From owner-modality@LISTSERV.ARIZONA.EDU Sat Apr 3 12:55:28 1999 Date: Sat, 3 Apr 1999 13:53:59 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Anthony T Lane Subject: dogs To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO In Dave's response to Angela re the capabilities of the "ideal conceiver", he suggests that, "...Arguably, I can conceive roughly of what it's like visually to be a dog, even though I'm not a dog." This seems to be accurate. Whereas we have rod and cone structures in our eyes, dogs only have rod structures (I think..). Given that we have good reasons for thinking that cones are responsible for our experience of color, we think that dogs see in black and white. It seems quite straightforward to imagine the visual experiences of a dog-- they are probably somewhat like the experience of seeing a black and white movie. It seems to me that it is somewhat more difficult to conceive of a dogs olfactory experiences. Avalanche rescue dogs are capable of quickly detecting the odor of people buried beneath considerable amounts of very hard snow (3 meters or more, sometimes). They are further able to distinguish the smell of the buried victim from the smell of the person handling them and other rescuers on the scene. I was once buried in a snowcave some six feet below the surface, and the rescue dog found me and had dug down to me in about 2 minutes. I don't have the faintest idea what it would be like to be able to make the kinds of olfactory discriminations that dogs are capable of making. It does not seem to be me to be particularly likely that even an ideal conceiver would be able to do this. I sems fairly straightforward to imagine a what a dog's visula experiences are like because we assume that their visual experiences are somehow less rich than our own. But, given that dogs have olfactory capabilities far superior to ours, itjust doesn't seem plausible to suggest that any non-canine being, however idealized and with whatever instrumentation you please, could conceive of what a dogs olfactory experiences are like. I'm not sure if all of this amounts to anything (except praise for the virtues of dogs). Anthony From owner-modality@LISTSERV.ARIZONA.EDU Wed Apr 14 18:51:55 1999 Date: Wed, 14 Apr 1999 18:48:45 -0700 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: inscrutability, inconceivability, etc. To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Josh writes: >Turning to yesterday's discussion, I'm unclear exactly what the >difference between and inscrutable truth and an open inconceivability >is. Both, I take it, are cases that are negatively conceivable but >not positively conceivable. Both are possible. Perhaps open >inconceivabilities are things that are to weird to have a concept of, >but I don't think we want to use "weirdness" as a formal distinction >(though I admit it is an informal distinction). So what did I miss? Both open inconceivabilities and inscrutabilities fall into (or are closely related to) the "twilight zone" of statements that are negatively conceivable but not positively conceivable, but they are different. Informally, the difference is that open inconceivabilities (e.g., "there are inconceivable features of the world"?) are verified as *false* in all complete positively conceivable scenarios, whereas inscrutabilities (e.g., some wondrous mathematical statement?) are *unsettled* in some complete positively conceivable scenarios. I'll work through this and some other relevant things in more detail. To start, here are some of the definitions. S is in the TWILIGHT ZONE if S is negatively conceivable but not positively conceivable. [N.B. "Ideal primary" is assumed throughout.] NEGPOS PRINCIPLE: If S is negatively conceivable, S is postively conceivable. [I.e., there is nothing in the twilight zone.] D is PC-COMPLETE if (a) D is positively conceivable, and (b) if D-and-F is positively conceivable, then D implies F. Note: This is a slight variation on the definition of PC-completeness given before. It's straightforwardly equivalent to the previous definition, but is hopefully a little more intuitive. Also, recall that "A implies B" should always be read as "`A -> B' is a priori". S is an INSCRUTABLE TRUTH if S is true and some PC-complete truth D does not imply S. SCRUTABILITY OF TRUTH: If D is a PC-complete truth and S is true, then D implies S. [I.e., there are no inscrutable truths.] Explanation: scrutability of truth says that all truths are implied by a complete qualitative description (i.e. a PC-complete description) of the actual world. Next we generalize this to all possible worlds: S is a GENERALIZED INSCRUTABLE if there exists some PC-complete D such that D-and-S is negatively conceivable, but D does not imply S. D is NC-COMPLETE if (a) D is negatively conceivable, and (b) if D-and-F is negatively conceivable, then D implies F. GENERALIZED SCRUTABILITY: If D is PC-complete, D is NC-complete. Note: I didn't define generalized inscrutables, as opposed to generalized scrutability, in the lectures or notes, but the definition here is the natural one. Basically we take the definition of an inscrutable truth and relax the requirement that D hold of the actual world, and that S be true there. Similarly, generalized inscrutability is the natural generalization of the inscrutability of truth. In effect, it says that for any possible world, a complete qualitative description of that world leaves nothing unsettled. It's easy to see that generalized scrutability as defined says precisely that there are no generalized inscrutabilities. (If this isn't obvious to you, working through it should be a worthwhile exercise.) S is an OPEN INCONCEIVABILITY if S is negatively conceivable, but for all PC-complete D, D implies not-S. NOINCONCEIVABILITY: No S is an open inconceivability. It's straightforward to see that if S is an open inconceivability, it cannot be a generalized inscrutable, and vice versa. If S is an open inconceivability, it is verified as false in every PC-complete scenario, so it can't be left unsettled in any PC-complete scenario, which is what inscrutability requires. More formally: S is an open inconceivability <-> for all PC-complete D, D implies not-S <-> there is no PC-complete D such that D-and S is negatively conceivable -> S is not a generalized inscrutable. It's nevertheless true that both open inconceivabilities and generalized inscrutables, if they exist, would provide counterexamples to NEGPOS and inhabitants of the twilight zone. To see this: (i) If S is a generalized inscrutable, then there is some PC-complete D such that D-and-S is negatively conceivable but D does not imply S. In this case D-and-S is negatively conceivable but not positively conceivable (if D-and-S were positively conceivable, D would not be PC-complete). So D-and-S is in the twilight zone. (ii) If S is an open inconceivability, then S is negatively conceivable. But S is not positively conceivable: If S were positively conceivable, then there would be some PC-complete D such that D implies S.** So S is in the twilight zone. [** This is a general principle about PC-completeness that is nontrivial but not hard to argue for. If S is positively conceivable, one can in effect construct D by conjoining as much as one can to S while still retaining positive conceivability. A maximal such conjunction will be PC-complete. The nontrivial part is making the case that there will always be a maximal conjunction here, as opposed to an ever-increasing series without a maximal element, but I think that case can plausibly be made. If the claim is doubted, we need to slightly modify the definition of open inconceivabity.] So, if NOINCONCEIVABILITY or GENERALIZED SCRUTABILITY is false, then NEGPOS is false. The reverse also holds, as I said in class. Here's the proof I didn't have time to give. Just say NEGPOS is false, and there is some S that is negatively conceivable but not positively conceivable. Is S ruled out by all PC-complete D? If yes, then S is an open inconceivability, so NOINCONCEIVABILITY is false. If no, then there is some PC-complete D such that D does not imply not-S. But D cannot imply S (if it did, then S would be positively conceivable). So S is a generalized inscrutable, and GENERALIZED SCRUTABILITY is false. All that is to say that NEGPOS is equivalent to NOINCONCEIVABILITY and GENERALIZED SCRUTABILITY. Or more informally, that the character of the twilight zone is exhausted by open inconceivabilities and generalized inscrutables. All this provides a closer analysis of modal rationalism. Pure modal rationalism says that S is positively conceivable iff it is negatively conceivable iff it is possible. Weak modal rationalism says just that if S is positively conceivable, it is possible (i.e. there are no strong necessities). Strong modal rationalism says that if S is negatively conceivable, it is possible (i.e. there is nothing in the twilight zone). PURE MODAL RATIONALISM: pos con <-> neg con <-> pos WEAK MODAL RATIONALISM: pos con -> poss STRONG MODAL RATIONALISM: neg con -> poss NEGPOS: neg con -> pos con It's not hard to see that pure modal rationalism is equivalent to the conjunction of weak modal rationalism and the negpos principle. Clearly, PMR entails WMR and NEGPOS. In the reverse direction, if WMR and NEGPOS hold, then neg con -> pos con -> poss. It's obvious that poss -> neg con (if something is possible, it can't be ruled out a priori), so the circle is closed: neg con <-> pos con <-> poss. So, pure modal rationalism is equivalent to the claim that there are no strong necessities and nothing in the twilight zone. This in turn is equivalent (via the previous discussion) that there are no strong necessities, no generalized inscrutables, and no open inconceivabilities. Hence these three problems exhaust the obstacles to pure modal rationalism. Personally, I am fairly confident about ruling out strong necessities (and thus about weak modal rationalism), e.g. by the arguments we've discussed in class. I am less sure about ruling out generalized inscrutables, but I'd like to think there are none, and that there are at least some decent arguments for thinking so. And I'm less sure in turn about open inconceivabilities, which seem to be harder to get a grip on (as we found in class). If we were able to rule out all three, though, the resulting pure modal rationalism would certainly give a very clean and elegant picture of the epistemology of modality, and of the shape of the modal universe. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Thu Apr 15 10:34:12 1999 Date: Thu, 15 Apr 1999 10:14:42 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Anthony T Lane Subject: Open inc's To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Dave specified that the only properties that we consider when talking about the possibility of open inconceivabilities are those that are non relational and nondispositional. I suppose that this rules out the case I'm about to make. At the moment, however, I'm not entirely clear how we can distinguish properties in this way. Consider two universes, U1 and U2. Both contain only two particles, asnd in each the distance between the two particles is diminishing such that, at a certain point, the two particles will pass very close to each other and then continue off for a very long voyage into infinity, assuming no interaction between them. In U1, however, there is a force of gravity such that the two particles will actually start orbiting around each other. The two particles never get very close together, however, as, in each world, one of the particles spontaneously decays with an infinitessimal release of energy (and it's the same particle in each worlf that decays, if one can make sense of such a notion). U1 and U2 have identical histories and, it seems, there is no possible way for any conceiver to tell the two worlds apart. I'm assuming that not even an ideal conceiver will be able to discern differences in uninstantiated laws, as our ideal conceiver would then be, I presume, quite inconceivable to us. Now I assume my example is ruled out given thaty the only difference in U1 and U2 is in terms of a dispositional property. Presumably, a red thing that exists in this univers but which will not, at any time in its history, interact with any other matter (it spontaneously coalesced out of a little patch of energy, suppose) is still red. It seems to me that in both cases, if it is meaninful to talk of properties at all, it is only insofar as a property has the potential for being discerned by some agent. But it is not clear to me why a property may then be ruled out simply because the event the event that would have instantiated that property never came to pass. From owner-modality@LISTSERV.ARIZONA.EDU Thu Apr 15 18:30:05 1999 Date: Thu, 15 Apr 1999 18:25:29 -0700 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: Open inc's To: MODALITY@LISTSERV.ARIZONA.EDU Status: RO Anthony writes: >Dave specified that the only properties that we consider when talking >about the possibility of open inconceivabilities are those that are non >relational and nondispositional. I suppose that this rules out the case >I'm about to make. At the moment, however, I'm not entirely clear how we >can distinguish properties in this way. Actually, I didn't mean to rule out the possibility that dispositional or relational properties could be inconceivable. I was just pointing to intrinsic (nondispositional, nonrelational) properties as a particularly fruitful class where inconceivable properties might be found. But that's not to say that they might not be found elsewhere. Anthony goes on to give the example of two universes with the same histories, but different uninstantiated laws, and suggests that the difference between these worlds might be inconceivable. The way I see things, both worlds are quite conceivable, and indeed Anthony did a nice job of describing them for us so we can conceive of each of them. What's true is that the difference between the worlds is not *perceivable* -- they will "look" the same to a being within those worlds, for example. But this just suggests that there is more to conceiving that perceiving. There are other examples of that: e.g. where one conceives of different theories compatible with the same empirical evidence (as e.g. in interpreting quantum mechanics), or where one conceives of unperceivable objects (e.g. a "peekaboo unicorn"). >It seems to me that in both cases, if it is meaninful to talk of >properties at all, it is only insofar as a property has the potential for >being discerned by some agent. But it is not clear to me why a property >may then be ruled out simply because the event the event that would have >instantiated that property never came to pass. Hmm, I think many would question the first claim. One could argue that in just this case, there's a real difference in properties of the worlds (a difference in laws) that would not be "discerned" by an agent in those worlds. I suppose one could argue that it is "discernible" in that there are some counterfactual states of affairs where the difference could show up. But in some cases (e.g. the unicorn or the theories) even this might not hold. Nevertheless, in all these cases the properties can be *conceived* by some agent. I think that corresponds to the thesis some people were defending in class -- that it's not meaningful to talk about properties that can't be conceived by any agent. I'd be interested to see any arguments for that thesis. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Fri Apr 16 14:35:15 1999 Date: Fri, 16 Apr 1999 14:33:30 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Timothy J Bayne Subject: An inscrutable truth? Probably not. To: MODALITY@LISTSERV.ARIZONA.EDU Status: R Here's an argument for an inscrutable truth. It's open to a number of objections - one of which I think is good. (1) There are truths about bat experiences. There are tue statements of the form "Being a bat is like. . .". Call these truths BE-truths. (2) Bats have bat experiences, but they cannot believe any truths about bat experiences, because the lack the requisite concepts. (3) You can only believe a truth about bat experiences if you yourself have battish (batty?) experiences. (4) Nothing but a bat can have battish experiences. Therefore, (5) nothing can know BE truths. In short: The only things that could are bats, but they can't, cos they don't have justified beliefs at all (or, at least, they don't have justified beliefs about their experiences.) Some comments. You might reject (1). I take it Dave (and Nagel) won't. They hold that there are intrinsic aspects of experience, that can only be known by having the experience. You might reject (2). You might say that having an experience is not just necessary for having knowledge about that experience, but that it is also *sufficient*. We're in deep water here, but I think most would grant that you need more conceptual sophistication to know about your own experience than just the experience itself. You might reject (4). Why can't there be superbats? Superbats are just like bats - they have the same experiences as bats - but they have the conceptual sophistication to know about their own experiences. This is probably the weakest link in the argument. In fact, I guess I think that this objection is right. Tim 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 Apr 16 14:49:46 1999 Date: Fri, 16 Apr 1999 14:47:16 -0700 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: An inscrutable truth? Probably not. To: MODALITY@LISTSERV.ARIZONA.EDU Status: R Tim's bat example is interesting. I should note that this isn't really being put forward as an instance of an inscrutable truth, but rather as an instance of an unknowable (perhaps even unbelievable) truth. It's important to note that these are not the same thing (though arguably inscrutable truths, if they exist, will be unknowable). Of the categories we've been talking about, the truths in question are probably most relevant to open inconceivability than anything else. One might use Tim's argument to say: (i) to conceive of these truths, one needs to have battish experiences; (ii) only bats can have battish experiences; (iii) bats don't have the relevant concepts; so (iv) no possible being can conceive of these truths. Tim's superbat objection would strike at premise (ii) here, and indeed that seems reasonable. One might try getting around this by talking about "dumb bat experiences", the sort that can only be had by an unintelligent bat. Superbats could never have these experiences, I presume! But maybe they could form the concepts some other way. Another relevant route is to question (i) (which corresponds to Tim's (3)). Arguably one can conceive of some experiences without having had those exact experiences, but rather by having experiences that are relevantly related to them. Hume's "missing shade of blue" is an example. And maybe e.g. we could use our own experiences to form concepts of certain "proto-experiences" that compose them, and then recombine the protoexperiential concepts to form concepts of experiences that we've never had. Maybe that sort of thing could lead a being without bat experiences to form concepts of bat experiences (and even of "dumb bat experiences"). It's not obvious that this is possible, but it's not obvious that it's impossible, either. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Sun Apr 18 22:59:45 1999 x-sender: agillies@pop.u.arizona.edu Date: Sun, 18 Apr 1999 23:13:49 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Anthony S Gillies Subject: open inconceivability To: MODALITY@LISTSERV.ARIZONA.EDU Status: R All, some thoughts on open inconceivabilities--OI's--(and why I can't conceive of any). The general strategy is to show that, on the assumption that S is negatively conceivable we can show that there is a PC-complete D such that D --> S. After doing that, we can look at problematic points of the argument. So, let S be any (primarily ideally) negatively conceivable statement. Let R be an ideal rational agent. S is thus negatively conceivable by R in the limit. That is, there is some ordinal stage of R's reasoning sigma, such that: R sees no contradiction in S at sigma and R sees no contradiction in S at all ordinals > sigma. Let D be the generalized conjunction of a maximal consistent subsets of the statements that R has marked as "conceivable" such that S is a conjunct of D. D is a complete description of a world, we just need to see that it is a *PC*-description. To see this, note that if R tags a statement as noncontradictory in the limit, then the statement is noncontradictory. That is, it has a satisfying structure. So D has a satisfying structure. But surely D must then be PC-complete. D-->S, and so we're done. Two things worth pointing out: (1) Most of the work above is being done by tacit appeal to the equivalence I see between NI-in-the-limit to PC: if we are serious about full rationality in the limit (to the point of fixed points of ordinal updates), then this seems to me to be plausible. What else could we mean by NC in this case other than something like satisfiable? (2) We need to explain away claim to OI status that statements like "There are inconceivable features of the world" might seem to have. I think that's pretty easy. First, they equivocate on "inconceivable"--sure there might be features which are inconceivable to us, but that's no surprise. And that doesn't establish that there are OI's. What's relevant is whether there are statements which are in the limit NC but not PC (and we're nowhere near the limit). But, if a statement is in-the-limit NC iff it is PC, then this is not possible. Thony "Curious green ideas sleep furiously." From owner-modality@LISTSERV.ARIZONA.EDU Mon Apr 19 00:11:21 1999 Date: Sun, 18 Apr 1999 23:59:53 -0700 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: open inconceivability To: MODALITY@LISTSERV.ARIZONA.EDU Status: R Thony puts forward a general argument against the existence of open inconceivabilities. In fact he gives much more: an argument that anything that is ideally negatively conceivable is also positively conceivable. If successful, such an argument would establish the negpos principle, and rule out both open inconceivabilities and inscrutable truths. >So, let S be any (primarily ideally) negatively conceivable statement. >Let R be an ideal rational agent. S is thus negatively conceivable by R >in the limit. That is, there is some ordinal stage of R's reasoning >sigma, such that: R sees no contradiction in S at sigma and R sees no >contradiction in S at all ordinals > sigma. Let D be the generalized >conjunction of a maximal consistent subsets of the statements that R has >marked as "conceivable" such that S is a conjunct of D. D is a complete >description of a world, we just need to see that it is a >*PC*-description. To see this, note that if R tags a statement as >noncontradictory in the limit, then the statement is noncontradictory. >That is, it has a satisfying structure. So D has a satisfying structure. >But surely D must then be PC-complete. D-->S, and so we're done. Hmm. I'm not certain what you mean by "consistent" above. I think you mean something like "not contradictory". If so, then there will plausibly be such a D, and the D in question will be NC-complete. The crucial question, as you say, is whether D is also PC-complete. If D is PC, it will be PC-complete. But what reason is there to believe that D is PC? Your argument turns on a "satisfying structure". I need to know just what a "satisfying structure" is. I see about three possibilities: (1) If a satisfying structure is merely a model that gives an interpretation for the terms such that D comes out true in that model, then there will certainly be a satisfying structure, but it's not at all obvious why the existence of such a structure implies positive conceivability, given that e.g. such a model may assign an interpretation of the terms in D quite different from our ordinary interpretation. (2) If a satisfying structure is a model such that the interpretation of D in the model is determined by the ordinary semantics of the terms in D, then it's less obvious that such a model will exist. If one is loose enough about what counts as preserving the ordinary interpretation of the terms, perhaps there will be a model, but in this case the inference from the existence of a model to the positive conceivability of D is again far from clear. (3) If a satisfying structure is just a possible world in which D is true, or even a positively conceivably situation in which D is true, then it's not obvious why such a satisfying structure should exist. Or at least, that's just what's at issue. To clarify, you might show in detail how the argument would run for (a) D = the conjunction of all the physical truths about the world with some purported inscrutable truth (e.g. a statement about tallness, on the epistemic theory; or a mathematical statement that is true but not a priori), and (b) a purported open inconceivability such as "there are inconceivable features of the world". As it stands your argument seems to rule these things out perhaps just a little too easily, but maybe the details will help. >Two things worth pointing out: >(1) Most of the work above is being done by tacit appeal to the >equivalence I see between NI-in-the-limit to PC: if we are serious about >full rationality in the limit (to the point of fixed points of ordinal >updates), then this seems to me to be plausible. What else could we mean >by NC in this case other than something like satisfiable? Well, NC in the limit means absence of any contradiction detectable through reasoning. That definition doesn't make any obvious claims about satisfiability. Certain weak forms of satisfiability may follow, but these don't obviously guarantee positive conceivability. And for strong forms of satisfiability that guarantee positive conceivability, it's not obvious why satisfiability in these senses follows from NC-completeness. At least, such a link is just what's at issue. But any argument for the link would be gratefully received! --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Mon Apr 19 02:38:34 1999 X-Priority: 3 X-MSMail-Priority: Normal X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3 Date: Mon, 19 Apr 1999 02:36:15 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Brad Thompson Subject: A problem for the notion of a single ideal conceiver To: MODALITY@LISTSERV.ARIZONA.EDU Status: R The case Tim mentioned about "dumb bat experiences" makes me think of a related possible area of open inconceivabilties. Actually, I want to present a case in which there are truths which can be conceived of by some being but not by an ideal conceiver. I'm not sure whether such truths should be called open inconceivabilities. Such a case might require us to either 1) revise our notion of conceivability such that it doesn't depend on a single ideal conceiver but rather on being conceivable by *some* being, or 2) reject modal rationalism. I actually prefer 1. (I can't recall whether the notion of a single ideal conceiver was crucial in the first place. We seem to have presupposed it in discussion.) OK, so the case goes like this. It seems that there *could* be a being whose conscious experience is holistically tied to its concepts. This is like an extreme case of Tim's bat case. The idea is that when one of these beings acquires a new concept, all of its conscious experiences are modified (slightly or radically , however you like). Assuming that having a concept of this being's phenomenal states requires having that state, then *any* being which has different concepts (and certainly an ideal conceiver which has all possible concepts) cannot have a concept of these experiences. Actually, as the preceding sentence shows, there is no being which can have all possible concepts. Here is a similar case. Imagine a being whose conscious experiences are holistically tied to its *beliefs* rather than its concepts. Any change in belief changes all the qualitative feels of its experiences. So such a being who believes that P will have perhaps radically different types of phenomenal experiences from an identical being that believes that not-P or believes neither P nor not-P. Now, surely our ideal conceiver believes P, not-P, or has neither belief. If it believes P, then it can't have the experience of a being (in the hypothetical species) that believes not-P or neither, etc.... Now surely the beings that I've imagined are possible. I think that even human beings have some of these properties. In the conceptual case, imagine the difference between seeing a chessboard and lacking the concept of chess and seeing a chessboard *as* a chessboard. In the belief case, imagine the difference between eating cowbrain and not knowing what you are eating and eating the brain and knowing it (yuk!). I don't think we want to call the cases I've discussed open inconceivabilties. After all, they are conceivable by *some* being, just not by the ideal conceiver we've hypothesized. Do we need a single ideal conceiver? It seems like we don't, though I haven't thought this out. Things seem to get pretty messy without the notion of a single ideal conceiver. But I suppose we already had a gap between human conceivability and ideal conceivability. Now we have a gap between *any* being and "total conceivability"--that is, between what any particular being can conceive and the totality of things conceivable by atleast one being. Brad -----Original Message----- From: David Chalmers To: MODALITY@LISTSERV.ARIZONA.EDU Date: Friday, April 16, 1999 2:46 PM Subject: Re: An inscrutable truth? Probably not. >Tim's bat example is interesting. I should note that this isn't >really being put forward as an instance of an inscrutable truth, but >rather as an instance of an unknowable (perhaps even unbelievable) >truth. It's important to note that these are not the same thing >(though arguably inscrutable truths, if they exist, will be >unknowable). > >Of the categories we've been talking about, the truths in question are >probably most relevant to open inconceivability than anything else. >One might use Tim's argument to say: (i) to conceive of these truths, >one needs to have battish experiences; (ii) only bats can have battish >experiences; (iii) bats don't have the relevant concepts; so (iv) no >possible being can conceive of these truths. > >Tim's superbat objection would strike at premise (ii) here, and indeed >that seems reasonable. One might try getting around this by talking >about "dumb bat experiences", the sort that can only be had by an >unintelligent bat. Superbats could never have these experiences, I >presume! But maybe they could form the concepts some other way. > >Another relevant route is to question (i) (which corresponds to Tim's >(3)). Arguably one can conceive of some experiences without having >had those exact experiences, but rather by having experiences that are >relevantly related to them. Hume's "missing shade of blue" is an >example. And maybe e.g. we could use our own experiences to form >concepts of certain "proto-experiences" that compose them, and then >recombine the protoexperiential concepts to form concepts of >experiences that we've never had. Maybe that sort of thing could lead >a being without bat experiences to form concepts of bat experiences >(and even of "dumb bat experiences"). It's not obvious that this is >possible, but it's not obvious that it's impossible, either. > >--Dave. > From owner-modality@LISTSERV.ARIZONA.EDU Mon Apr 19 21:14:06 1999 Date: Mon, 19 Apr 1999 21:13:48 -0700 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: A problem for the notion of a single ideal conceiver To: MODALITY@LISTSERV.ARIZONA.EDU Status: R Brad discusses some problems for the notion of an ideal conceiver. I note that we decided that this notion might be problematic and was unnecessary: instead of defining ideal conceivability in terms of what an ideal conceiver can conceive, we can define it in terms of conceivability "in the limit" among non-ideal conceivers who get smarter and smarter (e.g., P is ideally conceivable if there is some X such that P is conceivable for X and P is conceivable for all beings less limited than X). That being said, the problem still raises issues even for the modified view of ideal conceivability. >OK, so the case goes like this. It seems that there *could* be a being >whose conscious experience is holistically tied to its concepts. This is >like an extreme case of Tim's bat case. The idea is that when one of these >beings acquires a new concept, all of its conscious experiences are modified >(slightly or radically , however you like). Assuming that having a concept >of this being's phenomenal states requires having that state, then *any* >being which has different concepts (and certainly an ideal conceiver which >has all possible concepts) cannot have a concept of these experiences. >Actually, as the preceding sentence shows, there is no being which can have >all possible concepts. > >Here is a similar case. Imagine a being whose conscious experiences are >holistically tied to its *beliefs* rather than its concepts. Any change in >belief changes all the qualitative feels of its experiences. So such a >being who believes that P will have perhaps radically different types of >phenomenal experiences from an identical being that believes that not-P or >believes neither P nor not-P. Now, surely our ideal conceiver believes P, >not-P, or has neither belief. If it believes P, then it can't have the >experience of a being (in the hypothetical species) that believes not-P or >neither, etc.... Interesting cases. These seem to be somewhat similar to the "dumb bat experiences", in that they are experiences that can only be had by beings below a certain level of sophistication. As such, then none of our "sufficiently smart" beings will be able to have those experiences. Does it then follow that these experiences are not ideally conceivable? That conclusion seems to follow, *unless* there is a way the beings in question can conceive of the experiences without actually having them. We've seen that this can happen at least in some cases, e.g. Hume's "missing shade of blue", and novel experiences that involve recombination of elements of familiar experiences. Perhaps something like that could happen here. E.g., perhaps a smart being with a rich conceptual systen could still conceive of the phenomenology of a being with a limited conceptual system, perhaps by "scaling down" their own phenomenology in their imagination, or by some sort of analogical conception. Similarly, a being that believes P can arguably conceive what it's like to believe that not-P, perhaps by "changing the polarity" of their current belief in the imagination. E.g., even though I believe I was born in Australia, I think I can conceive of the phenomenology of believing I wasn't born in Australia, even though I don't have that phenomenology myself. Of course it isn't obvious that these methods will work for every possible case. But it's not obvious that they won't. If they do, things are OK. If they don't, it would cause trouble for the concept of ideal conceivability, and certainly for the link between ideal conceivability and possibility. The possibilities in question would come out to be open inconceivabilities by the official definition, which might seem odd. An alternative would be to create a new class for statements that are conceivable by some beings but not by smarter beings; but then the question is how to distinguish between the "coherent" cases like this (e.g. the ones just discussed) from the "incoherent" ones that are implicitly contradictory. That's particularly problematic if there are cases (a) involving concepts that can only be had by limited beings and (b) that involve subtle incoherenies on the part of those beings. We want to class these with the "incoherent" cases, but clearly the criterion of being undermines by rational reflection won't do, as beings that could perform the rational reflection wouldn't have the concepts! Maybe there is an alternative way of running the counterfactuals -- e.g. they're incoherent because *if* a being could have the concepts and be smarter, they would uncover a contradiction. Such a counterfactual would have an impossible antecedent and might thus be considered problematic, but arguably counterfactuals with impossible antecedents can sometimes have truth values. In any case, this would clearly be a bit of a mess. So let's hope that such cases can't really arise! I.e., let's hope that the experiences in question can at least be conceived of by smarter beings, even if they can't be had by smarter beings. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Tue Apr 20 11:16:33 1999 Date: Tue, 20 Apr 1999 11:14:32 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Erik A Herman Subject: ideal concievers To: MODALITY@LISTSERV.ARIZONA.EDU Status: R I think Brad's example of "a being whose conscious experiences are holistically tied to it's beliefs" is exactly the case with humans in the actual world. A change in belief DOES change the qualitative feel of things--and in a new way. As far as the ideal conceiver goes though, being ideal and all, can't it conceive of what it's like to beleive not-p while believing p? The ideal conceiver has considered both of them already and decided that p is the case, but probably "remembers" what it was like to belive not-p. So, using Brad's example, the ideal conceiver would be able to conceive of how you feel when you're eating cow brain in the case where you know what it is AND the case where you don't. Of course, there is going to be a different qualitative feel to the ideal conceiver because it can't remove itself wholly from being an ideal conceiver as opposed to the person actually EATING cow brain, but we would call what it is doing, "remembering", to be a form of "concieving of": [what it is like to eat cow brain without knowing that it is cow brain] wouldn't we? erik h. PS ( I hope it is OK to refer to the ideal conceiver as "it" ) From owner-modality@LISTSERV.ARIZONA.EDU Tue Apr 20 12:33:45 1999 Date: Tue, 20 Apr 1999 12:12:12 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Josh Cowley Subject: open conceivabilities or the lack there of To: MODALITY@LISTSERV.ARIZONA.EDU Status: R Here is an argument that I think shows there are no open conceivabilities. First some of Dave's definitions: > S is an INSCRUTABLE TRUTH if S is true and some PC-complete truth D > does not imply S. > S is a GENERALIZED INSCRUTABLE if there exists some PC-complete D such > that D-and-S is negatively conceivable, but D does not imply S. > S is an OPEN INCONCEIVABILITY if S is negatively conceivable, but for > all PC-complete D, D implies not-S. Because of the truth constraint in the def of INSCRUTABLE TRUTH I'm going to focus on a comparison of generalized inscrutable (GI) and open inconceivability (OI). The primary difference between GI's and OI's is that for GI's no D implies S while for OI's all Ds imply ~S. However, S is negatively conceivable for both GI and OI. What I'm not getting is how something can be negatively conceivable and yet be false in every conceivable world. Let's say that I know S is false in some PC-complete D. This is the same as saying that if I add S to D then I get a contradiction. Now if S implies a contradiction in every PC-complete world I can imagine then it seems to me that S leads to an obvious contradiction. But that is to deny that S is negatively conceivable. Now if S isn't even negatively conceivable then I see no reason for thinking S is possible. On the other hand if S doesn't imply a contradiction in every world then there is some D such that S is either true or undecided. In the latter case we just have an inscrutable truth. Josh From owner-modality@LISTSERV.ARIZONA.EDU Tue Apr 20 15:15:20 1999 Date: Tue, 20 Apr 1999 15:11:14 -0700 Sender: "Philosophy 596B: Mind and Modality" From: David Chalmers Subject: Re: open conceivabilities or the lack there of To: MODALITY@LISTSERV.ARIZONA.EDU Status: R Reply Josh writes: >Here is an argument that I think shows there are no open >conceivabilities. First some of Dave's definitions: > >> S is an INSCRUTABLE TRUTH if S is true and some PC-complete truth D >> does not imply S. > >> S is a GENERALIZED INSCRUTABLE if there exists some PC-complete D such >> that D-and-S is negatively conceivable, but D does not imply S. > >> S is an OPEN INCONCEIVABILITY if S is negatively conceivable, but for >> all PC-complete D, D implies not-S. > >Because of the truth constraint in the def of INSCRUTABLE TRUTH I'm >going to focus on a comparison of generalized inscrutable (GI) and open >inconceivability (OI). > >The primary difference between GI's and OI's is that for GI's no D >implies S while for OI's all Ds imply ~S. However, S is negatively >conceivable for both GI and OI. What I'm not getting is how something >can be negatively conceivable and yet be false in every conceivable >world. Let's say that I know S is false in some PC-complete D. This >is the same as saying that if I add S to D then I get a >contradiction. Now if S implies a contradiction in every PC-complete >world I can imagine then it seems to me that S leads to an obvious >contradiction. But that is to deny that S is negatively conceivable. Hmm. Let's take a statement such as "there are inconceivable features of the world". It's true that this is false in any PC-complete D, and combined with any such D leads to a contradiction. But that's not to say that S alone leads to a contradiction, or that S is negatively conceivable. That would requires that not-S is a priori. Now *if* one could know a priori that there is some PC-complete description of the actual world, then it would more or less follow from the above that one could know a priori that not-S. But it's not obvious that we can know the antecedent a priori, and indeed that's what's at issue. So, it seems that for this argument to rule out open inconceivabilities, we'll first need a way to rule out open inconceivabilities! >Now if S isn't even negatively conceivable then I see no reason for >thinking S is possible. On the other hand if S doesn't imply a >contradiction in every world then there is some D such that S is >either true or undecided. In the latter case we just have an >inscrutable truth. Well, for an open inconceivability S, there is the chance that (i) S may imply a contradiction in all positively conceivable worlds, but (ii) S may not imply a contradiction in all possible worlds. If there are open inconceivabilities, there may be more possible worlds than positivly conceivable worlds. --Dave. From owner-modality@LISTSERV.ARIZONA.EDU Tue Apr 20 14:55:55 1999 Date: Tue, 20 Apr 1999 14:53:12 -0700 Sender: "Philosophy 596B: Mind and Modality" From: Rachael J Parkinson Subject: Re: A problem for the notion of a single ideal conceiver To: MODALITY@LISTSERV.ARIZONA.EDU Status: R Dave responded to Tim and Brad's cases by suggesting that beings can conceive of experiences without really having them. Hume's "missing shade of blue" may be such a case, and yet it seems to be much more simple then the "dumb bat" cases we've been discussing. I am not sure that it is possible for an ideal conceiver to conceive of what it's like to be a bat simply by scaling down his own phenomenological experiences. This is because the difference in phenomenological feel between an ideal conceiver and a bat seems to be a difference of kind and not of degree. I can not conceive of what it's like to be an ant by simply imagining myself, only much stupider. Also, even if an ideal conceiver were to conceive of what it's like to be a bat by simply scaling down his own phenomenology, it seems that he would still have a lot of extra baggage in virtue of being an ideal conceiver that would taint his conception. I can remember what it's like to be five, but I don't know that I really have a pure concept of the feel of what it's like to be five because I have so much 'experience and wisdom' that lurks in the background when I try to conceive of it. (I can never really conceive of the wonder I felt toward the world because now I can better explain the world... Of course I could just take the little wonder that I feel now and multiply it, but I'm not sure that that really does the trick.) I guess a lot rides on what is required to have a positive conception of another's experience- my intuitions seem to be that you do need to have the experience in order to conceive of it. (Of course, a person who has never had a headache can presumably conceive of what it's like to have a headache, but that brings up an epistemological problem- unless he's had the headache- how does he know that he's getting it right? Likewise for the ideal conceiver and the dumb bat.) -Rachael > That conclusion seems to follow, *unless* there is a way the beings in > question can conceive of the experiences without actually having them. > We've seen that this can happen at least in some cases, e.g. Hume's > "missing shade of blue", and novel experiences that involve > recombination of elements of familiar experiences. Perhaps something > like that could happen here. E.g., perhaps a smart being with a rich > conceptual systen could still conceive of the phenomenology of a being > with a limited conceptual system, perhaps by "scaling down" their own > phenomenology in their imagination, or by some sort of analogical > conception. Similarly, a being that believes P can arguably conceive > what it's like to believe that not-P, perhaps by "changing the > polarity" of their current belief in the imagination. E.g., even > though I believe I was born in Australia, I think I can conceive of > the phenomenology of believing I wasn't born in Australia, even though > I don't have that phenomenology myself. >