Kripke's discussion: The PI of 'yard' picks out the Henry length in a given world. Even so, it's not necessary that the Henry length is a yard. To see this, take a world where Henry's arm was foreshortened by an accident. Considering that world as counterfactual, it's a world where Henry's arm is less than a yard. I.e., "the Henry length is a yard" is false in that world considered as counterfactual. I.e., the SI of "the Henry length is a yard" is false in that world. So "the Henry length is a yard" is not necessary. This isn't because the PI of "yard" is a cluster. The point works even when the PI is a non-cluster (as the PI used here shows). Rather it arises because the SI of "yard" is different from the PI, picking out the actual length at all worlds. (Something else one can say, though Kripke doesn't say it explicitly: It's a priori that the Henry length is a yard. If we take the foreshortened world and consider it as actual, "the Henry length is a yard" comes out true. I.e. the PI of "the Henry length is a yard" is true in that world. Similarly for any world. So "the Henry length is a yard" is a priori. This reflects the fact that at all worlds, the PI of "yard" picks out the Henry length in that world.) ---------------------------------------- "cats": PI: picks out "that kind of thing", i.e. the things I'm acquainted with as 'cats' or some such, at a given centered world. SI: picks out a specific biological kind (DNA, evolution history, etc?) at all worlds. p. 122: "Cats are animals" is not a priori. To see this, take a world where the catlike things are demons, and consider it as actual. Considered this way, we'll say that cats are demons and are not animals. I.e., "cats are animals" is false in this world considered as actual. I.e., the PI of "cats are animals" is false in this world. So "cats are animals" is not a priori This reflects the fact that the PI of "cats" doesn't pick out animals in all worlds. In fact it doesn't pick out any qualitative dictionary definition (for any definition D, "cats are D" isn't a priori), but rather picks out "that sort of thing". p. 126: Given that cats are in fact animals, then "cats are animals" is necessary. To see this, take the world where catlike things are demons, and consider it as counterfactual. Considered this way, it's a world where the demons aren't cats. So "cats are demons" is false in that world considered as counterfactual. Similarly for any world in which catlike things aren't animals. Considering such a world as counterfactual, we'll always say the nonanimals aren't cats. So "cats are animals" will be true in all such worlds considered as counterfactual. So the SI of "cats are animals" is true in all worlds. So "cats are animals" is necessary. The SI of "cat" picks out animals in all worlds. (Kripke also reiterates the PI point from p. 122: if we *discovered* that catlike things are demons, we'd say cats aren't animals. I.e. considering the catlike demon world as actual, "cats are animals" comes out false, i.e. "cats are animals" isn't a priori, and the PI of "cat" can pick out nonanimals. But here, we are considering the world as counterfactual, in order to determine what is necessary.) Similarly even for a world in which catlike things are animals with reptile internal structure. Considering such a world as counterfactual, we'll say it's a world where the catlike reptiles are fool's cats, not cats. So the SI of "cats" doesn't pick out the reptiles in that world. So the SI of "cats" picks out animals with a specific internal structure across worlds, and so it is necessary that cats have that specific internal structure. ------------------------------------------------- Of course I didn't expect precise replicas of this, but I was looking for the central points here to be made. Mostly people did a reasonably good job, showing at least a decent working knowledge of the framework. Often people could have been a bit more complete and explicit in the discussion, though. And people should work on being as precise and rigorous with the language as they can. A few points that came up repeatedly: (1) Quite a few people didn't distinguish the epistemic discussion on p. 122 from the modal discussion on p. 126 as carefully as they could have. Often people concentrated on the modal point (re rigid designation and the SI of "cat"), without going into Kripke's argument on the epistemic point (re considering the demon world as actual, and consequences for a priority and for the PI of "cat"). It's important to distinguish the epistemic from the modal point, and recognize that what's going on on p. 122 is really quite different from what's going on on p. 126. (2) Most people didn't address the point about clusters, and the reptile scenario. That's forgiveable, as they were minor points even in context. (3) Often people didn't quite follow my instructions re the consideration of hypothetical scenarios. In particular, it's important to note, when a scenario is being considered, whether it is being considered as actual or as counterfactual. Then one can note what results one gets when the scenario is considered this way, and what follows regarding the PI/SI of the concept, and for apriority/necessity. (In general, scenarios considered as actual have consequences for PI and for a priority, on the epistemic side; scenarios considered as counterfactual have consequences for SI and for necessity, on the modal side.) Being careful about this makes it a lot easier to disentangle the issues in (1) above. Note that Kripke's arguments almost always move from the specific to the general. He considers specific scenarios (as actual or as counterfactual), makes a considered judgment about how to describe them, and draws consequences about the way our terms refer across worlds (in effect about PIs and SIs), and draws conclusions about apriority and necessity. At least, that's the way the argument is presented: it's consideration of scenarios that establishes or supports the conclusions re a priority, necessity, and the like. That's a pretty common order of precedence in this area: one proceeds from a judgment about a scenario or scenarios (considered as actual or as counterfactual) to conclusions about PI/SI, apriority, necessity, etc, rather than vice versa. (4) A common looseness of language was to move too easily between intensions and extensions. E.g. people would say that the PI is the Henry length, or that the SI is the specific length, or that the SI picks out the actual PI in all worlds, and so on. Remember, a PI is a function, not a length. So one can say things like: the PI *picks out* a length in a given world; or, the *value* of the PI at a given world is a length. But it's best not to say that the PI itself is a length. For the phrases above, one could say e.g. that the PI picks out the Henry length in a given world, or that it picks out the actual Henry length in the actual world. Or one could just talk about the actual referent, instead of the PI, saying that the actual referent is the actual Henry length. For the second phrase, one could say that the SI picks out the specific length in all worlds. For the third, one could say that the SI picks out the actual referent in all worlds, or that it picks out the actual value of the PI at all worlds, or just that it picks out the actual Henry length at all worlds. There are various ways to do things, but it's important for many purposes not to mix up intensions and extensions too easily. The central thing is that an intension is a function from worlds to extensions. (Where extensions are things like objects, people, species, lengths, properties, etc.) The intension is not an extension, but it picks out an extension at any given world. I don't think anyone is really confused about this, but it's good to keep the language straight. (5) Another point about the language concerns just how we talk about evaluating statements at worlds. Given the ambiguity between PI and SI evaluation (i.e. between considering as actual and as counterfactual), one has to be careful about this. It's easy to be unclear by not specifying the sort of evaluation, or in other ways. Here are the sorts of locutions that are reasonable, where W is a world, and "A is B" is a statement. For PI evaluation (or considering as actual), the following locutions are probably best: In W considered as actual, "A is B" is true. The primary intension of "A is B" is true in W. The two above are probably the most precise, but one can also equivalently say the following without too much inaccuracy: "A is B" is true in W considered as actual. Considering W as actual, it's a scenario where A is B. Considering W as an epistemic possibility, it's an epistemic possibility in which A is B. If W turns out to be actual, it will turn out that A is B. If W is actual, A is B. Exactly the same goes if you substitute "counterfactual" for "actual" and "SI" for "PI" in the above, except for the last two, which don't have direct analogs. (Maybe a subjunctive such as "If W were the case, A would be B"; more on subjunctives and indicatives later.) Something similar goes for evaluating terms rather than sentences. Here you can say things like: In W considered as actual, "A" refers to X. The primary intension of "A" refers to X in W. If W is actual, "A" refers to X. [N.B. "Picks out" is as good as "refers to" here.] Things one *shouldn't* say: The PI of "A is B" is true in W considered as actual. (Redundant!) When W is considered as actual, the PI of "A" picks out X. (Redundant!) If W is actual, the PI of "A" picks out X (Redundant!) If W is actual, X is the PI of "A". (Intension/extension mixup.) The PI of "A" is actually X. (Intension/extension mixup.) One also has to be careful with things like: In W, "A is B" is true. In W, A is B. When we do this without a PI/SI indicator, or without an "actual" or "counterfactual", the usual practice is that we are considering W as counterfactual, so it is SI evaluation that's relevant. At least that is the standard practice in Kripke and in most contemporary philosophy. But in the context of the 2-D framework, it can be a little confusing and potentially ambiguous, so it's best to give some sort of marker where possible. ---- Anyway, as I said, people did pretty well. More precision and such (very important in this area!) should come with practice and experience. I'll probably set another assignment like this fairly soon, to give more practice at the framework. Maybe including an example or two not in Kripke, for you to give your verdicts on the PI and SI, etc. --Dave. P.S. Now that the assignment is out of the way, I'm looking forward to everyone's comments on the week 5 issues as soon as possible. From owner-modality@LISTSERV.ARIZONA.EDU Fri Mar 12 14:58:13 1999 Date: Fri, 12 Mar 1999 15:54:59 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Rachael J Parkinson Subject: Re: your mail To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Like Angela, I am confused about what the secondary intension points to in the case of the forty-second president of the United States. To elaborate on a point that Brad made about Deep Throat in our last meeting, I can think of a few cases where rigid designators don't seem to work the way Kripke intends. For example, Mark Twain; it would seem, given in our actual world that Mark Twain is Samuel Clemmons, Mark Twain would be Samuel Clemmons in all counterfactual worlds (I guess Samuel Clemmons isn't really a secondary intension?). But I can easily imagine a world where Mark Twain was Joe Schmoe. Likewise, I can imagine a world where Marilyn Monroe was not Norma Jean and Aristotle was a woman. I guess this is possible because I am assuming that the primary intension of those names apply to Mark Twain,the author; Marilyn Monroe, the actress; and Aristotle, the philosopher. But Kripke is intent on showing that names are not descriptive. The name, Marilyn Monroe, refers to, roughly, that woman that we call Marilyn Monroe. I think that Kripke can make a good case in respect to Aristotle (because Aristotle refers to the man and not to the Greek philosopher etc. it makes sense to say that Aristotle could have been a blacksmith.) I think that it does not make as much sense, however, to say that Deep Throat could have been a butcher, Mark Twain could have been a baker, etc. I guess I think that for these sorts of names, something like a descriptive theory is needed. I would appreciate if someone would expand on how primary and secondary intensions work in respect to definite descriptions and proper names. Thanks- Rachael From owner-modality@LISTSERV.ARIZONA.EDU Fri Mar 12 12:36:43 1999 Date: Fri, 12 Mar 1999 13:33:09 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Angela J Burnette To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO In our last meeting, Dave claimed that definite descriptions like 'the current president of the United States' would not refer to Bill Clinton in a poss. world where he was not the current prez even when that world was considered as counterfactual, is this because definite descriptions aren't rigid designators? Also, I'm confused (in these cases) as to what the primary and secondary intensions of such descriptions are like...because it seems as though, given that the referent of a description like 'the current president' doesn't pick out Bill Clinton in all poss. worlds *when considered as counterfactual*, that there...is no? secondary intension, or if there is, it's not serving the same...function? as the SI of a rigid designator (because it doesn't seem to fix the reference across worlds given that the actual world is the way it is). A little clarification here would be helpful... Also, regarding the concept of a priori, or rather the definition that ' x is knowable a priori if x can be known prior to experience,' I don't see why problems like the one presented by Godel's theorem aren't more troubling...if we have to idealize away from our own cognitive limitations to come up with an appropriate idea of 'in principle a priori knowability' such that we are conceiving of a being that is not subject to the sorts of mathematical counterexamples like Godel's theorem, then it seems as if our concept of a priori knowable breaks down...i.e., how are we to conceive of the distinction between what is knowable prior to experience and what is not for an idealized being like the one needed to avoid the Godel counterexample? It's not clear to me what such a being could or couldn't know without actually having to look...if a priori were defined strictly in terms of the analyticity of statements, then there seems to be less of a problem keeping the distinction between a priori and a posteriori straight when idealizing so far from our own case...but this doesn't seem to be the way Dave wants, centrally, to define the a priori...so is it the case that "knowable prior to experience" is more akin, or closer in kind to an idea like "analyticity" than I think, or what? angela From owner-modality@LISTSERV.ARIZONA.EDU Sun Mar 14 15:29:45 1999 Date: Sun, 14 Mar 1999 15:29:28 -0800 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Primary and secondary intensions To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Angela, Rachael, and Erik L. raise a few lingering questions about the two-dimensional framework, primary and secondary intensions. Angela writes about the secondary intensions of descriptions: >In our last meeting, Dave claimed that definite descriptions like 'the >current president of the United States' would not refer to Bill Clinton in >a poss. world where he was not the current prez even when that world was >considered as counterfactual, is this because definite descriptions aren't >rigid designators? Yes, this more or less comes down to the fact that descriptions aren't rigid designators. Let's take "the 42nd president of the United States" as our description to avoid difficulties due to indexicals and centering. And let's consider a world in which Paul Tsongas is the 42nd president of the United States -- and let's not consider it as actual (as an epistemic possibility), but as counterfactual (as a way the world might have been). In that counterfactual world, what does "Bill Clinton" pick out? I.e., who is Bill Clinton in that counterfactual world. Certainly not Paul Tsongas! It's Clinton, who presumably is somewhere back in Arkansas in this world. In that counterfactual world, what does "the 42nd president of the United States" pick out? I.e., who is the 42nd president of the United States in that world? Certainly not Clinton! It's Paul Tsongas, who is busy being president in this world (at least until he has a heart attack). All this is borne out by the fact that we describe this as a counterfactual world in which Tsongas (not Clinton) is the 42nd president of the US. If "the 42nd president" picked out Clinton in this world considered as counterfactual, it would come out as a counterfactual world in which Clinton is the 42nd president, which seems wrong. This brings out Kripke's central distinctions between names and descriptions, which is that names are rigid designators and descriptions are not. That is, names pick out the same individual across all counterfactual worlds, but descriptions need not -- they pick out whatever individual satisfies the description, even in counterfactual worlds. For example, "Clinton" picks out the same person in all worlds considered as counterfactual; i.e. the SI of "Clinton" picks out the same person in all worlds; i.e. "Clinton" is a rigid designator. Whereas "the 42nd president" picks out whoever is the 42nd president in a given world considered as counterfactual; i.e. the SI of "the 42nd president" picks out different individuals in different worlds; i.e. "the 42nd president" is not a rigid designator. So the SIs of "Clinton" and "the 42nd president" are different. If they were the same, then "Clinton is the 42nd president" would be true in all counterfactual worlds, i.e. it would be necessary (in the Kripkean sense) that Clinton is the 42nd president, which seems wrong. The Tsongas world itself seems to bear witness to the fact that it is not necessary that Clinton is the 42nd president. We might say that SI of "Clinton" goes with the person, while the SI of the "42nd president" goes with the office. There is a temptation sometimes to think that the secondary intension of a term is just *defined* as the function that picks out the actual referent (e.g., Clinton for "the 42nd president") in all worlds. (Even some well-known philosophers have interpreted the framework that way). Of course one could define an intension that works that way, but that isn't the way secondary intensions (as I've defined them) work. One simply takes a world, considers it as counterfactual, and sees what the term picks out there. Kripke's contribution was to make the nontrivial observation that names pick out the same referent in all worlds consider this way, whereas descriptions don't. That is, he made the nontrivial observation that names but not descriptions are rigid designators. If we defined secondary intensions in the alternative way suggested above, then it would be trivially true that names, descriptions, and everything else are rigid designators, so we would lose both the nontriviality of Kripke's observation. Indeed, if we did things that way, we'd lose any way to distinguish between names and descriptions (such as "Hesperus" and "the evening star") at all. Their primary and secondary intensions would be the same, and both expressions would pick out the same things in all worlds. So we'd lose the ability to account for the fact that it is not necessary that Hesperus is the evening star, and so on. Or at least, we'd break the tie between secondary intensions and necessity, which would seem undesirable. >Also, I'm confused (in these cases) as to what the >primary and secondary intensions of such descriptions are >like...because it seems as though, given that the referent of a >description like 'the current president' doesn't pick out Bill Clinton in >all poss. worlds *when considered as counterfactual*, that there...is no? >secondary intension, or if there is, it's not serving the same...function? >as the SI of a rigid designator (because it doesn't seem to fix the >reference across worlds given that the actual world is the way it is). A >little clarification here would be helpful... The PI of "the 42nd president of the US" picks out the 42nd president in all worlds. The SI picks out the 42nd president in all worlds. The actual referent is Clinton. The PI of "Bill Clinton" picks out very roughly "the guy I've heard of as `Bill Clinton'" in all worlds. The SI picks out that very guy in all worlds. The actual referent is Bill Clinton. This all reflects the fact that the SI of a name picks out the actual referent in all worlds, whereas the SI of a description doesn't. The description still has an SI, as one can still consider a counterfactual world and ask what the description picks out there. It's just that for a description, it needn't pick out the actual referent (as the term isn't a rigid designator). Basically, the SI of a name depends very heavily on the way the actual world turns out (it is very much a posteriori). But the SI of a description depends much less heavily on the way the actual world turns out (sometimes it may be a priori, or sometimes it may be somewhat a posteriori, e.g. due to the presence of a name within the description). It's still the case that when we consider the Tsongas world as counterfactual, we're holding fixed that the actual world has turned out this way, with Clinton being such-and-such a guy, being the 42nd president, etc. Given that a posteriori information, this tells us that "Clinton" picks out that very guy across all worlds. But even this a posteriori information doesn't tell us that "the 42nd president" picks out Clinton across all worlds. Even knowing how the actual world turned out, the Tsongas world in question is described as a counterfactual world in which Tsongas is the 42nd president, not Clinton. [All this is complicated a little by the fact that descriptions arguably have *a* reading on which they are rigid designators -- e.g. where one reads "the 42nd president" as something like "the actual 42nd president". On this reading, one could consider the Tsongas world as a world in which the 42nd president isn't president (i.e., the actual 42nd president, Clinton, isn't president), and on that reading, "the 42nd president" would pick out Clinton in the Tsongas world, not Tsongas (and would pick out Clinton in all worlds). But I think this would be a somewhat unusual reading of the description. At most, this would suggest that descriptions are ambiguous between a nonrigid and a rigid reading. The difference between the SIs of rigid terms and nonrigid terms would stay intact.] Rachael writes: >Like Angela, I am confused about what the secondary intension points to >in the case of the forty-second president of the United States. To >elaborate on a point that Brad made about Deep Throat in our last meeting, >I can think of a few cases where rigid designators don't seem to work the >way Kripke intends. For example, Mark Twain; it would seem, given in >our actual world that Mark Twain is Samuel Clemmons, Mark Twain would be >Samuel Clemmons in all counterfactual worlds (I guess Samuel Clemmons >isn't really a secondary intension?). But I can easily imagine a world >where Mark Twain was Joe Schmoe. Likewise, I can imagine a world where >Marilyn Monroe was not Norma Jean and Aristotle was a woman. I guess this >is possible because I am assuming that the primary intension of those >names apply to Mark Twain,the author; Marilyn Monroe, the actress; and >Aristotle, the philosopher. Hmm, interesting. I think one can certainly imagine a world *considered as actual* in which Marilyn Monroe wasn't Norma Jean, etc (just say we discovered that she was really born "Marilyn" and that the "Norma Jean" business was a myth, with the Norma Jean she pretended to be still living happily in Florida as a grandmother). But can we imagine a world considered as counterfactual in which Marilyn wasn't Norma Jean? Take a counterfactual world where the real Norma Jean (who in the actual world became Marilyn) never went into show business, and that some other blonde actress (Jayne Mansfield, say) took the name "Marilyn Monroe", had some plastic surgery, ended up looking and sounding and living just like the actual Marilyn. Is this a counterfactual world in which Marilyn was Jayne Mansfield? Or is it a counterfactual world in which Marilyn never went into show business, and in which someone else looked and lived like her? Kripke's intuition is the second, in which case the SI of "Marilyn Monroe" picks out Norma Jean in this world, not Jayne Mansfield. Do you have the first intuition instead? If so, then presumably the SI of "Marilyn Monroe" picks out Jayne Manfield in this world, and picks out roughly whoever plays the relevant actress role in a given world. My own intuitions tend to side with Kripke here. It does seem that we can consider counterfactual worlds in which Marilyn Monroe never went into acting, and lived a happy family life in Florida. If so, there are counterfactual worlds where the SI of "Marilyn" picks out Norma Jean, irrespective of whether she's an actress. And it is not obvious to me that we can easily consider a world where Jayne Mansfield really was Marilyn, as opposed to looking and living like her. If so, then the SI of "Marilyn" goes with the person, not the role. But maybe you have different intuitions here. If you're right, there is at least one reading of some names in which they are nonrigid: "Marilyn Monroe" will pick out whoever plays the actress role in a counterfactual world, irrespective of whether they are the same person as the actual Marilyn. If this is right, then the distinction between names (at least these names) and descriptions would be broken down further. Anyway, I'm not certain whether you're meaning to endorse these claims, or just the weaker primary intension claim (which applies only to worlds considered as actual). The weaker claim I think is very reasonable; the stronger claim would be unorthodox and many would disagree, but that's not to say that it's untenable. On this reading, I presume you'll have to disagree with Kripke, and argue that it is necessary that Hesperus is the evening star, in all counterfactual worlds, and so on? >But Kripke is intent on showing that names are not descriptive. The name, >Marilyn Monroe, refers to, roughly, that woman that we call Marilyn >Monroe. I think that Kripke can make a good case in respect to Aristotle >(because Aristotle refers to the man and not to the Greek philosopher etc. >it makes sense to say that Aristotle could have been a blacksmith.) I >think that it does not make as much sense, however, to say that Deep >Throat could have been a butcher, Mark Twain could have been a baker, etc. >I guess I think that for these sorts of names, something like a >descriptive theory is needed. > >I would appreciate if someone would expand on how primary and secondary >intensions work in respect to definite descriptions and proper names. Hmm, OK, so maybe you are endorsing the strong claim here. I think Kripke would argue that Deep Throat could have been a butcher. (One can imagine Nixon saying "if only that damned Deep Throat had never gone into public affairs and had become a butcher instead".) And I think he would argue that "Deep Throat" can't pick out other individuals in other counterfactual worlds. (Of course it can pick out other individuals in worlds considered as actual, as the primary intension is pretty clearly descriptive.) On your view, it looks like there will be two classes of names, some of which are rigid designators and some of which are not. The rigid ones pick out the same individuals across counterfactual worlds (they have a constant SI), whereas the nonrigid ones don't (they have a descriptive SI). I guess the nonrigid ones will be the ones that "look" descriptive in some sense, or which have strongly desciptive connotations -- Deep Throat, Jack the Ripper, Hesperus, etc, etc? Such names would presumably have very similar PI and SI, captured by the description in both cases. The rigid ones, by contrast (e.g. Rachael Parkinson, Joe Schmoe, etc) will have some sort of subtle PI (maybe captured causally or metalinguistically) and will have an SI that picks out the same person in all worlds. As I say, that would be an unorthodox but interesting position. (The orthodox position is that all names are rigid, so that their SI picks out the sme individual in all worlds.) It might be tricky to draw the line between the two sorts of names. I guess an alternative would be to say that many or all names are ambiguous between a rigid and a nonrigid reading, so that "Deep Throat" etc can be read both ways (on one reading, Deep Throat could have been a butcher, on the other reading, not). Maybe one could even make the case for all names, so that e.g. "Rachael Parkinson" has an alternative descriptive reading on which Rachael had to be called `Rachael' or some such. This would break down the barrier between names and descriptions still further, as we'd now have ambiguity between rigid and nonrigid readings in both cases. All that applies to secondary intensions, of course. I think the primary intensions of names etc will be fairly unaffected by what we say here about rigidity. Erik writes about causal primary intensions: >I have a brief question regarding the admissability of causal primary >intensions. This has come up at least twice; once in Tim's response to >Josh's BIV scenario, and then again in our first class meeting. So, the >problem is, I don't think primary intensions can ever be appropriately >thought of as "the cause of my sensation that" or "that which brought >about my experience of" or some such, because in any of these cases the >primary intension can vary wildly, depending on what the cause happens to >be. So some brain in a vat may have a PI of water that appeals to >electrodes and stimulation or whatever, and on earth this of course won't >make any sense. So "that which causes my x experience" makes a primary >intension the same as whatever the cause happens to be , and not the >essential notion or properties of the thing itself, apriori. So I think >the primary intension has to be necessarily connected to the essential >quality of the thing (at least as it is known apriori or conceptually). >What we mean by the concept of water, then, is not what causes the water >experiece (which may have nothing to do with the essence of water itself). Hmm, I'm not sure I've entirely grasped the problem here. It seems to me that the BIV case as you've described it is compatible with the PI of water being "that which causes my x experience". If that's the PI for the BIV, then presumably "water" will pick out whatever causes x experiences for the BIV, which will presumably be some assortment of chemicals and electrodes, etc, which is just what you say. Here there will be no difference between the PI of "water" in the BIV and in you and me. In both cases it might be "what causes my x experiences". Of course the *referent* will be different between the cases: for you and me, the PI will pick out H2O, whereas for the BIV, it will pick out electrodes and such. But the PI and the referent are different things. One shouldn't strictly speaking say that the PI for the BIV involves electrodes, etc. It's important to remember that the PI is a function from worlds to extensions, and the referent is just the extension at the actual world. For the BIV, the PI might be "what causes my x experiences". One can say that the PI *picks out* electrodes and such in its world, or one can say that the *referent* involves electrodes, but these are different claims. It may be that one wants to reject simple causal PIs for other reasons. E.g., one might reasonably argue that the term "water" for a BIV *doesn't* pick out the electrodes, even though they cause x experiences in the BIV. If that's right, then arguably the PI for "water" has a descriptive component over and above the pure causal component. But that's a subtle matter. Let me know if I've missed your point here. --Dave.