Ghosts and Sparse Properties Presenter: Philip Goff, University of Hertfordshire Philip’s Paper Commentator 1: Andrew Melnyk, University of Missouri Andrew’s Paper and Podcast or download the video in Ogg format Commentator 2: Esa Diaz-Leon, Manitoba University, Canada Esa’s Paper Share this:FacebookTwitterEmailRedditPrintPinterestTumblrLinkedInLike this:Like Loading... Related 25 Comments Hi there! I have some questions about Andrew’s commentary. I think he raises lots of interesting issues. Regarding his response to Philip’s ghost argument against funny physicalism, I really liked the idea that a property that is actually non-fundamental does not have to be essentially non-fundamental, that is, a property might be fundamental in one possible world but non-fundamental in another (I was wondering what an example of a plausible case would look like). I also liked Andrew’s responses to Philip’s arguments against a posteriori physicalism. Regarding Philip’s ghost argument against a priori physicalism, Andrew seems to respond that not all versions of a priori physicalism entail that it’s conceptually true that phenomenal properties are essentially higher-order: for instance, Lewis’ version of realizer-functionalism claims that phenomenal properties are identical to first-order physical properties (that is, the properties that realize certain functional roles, not the higher-order properties of having one or another first-order property satisfying the functional role, as role-functionalism would have it). I think this a great point against P2 of that argument. However, I was thinking that the conceivability of ghosts seems to pose a challenge to Lewis’ version of analytic functionalism anyway: if zombies are conceivable, then phenomenal properties could (conceivably) be instantiated even if they do not realize the corresponding functional roles, and this seems to be contrary to (analytic) realizer functionalism: according to this view, it is conceptually true that if M is a phenomenal property, then M meets certain causal/functional descriptions. Many, many, many thank for Andrew and Esa for dead good comments! Here are my responses to Andrew. Responses to Esa to follow, hopefully at some point today. Ghost argument against funny physicalism Andrew correctly identifies an important premise of my argument: ‘P2. If a property is actually non-fundamental, then it’s essentially non-fundamental (i.e., non-fundamental in all possible worlds in which it’s instantiated)’. However, he seems to take it that P2 features in my paper as a basic assumption which I don’t argue for; this is not the case: I give extensive argument for P2 in section XIII. Andrew points out that Shoemaker has a view which entails the falsity of this P2. But in the absence of a response to my arguments for P2 in section XIII, this merely shows that my argument has force against Shoemaker’s view as well as against funny physicalism (I am grateful for Andrew for pointing out this out; I’ve never liked the Shoemaker view). Ghost argument against a priori physicalism Andrew questions the following premise: ‘A priori physicalism entails that it is a conceptual truth that conscious states are essentially higher-order states (it is a conceptual truth that functional states must be realised by more fundamental causal goings on)’, on the grounds that a Lewis-style view might claim that functional properties are only involved in reference fixing of the concept ‘pain’, whilst the referent of the concept might be a fundamental property. I find Lewis’s view very confusing, so it would be very useful for me if we can clear this matter up. As I see it, on Lewis’s view, ‘pain’ is a flaccid designator, so my pain might not have been a pain (just as ‘the current president of the USA’, i.e Obama, might not have been the current President of the USA). For my pain to be a pain is for it to satisfy a certain causal description. But when I am conceiving of my ghost twin feeling pain, I am conceiving of my pain being a pain, whilst having no causal role, which is inconsistent with the Lewis/Armstrong style view. Have I got the view wrong? First ghost argument against a posteriori physicalism Andrew doubts the following premise, ‘If my mode of consciousness is a fundamental property, then a posteriori physicalism is false (I define ‘physicalism’ such that any form of physicalism holds that my conscious states supervene on the nature of fundamental physical entities)’, on the grounds that my mode of consciousness might supervene on fundamental physical entities without being less fundamental than fundamental physical entities (by being a proper subset of fundamental physical entities). This is a good point; I wasn’t quite precise enough with what I wrote in the brackets. I should have said: I take it to be part of the definition of ‘physicalism’ that (i) facts about the conscious states of an organism supervene on fundamental physical facts, (ii) those fundamental physical entities are more fundamental than facts about the conscious states of an organism. If you are not a ‘physicalist’ in this sense, then you’re not my enemy (in this paper). Second ghost argument against a posteriori physicalism Andrew doubts the following premise, ‘If my mode of consciousness is not an essentially higher-order state, then a posteriori physicalism is false (a posteriori physicalists wish to identify actual modes of consciousness with functional or neurophysiological states; such states look to be essentially higher-order, e.g. a state which is defined in terms of billions of neurons must be realised by neurons, each of which is more fundamental than the neurophysiological state they constitute)’. His first reason repeats the claim that P2 in the argument against funny physicalism may be false. But again, he says nothing in response to my extensive arguments for P2. Secondly, he suggests that proper parts may not be more fundamental than the whole they constitute. Again, I am just defining ‘physicalism’ (roughly) as the view that fundamental physical facts, which involve quantifying over dead little things like protons, electrons and quarks, are more fundamental than facts about brain states or the mental states of organisms (I probably should have been clearer about this). Finally, I completely agree with what Andrew has to say about the link between conceivability and possibility, and I don’t think Chalmers’ arguments against physicalism work for exactly the same reason Andrew doesn’t think they work. This paper wasn’t intended primarily as an attack on a posteriori physicalism. Fortunately, there is a knock down argument against a posteriori physicalism, which I outline in my forthcoming AJP paper: ‘A posteriori physicalists get our phenomenal concepts wrong’, which is available on my academia page: http://herts.academia.edu/PhilipGoff (shameless publicity). Also, someone called KELLY HARMAN/HARMON emailed me about this paper, and I accidentally permanently deleted the email before responding. If you see this, could you please send me the email again? Sorry! Reply to Esa’s comments: Esa has some worries that my treatment of a priori physicalism differs from my treatment of funny physicalism, and that my treatment of funny physicalism vis-à-vis ghost arguments differs from my treatment of funny physicalism vis-à-vis zombie arguments. Let us suppose for a moment that, with regards to each of the following three disjunctions, the first disjunct is inconsistent with the second disjunct: (A): ‘PhysicSal properties a priori entail phenomenal properties’ or ‘PhysicSal zombies are conceivable’. (B): ‘Deep physical properties a priori entail phenomenal properties’ or ‘Deep zombies are conceivable’ (C): ‘Deep physical properties a priori entail phenomenal properties’ or ‘Ghosts are conceivable’ I am completely happy to agree that, in the absence of an argument to the contrary, the a priori physicalist is entitled to hold that physicSal properties entail phenomenal properties. But there is an argument to the contrary: zombies seem upon reflection to be conceivable, and this is inconsistent with the first disjunct of (A). The matter is quite different when it comes to the funny physicalist’s claim that deep physical properties entail phenomenal properties, for no one has the concepts required to be able to conceive of deep zombies, and so we have no analogous zombie argument to the conclusion that the first disjunct of (B) is false. This is what justifies my different treatment of a priori physicalism and funny physicalism vis-à-vis zombie arguments. I am not saying that zombie arguments against a priori physicalists are knock down arguments, but there is a clearly a difference between the two cases: there is a zombie argument against a priori physicalism, but there is no zombie argument against funny physicalism. So we do not have a zombie argument for the claim that . But, on the assumption that the disjuncts of (C) are inconsistent, we do have a ghost argument to the conclusion that the first disjunct of (C) is false: we have good reason to think that ghosts are conceivable, and (we are assuming) this is inconsistent with the first disjunct of (C). This is what justifies my different treatment of funny physicalism vis-à-vis ghost arguments and funny physicalism vis-à-vis zombie arguments. Again, I don’t claim that this is a knock down argument, but there is clearly a difference between the two cases: there is no zombie argument against funny physicalism, but there is a ghost argument against funny physicalism. Furthermore, I am not sure that I want to grant the assumption that the disjuncts of (C) are inconsistent. This is an assumption Esa makes, but not one that I have made. Nor does it seem to me to be one that my argument is reliant on. I would like to know how Esa justifies this assumption and/or why she thinks my arguments is reliant on it. Esa worries that maybe our epistemic relationship to physicSal properties isn’t so different to our epistemic relationship to deep physical properties, given that ‘we do not have a complete account of dispositional physical properties as of yet’. I do suggest at one point that this may be seen to be a weakness of zombie arguments (although having said that, Chalmers, in ‘Consciousness and its place in nature’ argues that against this worry), but it is not a weakness of ghost arguments: we don’t have to wait for the end of science in order to conceive of a ghost. Esa next considers an argument for the claim that protophenomenal properties do not entail phenomenal properties. Given that I am attacking funny physicalism, this argument doesn’t seem to be a challenge to the claims of my paper. However, I do wonder how the premise of this argument that phenomenal properties are not conceptually connected to non-phenomenal properties can be justified. I completely agree with Esa that this paper presents no new arguments against a posteriori physicalism (It is worth noting that both my commentators are a posteriori physicalists, and this paper is not meant to raise any new challenge against a posteriori physicalism). Regarding Esa’s second lot of comments: As I said in my response to Andrew, my concern with Andrew’s objection to my premise ‘If a property is fundamental in one world, then it is fundamental in all worlds’, is that he doesn’t seem to given a response to my arguments for this claim in section VIII. He points out that Shoemaker’s view entails the falsity of this premise, but in the absence of a response to my arguments in section VIII, this merely shows that my argument works against Shoemaker too. Regarding the issue of whether my ghostly argument against a priori physicalism works against Lewis: mine and Esa’s understanding of Lewis’s view seem to be inconsistent. I thought Lewis could allow pain to exist in counterfactual possible worlds (although not in worlds considered as actual) without having the pain role (he calls it ‘mad pain’). But I still think my argument works against him, because in these worlds pain wouldn’t feel painful, whereas I can conceive of my ghost twin having pain that feels painful. If I have misunderstood Lewis, it would be really helpful to me to clear this up… Hi everyone, glad to see the discussion in here blooming! So, as an a priori physicalist I guess I ought to say something and stop having a posteriori physicalist fighting my battles 🙂 The way I see a priori physicSalism as being able to pull of the a priori deductions is as follows. First I take it that there is no in principle reason that we should not be able to a priori deduce nrual activity from the behavior of physicSal properties (Chalmers grants this). So what we need is a good theory of what we are trying to explain. Let’s take the higher-order thought theory as an example (I don’t think any version of it is true, but only that they could be and that is enough to make the point). Suppose that we accept as a pre-theoretic platitude that a mental state is conscious only when I am conscious of myself as being in it in an appropriate way (this is what Rosenthal calls the Transitivity Principle). Suppose then that we are able to identify the states in the brain that represent perceptible properties like red, green, etc (suppose that it is synchronous neural activity in a certain frequency among neurons distributed through-out the brain (or a region of the brain). Suppose then that we found that certain higher-order states occurred and that we could tell that they were thought about the first-order states (this may turn out to be in the pre-frontal areas of the brain). Using the transitivity principle we can a priori deduce that there being something that it is like for me consists in my being conscious of myself as being in a state with qualitative properties (or in respect of the first-order state’s qualitative properties). We could then identify the higher-order thought, in virtue of which there is something that it is like for me to have the experience, with the neural activity in the pfc (say). So in this sense my view is like Lewis-Style realizer functionalism. But I think that this identity, just like all Kripkean “a posteriori” identities has a contingent primary intension and a necessary secondary intension. Given this Ghosts pose no problem what so ever. Compare: that it is conceivable that water is xyz is no threat to the identity of water and H20. This is because “the watery-stuff is H20” is contingent. So too “the painful stuff is activation in pfc” is contingent. So it is conceivable that the painful-stuff be other than activation in the pfc. Are ghosts ideally conceivable? I think that they might be prima facie negatively conceivable (there is no obvious contradiction in the process of Cartesian doubt) but I can’t make any sense of the idea that they are positively conceivable. In order for that to be the case we would need quite a lot more detail about nonphysical properties (as of yet we have none). You seem to think that this is obvious but when I poll my undergraduates a lot of them are dumbfounded with the suggestion that we might be ghosts. Finally, I wonder what you make of the physicSalist counter-conceivability arguments. So, take what I call shombies. A shombie, in your terms, is a creature that (1) is conscious (2) is such that its modes of consciousness DO supervene on some more fundamental property (3) has no fundamental properties other than the physicSal ones Shombies are conceivable (i just outlined how to do it). So you can get a zombie-style argument against fundamentalism. Can you get a ghost like argument? I think so. I have called these things ‘zoombies’. A zoombie, as I originally put it, is a creature that is nonphysically identical to me but which lacks qualitative properties. It has all of the very same nonphysical properties as I do but no (non-physical) qualitative properties. To put it in your terms a zoombie is a creature that no fundamental conscious properties but none the less has ALL of the very same fundamental properties as I do. Richard: So let’s call ‘Lewisian physicalism’ the view that our mental concepts are rigid designators which pick out certain states in terms of their actual causal role. I take it that this view implies that in all worlds considered as actual, conscious states have the causal role associated with the corresponding mental concept (just as in all worlds considered as actual water is watery). I also take it that the in counterfactual worlds where, say, pain, does not have the causal role associated with the concept of pain, it does not feel painful (for we pick out pain in terms of its feeling painful, just as we pick out water in terms of its being watery). Ghosts seems to me to be problematic to this view for the following two reasons: 1. It is conceivable that in the actual world there is a ghost (when I get to the end of the process of Cartesian doubting I am conceiving that in the actual world only I exist and that I am a ghost). This seems to me inconsistent with Lewisian physicalism, as it implies that we can conceive of a world considered as actual where conscious states have no causal role. 2. When I conceive of a ghost in pain I conceive of its pain as feeling painful, even though its pain does not play the pain role. This is inconsistent with Lewisian physicalism, as that view implies that in worlds where pain doesn’t play the pain role, pain isn’t painful. I think shombies are conceivable, but not possible. Why should I think they are possible? Andrew and Esa: I made a mistake in my comments to you on the Lewisian view. I should have said that pain is a rigid designator on this view, see my last post for my latest thoughts (which may be horribly wrong…) on ghosts and the Lewisian view. Hi Phil, hi everyone. Really interesting paper and comments. I have a couple of questions. I’m not sure quite how they relate to questions already raised, but I’ll ask them anyway. Apologies for any overlap. My main question is about the role that principle CP plays in the argument. You’re arguing that if the funny physicalist accepts CP, she can’t then maintain that there’s a conceivability/possibility gap with respect to the existence of ghosts. And you’re arguing that the funny physicalist should indeed accept CP, because if she doesn’t then her view will collapse into a more standard version of a posteriori physicalism. I’m not sure about the second part of this argument: I think that the funny physicalist can deny CP without her view collapsing into standard a posteriori physicalism. What the funny physicalist needs is a principle which, in conjunction with principle PCD, entails that standard a posteriori physicalism is false. But the funny physicalist doesn’t need anything as strong as CP to do that. All she needs is a principle which specifically concerns the relation between phenomenal and physicSal concepts. It also seems to me that we have independent reasons for denying CP as it stands. Consider the rigidified description ‘John’s favourite actual property’. It’s conceivable that John’s favourite actual property is phenomenal property P. But if it turns out that John’s favourite property is actually property Q, then it’s metaphysically impossible that John’s favourite actual property is P. So we have a conceivability/possibility gap of the relevant sort here. So I think that it’s open to the funny physicalist to reject CP, and that she has independent reasons for doing so. Given that, I think that the funny physicalist can plausibly concede the conceivability of ghosts but deny their possibility. My other question is about the notion of genuine resemblance. As I understand it, you argue that you genuinely resemble your ghost twins, and hence (by Lewis’s principle of genuine resemblance) that your mode of consciousness is a sparse property. But I don’t see why the funny physicalist needs to concede that you genuinely resemble your ghost twin, in Lewis’s sense of genuine resemblance. What’s to stop the funny physicalist from saying this: ‘Of course there’s *a* sense in which you resemble your ghost twin – your conscious experiences are qualitatively identical, after all – but you don’t *genuinely* resemble your ghost twin, precisely because your mode of consciousness is not a sparse property.’ P.s. to Richard…forgot to say summit about your worries about conceiving of ghosts…(comments to Charlie-boy to follow) 1. Isn’t the negative conceivability of ghosts worrying enough? 2. I’m not sure what more you need to know about the properties of a ghost. Its only property is its mode of consciousness. It’s conceivable that that very property exhausts the nature of a thing (I might add that it seems to be a lot less clear whether it is conceivable that physicSal properties, which are a matter of having certain dispositions or mathematically describable structure, exhaust the nature of a thing). My ghost isn’t a creature whose mode of consciousness is realised in some ectoplasmic non-physical stuff; I agree that we would want to know more about the nature of the ectoplasm to conceive of such a thing. My ghost just has its mode of consciousness as its one and only fundamental property. 3. Have you tried to get your undergrads to conceive of my kind of ghost, i.e. nature exhausted by consciousness, rather than just a non physical thing? If you just ask them to conceive of a non-physical mind, then they’ll inevitably try to imagine some something whose consciousness is grounded in some kind of ectoplasmic non-physical stuff, and I agree that’s not conceivable till we know more about the nature of ectoplasm (answers on a post card…). I recommend explaining what a ghost is in my sense, and telling them to go through the Cartesian doubting process, and ask themselves at the end whether they are having a positive conception of themselves as ghosts. I think you’ll get better results! Hi Philip, thanks for the interesting response! I am not sure why you think that in the Ghost world conscious states have no causal role. Allowing that what you say in 1 is conceivable (I don’t really think it is) then this is just a case of nonphysical states playing the roles (something functionalism used to be celebrated for). So the Ghost world, considered as actual, is one where the primary intension of pain picks out nonphysical states but this is no problem because in that world they are picked out in just the same way that we pick ours out (this is a consequence by the process of Cartesian doubting). This means that we are picking them out via the same causal roles. Shombies are possible because of the link between conceivability and possibility; that’s what all of these a priori arguments depend on. Thanks for these further remarks Phil! Re 1: It would be if it were ideal negative conceivability but I only allowed prima facie negative conceivability and that’s not worrying. Re 2 & 3: I can’t make sense of what this is supposed to mean. It isn’t what Descartes has us doing in the Meditations. This is because a mode has to be a modification of something. This is why so many people read Descartes as a substance dualist (though I suspect that you might not, having studied with Galen and all..). Here is another way to make the point. You say that it might actually be the case that we are ghosts. Suppose that we do in fact live in the ghost world. Can’t I still conceive of zombies? If so what am I conceiving? A creature just like me but which lack consciousness. Since (ex hypothesi) I am in the ghost world that means that I am conceiving of a world where there are creatures with the same fundamental properties but which lack consciousness. Call these creatures ‘Ghombies’. Ghombies are conceivable and they show that consciousness can’t be a fundamental property in the way your (2) has it. P.S. i think there is a really interesting XPhi experiment waiting to happen here with respect to (3) if anyone wanted to do it… I want to discuss part of Philip’s response to my comments on his paper (see comment #2). He writes: Andrew correctly identifies an important premise of my argument: ‘P2. If a property is actually non-fundamental, then it’s essentially non-fundamental (i.e., non-fundamental in all possible worlds in which it’s instantiated)’. However, he seems to take it that P2 features in my paper as a basic assumption which I don’t argue for; this is not the case: I give extensive argument for P2 in section XIII [he means VIII, surely]. Philip is right, so let me explain why I find his case for P2 weak. He argues as follows (pp. 13-14): By the definition of ‘ghost’, my ghost twin’s mode of consciousness is a fundamental property of its world. It follows straightforwardly, given the principle of fundamentality, that my ghost twin’s mode of consciousness is a sparse property of its world. ..… [T]he fact that my ghost twin’s mode of consciousness is a sparse property of its world entails that my mode of consciousness is a sparse property of the actual world,….. If my mode of consciousness is a sparse property, it follows, given the principle of fundamentality, that my mode of consciousness is a fundamental property. In effect, Philip argues from the fundamentality of my ghost twin’s mode of consciousness in his world to the fundamentality of my mode of consciousness in the actual world, which is approximately the contrapositive of P2. Here is the principle of fundamentality, to which he appeals twice: “The sparse properties in a world are the fundamental properties in that world, i.e. the properties in the supervenience base at that world.” I take this principle to say that, in any possible world, property P is sparse iff it’s fundamental. It is the principle when read from right to left that is appealed to first; it is the principle when read from left to right that is appealed to second. My problem with Philip’s argument is that I see no sufficient reason to endorse the principle of fundamentality when read from left to right, i.e., as claiming that, in any possible world, every sparse property is fundamental. I don’t see why a sparse property—a property that grounds genuine similarity—shouldn’t be non-fundamental. Philip presents three reasons for thinking that “it is a very good idea indeed to go on to identify the Real/natural properties with the fundamental properties” (p. 16); and I assume he would be happy to replace “Real/natural” with “Real/natural/sparse”. So his three reasons are reasons for accepting the principle of fundamentality. 1. The crux of Philip’s first reason is that “there is no need to suppose the existence of any non-fundamental Real properties to serve as truth-makers for p. ” (p. 16). But this doesn’t give a positive reason to think it true that every sparse property is fundamental; it merely says there’s no reason (of one particular kind) to think the generalization false. Moreover, there might be other reasons, unrelated to truth-making, to think there are sparse but non-fundamental properties. 2. The heart of Philip’s second reason is this: “…it seems fair to say that the problem [of mental or special-scientific causation] looks a lot less troubling if we suppose that higher-level properties are not Real properties, ontologically additional to the Real properties at the fundamental level, but are merely qualitative truths about objects, the truth-makers for which reside at the fundamental level.” But to note, even correctly, that a claim would make an admittedly unsolved problem look less troubling is not much of a reason to believe it. In any case, higher-level properties that are sparse but non-fundamental wouldn’t be “ontologically additional” to the sparse and fundamental properties if the sparse but non-fundamental properties could be realized (e.g., in Shoemaker’s sense, which is exactly what he claims) by the sparse and fundamental ones. So the denier of the principle that all sparse properties are fundamental might be just as well-placed to solve the problem of higher-level causation as the believer in that principle. 3. Finally, here is Philip’s third reason: “Consider a world with fundamental facts f and some non-fundamental property p. Any world where f obtains is a world where p exists. The existence of f necessitates the existence of p. If p is a Real property, then we have a necessary connection between distinct existences.” True, though if p is multiply realized, the necessary connection will run in one direction only, from f to p, which removes the problem of saying what the distinctness of the two existences consist in. Is it clear that there is some other objection to uni-directional necessary connections between distinct existences? Regarding Charlie’s claim that the funny physicalist can make do with some principle relating physical and phenomenal concepts rather than CP: If you look at the specific principles endorsed by those who reject physicSalism, which I give examples of, they all entail C. For example, Bealer thinks that phenomenal concepts are semantically stable, and there is no gap between conceivability and possibility when we have semantically stable concepts. So I agree in principle that, if she could come up with some modal principle which just entailed that there is no conceivability/possibility gap when we’re dealing with phenomenal *and* physical properties, such that this principle has no implications for when we’re just dealing only with phenomenal properties, then the funny physicalist could defend her view with that principle. But I can’t imagine how such a principle could be justified: why would it be that it is only when we have *both* physical and phenomenal concepts that we there can’t be a con/poss gap? Secondly, in your alleged counterexample, to CP, the gap between con and poss opens up because we are conceiving of a certain property as ‘John’s favourite property’: it is conceivable but not possible that that property is phenomenal. Finally…take my two ghost twins, Bill and Ben. Bill and Ben perfectly resemble. But, regarding our conscious experience, the resemblance between Bill and I does not fall short of the resemblance between Bill and Ben. The last sentence is a premise of the argument that one could in principle deny, but I think it is a very plausible premise. Response to Richard: Lewis/zombie style ghosts are things whose mental properties have causal roles, realised in some non-physical stuff. But my ghosts don’t. My ghosts are what you get when you get to the end of Cartesian doubting. I can doubt that there is any causation in the actual world, and so the thing I end up conceiving of is not something with causal powers. I do read Descartes as a substance dualist (or at least I think it’s coherent to see his view as substance dualism). A ghost’s mode of consciousness *is* a modification of something: it’s a modification of thing that has no properties other than that mode of consciousness. If you find this metaphysically troubling, then you’re off on an unhealthy regress, because at some point you’re going to reach the conjunction of all the properties of a thing, call that conjunction F, and then you’ll ask: ‘But what is it that has F?’. The only answer is: a thing that has no properties other than F (perhaps you think we need to believe in substrata, but I could just say that if objects need substrata, then my ghost twin has a substratum). I could raise exactly the same worry you raise concerning a purely physical being, call it TABLE. Take the conjunction of all of TABLE’s physical properties, and call that property a ‘mode of physicality’. Here’s the worry: what is that mode of physicality a modification of? Well, it’s a modification of a particular thing, i.e. TABLE, which is such that its being is exhausted by that mode of physicality. In the same way, a ghost is a thing such that its nature is exhausted by its mode of consciousness. To repeat, if you just keep asking, ‘But what is the thing that has that property?’, then at some point you’re going to run out of properties (when you get to the conjunction of all the thing’s properties). The only sensible question is: which properties can exhaust the nature of a thing. It’s conceivable that a mode of consciousness can exhaust the nature of a thing (indeed I don’t think you’ve given me any reason to think it’s not ideally conceivable). It’s a lot less clear than we can conceive of physicSal properties exhausting the nature of a thing. If my ghost twin is conceiving of its zombie twin, presumably this is because it mistakenly thinks it has physical properties. It is conceivable for my ghost twin that ‘there is something just like me which doesn’t have consciousness’, but this is not possible. A gap between conceivability is opened up by the indexical reference to itself (compare ‘that stuff could exist without H20 molecules’ pointing to water). Hi Phil – thanks for the response! Just a quick follow-up on principle CP. Take the concepts ‘pain’ and ‘that which is actually John’s favourite phenomenal property’. It’s a priori that both concepts refer to phenomenal properties. Now take the proposition: Pain = that which is actually John’s favourite phenomenal property. This is conceivable. But it’s metaphysically impossible if John’s favourite phenomenal property is not in fact pain. So we have a conceivability / possibility gap here. Suppose we say that this gap opens up because of the descriptive nature of the concept ‘that which is actually John’s favourite phenomenal property’. The question then arises whether the funny physicalist’s conception of the relevant ‘deep’ phenomenal/protophenomenal properties is similarly descriptive. And it seems to me that it is. The conception is something like ‘those properties which actually subvene the relevant physicSal/conscious properties’. If this is right, then I can’t see what’s to stop the funny physicalist from claiming that a conceivability /possibility gap opens up w.r.t. the existence of ghosts. Hi all, thanks for the great discussion! I would like to make a brief comment to Charlie’s last post. I think the objection stands with respect to the version of funny physicalism that he mentions, that is, one that holds that (1): ‘phenomenal properties = those that actually subvene physicSal properties’ is an a posteriori metaphysical necessity. If (1) is a posteriori true, we could conceive of scenarios where it does not hold, for instance ghost worlds, in which phenomenal properties would be instantiated (and therefore the deep properties that are identical to them would also be instantiated) but they would not be the properties that subvene physicSal properties in that world. Therefore, (1) would be true in that possible world considered as actual but not as counterfactual. However, if I got it right, this is not the version of funny physicalism that Philip is attacking. For all (1) says, phenomenal properties might be fundamental after all. But Goff’s target are those versions of physicalism that hold that phenomenal properties supervene on fundamental properties. So the corresponding version of funny physicalism would be something like (2): ‘phenomenal properties = properties Q that supervene on those proto/phenomenal properties that actually subvene physicSal properties’. Now the question is: could funny physicalists maintain that (2) is also an a posteriori necessity? I think that in order to argue that it is not, based on conceivability considerations, it has to be argued that (2) is actually a priori false. If the argument was just that it is conceivably false, it could always be responded that (2) is an aposteriori necessity like (1) is. So I think that Goff’s ghost argument against (2) should be understood as an argument for the following being a priori true: (3) phenomenal properties are essentially fundamental. I guess that the argument for the conceivability of ghosts plus the principle of genuine resemblance plus the principle of fundamentality are supposed to provide a priori grounds for (3). If so, then it is clear that (2) cannot be an a posteriori truth because it could be ruled out a priori. Typo: in the ghost world above, (1) is False when considered as actual but True when considered as counterfactual (that is, (1) is an a posteriori necessity). Hi Philip, thanks for the response. Consider the Ghost world as actual. Then when we talk about causation we talk about our experiences being ordered in a certain way. The phenomenal properties that “exhaust the nature” of the ghost are ordered in various ways; for instance the experience of being kicked is followed by the experience of pain unless one has had the experience of anesthesia, etc. This is all that’s needed. I don’t know what you get at the end of Cartesian doubt…it seems to me at best prima facia negatively conceivable. You haven’t given any reason to think that it is positively conceivable.I just can’t make any sense out of the notion that my mode of consciousness exhausts my nature. So, I agree that the question is which properties can exhaust the nature of a thing and I just can’t picture what it would look like for a mode of consciousness to do that; in fact I doubt that you can either. A gap between conceivability [and possibility] is opened up by the indexical reference to itself (compare ‘that stuff could exist without H20 molecules’ pointing to water). I am not sure that I follow this. It is conceivable that ‘that stuff exist without H2O molecules;’ that’s just Twin Earth xyz…that’s the kind of conceivability that matters here and there is a link between conceivability and possibility at this level. The point is that I can conceive of the fundamental properties that you claim exhaust the nature of the ghost being instantiated without consciousness (the ghombie world which we arrive at by thinking about ghosts conceiving of zombie-like creatures) and so consciousness cannot be a fundamental property and so ghosts as you describe them are not ideally conceivable. Notice that if you are right it undermines the ghost argument. If you are conceiving of ghosts, presumably this is because you mistakenly think you have nonphysical properties. Response to Andrew’s responses to my three short arguments for the principle that the natural properties are the fundamental properties These three arguments depend on the assumption that we are being ‘ontologically serious’ about the natural properties, i.e. taking them to be Real properties (genuine universals or sets of tropes), rather than, say, just sets of objects. 1. My first argument is supposed to be an argument from parsimony. If we already have sufficient non-fundamental Real properties to make it the case ‘x is F’, then we don’t need to invest in a Real property (and hence a natural property) corresponding F to make ‘x is F’ true. I take it that this constitutes a reason to believe there is no Real property corresponding to F. 2. I don’t agree with Andrew’s claim the fact that a certain metaphysical view helps makes sense of mental causation doesn’t constitute a reason to believe that view. I would like to hear more about the relation of ‘realisation’. If ‘realisation’ is a matter of Real physical properties serving as truthmaker for some functional description, where the functional predicates don’t correspond to Real properties, then I can make sense of the realised functional properties not being ontologically additional to the physical properties. But if both the functional properties and the physical properties are Real properties, then I find it difficult to make sense of a relation which holds between them such that the former are not ontologically additional to the latter. Tricky issues though. It’s funny, most philosophers of mind find the notion of ‘ontological free lunches’ quite unproblematic, whilst many metaphysicians find the notion unintelligible (unless it’s the free lunch is just a matter of a truth the truthmakers for which we’ve already paid for). 3. I wasn’t thinking that the uni-directional nature of the necessary connection between truth and truthmaker makes it less problematic. Let me try to spell it out a bit more. Suppose we have a functional proposition , which is made true by some fundamental physical states of affairs. The proposition would have existed whether true or false, so nothing’s being necessitated into existence. We just have an internal relation between two things (if both the proposition and the truthmaker exists, they necessarily stand in the truthmaking relation). Now suppose we have a higher-level universal that is necessarily brought into existence by distinct fundamental universal. I take the latter kind of necessary connection to be much stranger. It’s strange to think that the existence of one concrete entity necessitates the existence of a distinct concrete entity. Reply to Charlie The description ‘John’s favourite phenomenal property’ necessarily refers to a phenomenal property, but when I’m thinking about that property under that description, I’m not thinking of it in terms of what it’s like. I am thinking of it as a phenomenal property, but I’m not thinking of it in terms of its determinate phenomenal nature. Because of this, this doesn’t count for me as thinking about that phenomenal property qua phenomenal property. I probably could have been clearer about this. As to your worry that a gap between con and poss opens up when we bring in protophenomenal properties, I don’t think that’s a problem for me. I want to argue from the conceivability to the possibility of ghosts, and when we’re conceiving of a ghost our conception involves only thoughts about phenomenal properties qua phenomenal properties. Esa and Charlie: Yes, I think Esa gets it right. The idea is that the funny physicalist has to believe that ghosts are possible, and if ghosts are possible then phenomenal properties are essentially fundamental, which is inconsistent with the funny physicalist claim that phenomenal properties supervene on proto-phenomenal properties (well, maybe Esa puts it a bit too strongly when she says I claim that (2) is a priori false: the claim is not that we can rule it out on conceptual grounds, we have to make some (very plausible) metaphysical assumptions). I only need there to be a link between con and poss when conceiving of ghosts. Richard, In response to your claim that ghosts are only negatively conceivable: when I get to the end of Cartesian doubting I am having a conception, therefore ghosts are positively conceivable (i.e. it’s not just that we can’t rule them out a priori, we actually end up conceiving of one when we get to the end of the process of Cartesian doubt). If you want to say that ghosts are not ideally conceivable, then I think you need to say that we cannot really complete the process of Cartesian doubt, or at least explain why the conception we end up with involves incoherence. Otherwise, your claim that ghosts are not ideally conceivable is just a leap of faith that there’s some incoherence involved in the conception we end up with at the end of Cartesian doubting, even though no one can explain to us why it is incoherent. If what we are conceiving of at the end of Cartesian doubting is coherent, then it is conceivable that consciousness exhausts the nature of a thing. I can’t see how on any theory of causation, a ghost with experiences ordered in a certain way is sufficient for causal relations to hold between those experiences. If we believe in causal powers/Armstrong-type laws of nature, then the matter is straightforward: a ghost world doesn’t have such things. But even if we hold a regularity/counterfactual theory of causation, it doesn’t seem that there are enough things in the ghost’s world to make it true that there are causal relations in that world. Furthermore, on any a priori physicalist theory, the relevant dispositions mediate not just between mental states, but between sensory inputs and behavioural outputs. A ghost has neither senses nor body. Thanks for comments and cool discussion everyone!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Very helpful! Comments are closed.