Dissolving the Hard Problem of Consciousness

Presenters: Glenn Carruthers, Macquarie University and Elizabeth Schier, Macquarie University & Berlin School of Mind Brain

Commentator 1: Janet Levin, University of Southern California

Commentator 2: Ellen Fridland, Berlin School of Mind and Brain & Humboldt University of Berlin

Commentator 3: Jennifer Matey, Florida International University

Advertisements

76 Comments

  1. A huge thank you to Richard, Janet, Ellen and Jennifer. It’s an honor to receive such thoughtful and helpful discussion of our work. Now how I’d like to approach this is to respond to each commentator individually as I think that each of you make distinct points worthy of their own consideration. I will start now with a response to the concerns Ellen has raised. Now these are my thoughts so don’t blame Liz for anything bizarre here!

    Why do we need a stronger notion of conceivability
    The main thrust of Ellen’s argument against our take on the zombie argument is that the notion of conceivability is too strong; too strong in the sense that it makes some things inconceivable that, in fact, are conceivable. This stronger notion of conceivability is necessary to allow that the argument around Zombies be informative. Let me explain.

    First of all it is important to note that she is right about what our notion of conceivability does, it does make it impossible for Ellen to conceive of her mother as a thief, for example (although not, I suppose for her to imagine her mother being a thief, or to write fiction about her mother stealing). Second I need to apologise for the sentence “We submit that we must mentally represent them in a way that coheres with our knowledge of (at least) other a priori truths.” As Ellen, and the others, point out the notion when end up with requires that something cohere with a posteriori truths. Now the phrasing there is my fault, not Liz’s, I was trying to gradually strengthen the notion of conceivability but clearly I have failed to communicate that, so yes it is important to note that ultimately the notion of conceivability we want requires coherence with a posteriori truths.

    Now this is a very strong notion of conceivability, but one, we think is that necessary in order to be charitable to Chalmers. I will explain why I think this using a couple of examples and Liz will add some further reasons later on.

    Ellen gives us two examples of what is conceivable in the weak sense but not our strong sense. These are that light might travel at faster than 1 billion metres per second and that her mother is a thief. Now I would certainly grant that such situations are imaginable. Being imaginable, however, is not useful for answering questions about consciousness or any other aspect of the world. If it were enough then we could do physics by imagining faster than light light, or put Ellen’s mother on trial for theft because we can imagine her stealing. I think it would be extremely uncharitable to Chalmers to allow his notion of conceivability these implications.

    Ok, but have I missed the point? Clearly, and again as Ellen has noted, I am interested in the actual world, whereas Chalmers’ interest is possibility. However, the principle of charity requires that we grant that Chalmers wants his argument to have so implications about the actual world. Not, of course, that Zombies could exist in our world, but that the concept of consciousness at play in the Zombie thought experiment be informative as to the nature of consciousness in the actual world. After all, if it were not so informative then we would wonder what the point of making the argument could be. So, then, this is why the stronger notion of conceivability is appropriate: it limits the conceived to what can be informative about the world.

  2. I want to add to Glenn’s thanks. It is a really great conference Richard and your thoughtful comments are really helping me think through this work in progress, Janet, Jennifer and Ellen.

    first a few clarifications:

    I agree with Jennifer that this argument does not invalidate the use of these intuitions in other ways, as we point out we are not even challenging their use in arguments for dualism or mysterianism. I also agree with Janet that there are other ways to challenge the argument for a hard problem and that questions about the relation between conceivability and possibility are important. But we have a much more modest aim in this paper, namely to undermine the support for the claim that the phenomenal quality of experience cannot be explained by structure and function. There is no evidence that there is a hard problem of consciousness.

    I want to acknowledge that I share Janet’s (in particular) skepticism about the move from conceivability to possibility. In relation to this she accuses us of preaching to the converted and ignoring the agnostics. I think that the more lines of attack that are mounted against a position the better and would not want to claim that the approach she favors is unimportant. But I deny the claim that our approach is obviously inferior. To trade in analogies, isn’t it worth trying to beat them at their own game than change the rules altogether?

    Now for the Interesting bit and the place where we air our disagreements about our paper in public (gulp!) because my response to Ellen’s worry is quite different to Glenn’s. Please bear with us as you get to see some of the processes behind producing a collaborative piece

    There is a general concern that our argument relies on us taking, as Ellen puts it, a too strong notion of conceivability, namely one that includes both a priori and a posteriori truths. We agree that this is not the notion of conceivability that Chalmers works with, but I do not think that it is the one that we are working with either (or perhaps more precisely that we need to work with).

    Our key point is that the intuition is being used as a premise in an argument. In this case an argument that the phenomenal qualities of experience cannot be explained by structure and function. Given this it is appropriate to ask what evidence do we have that the intuition is true. Now one reason, quite possibly the one that Chalmers’ would have at least initially appealed to, is that the intuition is self-evidently true. However, this is far from the case. As anyone who has taught this material can appreciate, someone need not have any strong pre-theoretical commitments to have trouble imagining that the phenomenal quality of experience makes no difference to one’s behavior. And as Janet points out many philosophers do not feel the pull of this intuition either.
    Given that the intuition is not self-evidently true, we need evidence to support its truth.(I had thought) our claim was that the only other evidence that could be provided, a priori, is a belief that the phenomenal quality of experience cannot be explained by structure and function. So to use Fridland’s analgoies, we are not saying that because we know a posteriori, that her mother is honest, it is inconceivable that she is a thief. Rather we are saying that Chalmers’ argument is like saying that because it is conceivable that her mother is dishonest we are justified in convicting her of robbery. Or to take the case of gravity, the analogous argument would be to argue that tables are unneccesary for holding up cups because it is conceivable that gravity does not exist.

    i want to note that I might ultimately agree that Glenn has the better notion of conceivability, but as I have already said, I think a better argumentative strategy is to accept the opponent’s rules and try and beat them at their own game …

  3. Here are some thought’s on Levin’s reply. I am worried that we may start talking passed each other, because we think that the advocate of the hard problem needs to use different notions of conceivability. Levin, I take it, holds that the hard problem advocate can work with a weak notion of conceivability, whereas I suspect they need a stronger notion. With that in mind here are a few comments aimed at trying (I’m not convinced succeeding!) to get us on the same page as to just what our argument is:

    Levin argues that it is still possible to be a materialist and hold each of the key intuitions. In particular she suggests that one can be a materialist, or take it that the hard problem is in principle soluble, and still find Zombies conceivable. There are a couple of things to note about this point. First, in the initial presentation we do take a Dennettian line and could be interpreted as saying that only dualists have the key intuitions. However, this is not what we were trying to communicate (again I think this is a problem of my writing not Liz’s!). The issue we want to highlight is not so much the having of the intuitions (indeed people can have inconsistent intuitions) but of being able to justify the truth of the intuition. What we were trying to communicate was the difficulty the hard problem advocate has in doing this without appealing to the hard problem.

    Further it is right that one can be a materialist and hold the zombie intuition if we use the weaker notion of conceivability, such that conceivability is akin to simply imagining- which does seem to be Levin’s position on p.3. However, on the stronger notion of conceivability, which as I claimed above, I think we should be using in order to allow the possibility that the Zombies argument is informative, it is not necessarily (depending on how the world turns out to be). So on the strong notion of conceivability the claim that Zombies are conceivable needs some justification. If we want to say the conceiving just is imagining that is fine, but I think it precludes the possibility of the Zombies argument being informative, so to be charitable to hard problem advocates we should use the stronger notion.

    Levin finds the Zombies argument unconvincing because the she does not find the link between the conceivability of Zombies and their possibility convincing. Although she does hold that conceivability is “in general good evidence for possibility”. This I take it is quite a common worry regarding the Zombies argument; indeed Chalmers spends time over several publications trying to establish the link between conceivability and possibility. Like Levin I find this argument unconvincing (likely because I prefer scientific realism to modal epistemology). However, what we have aimed to do in the paper is not deny this reply to Chalmers, but highlight an additional worry.

  4. Thanks Liz and Glenn for your responses. I must say, I’m enjoying doing this online, but a part of me wishes we could do it Berlin.

    I’ll respond to Glenn and Liz independently, since they each had a separate response to my comments. Though, I’m both outnumbered and flattered that they took my comments so seriously, I’ll try to make some useful points.

    Again, I, like Glenn prefer scientific realism to modal epistemology, but since we’re already arguing about zombies, we should take seriously what they could mean.

    Glenn: I like that you bite the bullet on this one and say that it is inconceivable that my mother is a thief. I think she’ll be happy to know that at least someone is defending her honor (even if it’s not her own daughter). That said, I wonder how many other useful conceptual tools we have to throw out, if our notion of conceivability is this strong. Can we use counterfactuals about the past? Can we, for instance, say: I’m really lucky that I took the tram and not the U-Bahn because otherwise I would have gotten stuck in the protest. I know that I did not take the U-Bahn. On your strong notion of conceivability, is it at all informative to say that I may have? I guess counterfactuals about the future are less problematic, but maybe not: can I say (supposing that I’m a medical Dr.) that if you exercise 5 times a week you will have a strong heart even if I know (say you’ve been my patient for a long time) that there is NO way that you are going to exercise that many times per week. Does all of this get thrown out as informative with your strong notion of conceivability? I’m not sure that it has to, but if it does, I would claim that we have a pretty strong prima facie case against using such a strong notion of conceivability.

    Also, if you want to use such a strong notion of concievability, I think you need to provide some argument for it *in the paper*. That is, I think you need to give an explicit defense of why a posteriori truths must be taken into account by any notion of conceivability.

    In your response to me, you say that a strong notion of conceivability is necessary in order to be charitable to Chalmers, that is, to see how anything that comes out of zombie thought experiments could be informative about consciousness in the actual world. But, I’m not sure. The Chalmers argument doesn’t proceed from the intuition about dualism to dualism is true. It is more subtle than that. The idea is that dualism is conceivable, if it is conceivable then it is metaphysically possible, if it is metaphysically possible then it tells us something about explanations of phenomenal properties in terms of structure and function. I guess I just don’t see why this strong conceivability is required for Chalmers to get to his conclusions. Of course, that doesn’t mean that it isn’t required, I just don’t yet see why it is. I think Liz is making a similar point to yours here, so I’ll switch to her comments now.

    Liz: First off, I guess I’m not sure that the hard problem is seeking to establish the truth of (some people’s) dualist intuitions. The point is not that the intuition is true, but the fact that we can have it tells us something about what is possible. Maybe your point is that if the intuition isn’t true then one shouldn’t be able to proceed to any further argument and maybe that’s right. This is curious. I guess I’m just not sure what truth-value intuitions need in order to do the work for Chalmers. Perhaps they function like hopes or wishes and thus, need not be either true or false in order to be relevant. I’m just not sure about this.

    Okay, on to your examples about how you see Chalmers argument as proceeding and thus, how you take your response to it as going: The examples are helpful and they show that you and Glenn are really on the same page about this. But I think that the analogies don’t work: you say, Chalmers argument is “like saying that because it is conceivable that her mother is dishonest we are justified in convicting her of robbery.” That would mean that Chalmers argument proceeds from “zombies are conceivable to consciousness cannot be explained by the structure and function of brain states.” But there are missing premises here. There is all the stuff about metaphysical possibility and what that means for explanation. Though I don’t think you’re being fair to Chalmers, I do think that you’ve hit on something very important. Not in the realm of conceivability, but in terms of the relation between metaphysical possibility and explanation in the actual world. You say that Chalmers argument is like saying that, “tables are unneccesary for holding up cups because it is conceivable that gravity does not exist.” I think this where the interesting stuff happens. You could say that though there may be a possible world where in fact it is true that the table is unnecessary to explain the cup not falling, it is much too quick to then transfer from that possible world, to our world, the notion that the table holding up the cup is explanatorily impotent in accounting for why the cup is being held up on my table now. That is, it is unjustified to claim that for explanations to be informative in this world, they must hold in every possible world. I think this strikes at the heart of the hard problem, but, of course, this then is not about conceivability and question begging nature of the intuitions behind zombies, but about the nature of explanation.

  5. Thanks so much to both Glenn and Liz for your comments! One clarification, though, about my views about the conceivability of zombies. I may have misstated things in the commentary, but I think that zombies remain conceivable even in the face of increasing knowledge of the structure and function of the brain and central nervous system (in contrast, for example, to what happens to the ‘intuition’ that water can be conceived in the absence of H2O (or vice versa) when we learn the relevant facts about the microstructure of water). I’m not sure this counts as a ‘weak’ notion of conceivability, since it’s not just that the zombie scenarios are initially imaginable or conceivable, but that they remain so when confronted by empirical information about the world. To be sure, we don’t (yet) know everything about the structure and function of the brain–and some people invoke this relative ignorance to explain the distinction between intuitions in the case of water and H2O, on the one hand, and conscious experiences and neural states on the other. But is it plausible that learning more truths such as the ones we now accept will help (given that invoking methodological considerations that rule out dualism shouldn’t count)? And if not, it seems that it’s best to have an argument that conceivability does not entail possibility (at least in conscious experience-neural state cases) waiting in the wings.

  6. This is great, everyone, you are certainly giving us lot’s to think about. Also I need to apologise to Jennifer for not replying to your commentary yet, sorry- no judgement on you I promise! I will get to it ASAP (for the record Liz has sprained her left ankle and burst a blood vessel in her right foot- feel free to laugh at her- which has rendered her almost completely immobile, so I havn’t had as much time as I would have liked to work on this over the weekend).

    @Ellen point taken on arguing for the necessity of the stronger notion in the paper, I’ll start developing an argument!

    @Janet I liked your claim that you think Zombies will remain conceivable no matter how much we know about the mind, this clarifies for me why you would want to hold onto the intuition despite being a materialist. However, I don’t think it would be possible for an advocate of a hard problem to extend this reasoning to argue for the existence of a hard problem. Let me explain:

    I think there is no way to predict what would remain conceivable given some generally accepted explanation of consciousness. If it turned out, for example, that Dennett was right and consciousness just is ‘fame in the brain’ then we couldn’t conceive of those processes of access, ie, ‘fame’ without consciousness anymore than we can conceive of a body of H2O molecules in close proximity at room temperature without water. I’m not saying that this is the right account of consciousness, just that we know of at least one account which, if accepted, makes zombies inconcievable. Let me grant for the sake of argument that Zombies may or may not remain conceivable.

    The argument for the existence of the hard problem, depends on Zombies being conceviable. But, we’ve just reasoned that they may or may not be conceivable depending on what consciousness turns out to be. Now if this is right then it seems we would want some reason to take a side on conceiveability and not just relay on pretheoretic intuitions (which seem to vary between people- who has the right intuition?).

    This then brings us back to the problem we raised in the paper, one could attempt to argue that Zombies are inconceivible by, say, giving the evidence for an identity between consciousness and fame in the brain. But, what argument could be given that Zombies are and will remain conceivible? It seemed to us, in the paper, that there was no way to provide such an argument that didn’t depend on assuming there is a hard problem.

    So even if we grant that it seems Zombies might remain conceivible no matter what we learn about consciousness I don’t think this is enough to get us to the existence of a hard problem.

    cheers
    Glenn

  7. Here are some thoughts on Jennifer’s reply.

    Jennifer’s core claim is that we havn’t shown that the knowledge argument for a hard problem of consciousness is circular or at least not circular in a damaging way. She offers several other interpretations of the knowledge argument that avoid the worry we raise.

    She says that: “This argument [Mary] would only be question begging if the intuition that Mary learns something new when she leaves the room were not consistent with other explanations that do not entail a hard problem.” In other words the knowledge argument isn’t circular if it can be true that there is no hard problem and Mary learns something new.

    And then “But, in fact other hypotheses have been offered that would explain the key intuition that Mary learns something new upon seeing red that are also compatible with physicalism”

    But there is a slip here between two uses of the knowledge argument. It can be used as an argument for the hard problem, or, with some additional considerations (i.e. that the hard problem leads to dualism) as an argument for dualism. It is only the former which interests us here.

    I think I agree with both things Jennifer says, but I don’t think she has shown that Mary can learn something new without there being a hard problem. To argue for this she reminds of us the ‘ability’ and ‘old fact’ replies which are physicalist responses to the knowledge argument. However, my reading of these replies (and it may be more useful for Liz to chime in here) is that these replies accept the existence of a hard problem, but deny that this leads to dualism as the hard problem is soluble. We’d like to get of the train one step earlier and deny that the knowledge argument gives reason to believe there is a hard problem, regardless of whether or not acceptence of a hard problem leads to dualism.

    I think Jennifer is right that because of these replies we cannot conclude that the knowledge argument begs the question for dualism. But, physicalists can accept that there is a (solveable) hard problem (as Ellen and Janet seem to) the issue of the existence of a hard problem is not the same as that of dualism. To show that the knowledge argument doesn’t beg the question for there being a hard problem it is thereby insufficient to show that it doesn’t beg the question for dualism.

    Jennifer’s next argument is to present the knowledge argument in a way such that the hard problem does not appear in the premises. She says:

    1. Mary knows all the physical and functional facts about colour perception
    2. Mary seems to learn something new when she first sees red
    3. It seems Mary’s new knowledge concerns what it is like to see red
    c. So consciousness cannot be expained just by an appeal to structure and function

    I worry that there are two things wrong with this presentation. First, premise 2 appears too weak to support the conclusion as it stands. Premise 2 for example states that it SEEMS Mary learns something new. Given this premise shouldn’t the conclusion be “so IT SEEMS consciousness cannot be explained just by an appeal to structure and function”. Now this looks very nitpicky and trivial but I think I hides exactly the slip in the argument we were trying to highlight in the paper. To get the conclusionas stated in c, we need a different premise namley:

    2a: Mary learns something new when she first sees red.

    When we put the premise is this form the second problem with the argument becmes apparent. Whilst we may not need to provide an argument for premise 2, thats just about seeming and I have no grounds to deny that it seems to Jennifer or anyone else that it seems to them that Mary learns something new. But premise 2 doesn’t support c. But when we replace 2 with 2a we are left with a claim that more transparently needs support, namely the claim that Mary learns something new. This is what worries us, after all what reason can be given to conclude that Mary learns something new? This then is where the question begging nature of the argument arises as we initially identified it in the paper.

    In the paper we also used the variability of intuitions about Mary to support the idea that the intuitions are theory dependent. But, as Jennifer points out, reasons aside from holding different theories can explain some of the variation. I think Jennifer is right that we can’t use variation alone to argue for theory dependence, but I think the other use we make of this variability, namely to suggest that the advocate of the hard problem ought not assume their intuition is right and should argue for it still holds despite this claim.

    Finally Jennifer points to a third use of the knowledge argument when she says:

    “Another constructive way of understanding thought experiments such as the one involving Mary is to see them as tools that help us unpack implicit commitments, better understand the nature of our concepts, and flesh out out the logical relations between our ideas and their entailments”

    Now of course- as Jennifer recognises, but I will make explicit anyway- this only tells us what we think, implicitly, about consciousness not what consciousness is. This is fine and worth doing, but why is it worth doing? The value of this exercise is to make explicit hidden commitments about consciousness. Once these commitments are made explicit then we can attempt to determine their truth. We ought not assume that they are true without argument.

    Lots of fun so far ;0

  8. thanks to glenn and liz for their interesting paper, and to janet, ellen, and jennifer for their commentaries. since i’m the main target of the paper i thought i’d offer a few thoughts.

    my main reaction is that there’s a misinterpretation at the heart of the paper. glenn and liz take me as arguing that there is a hard problem of consciousness by using zombie arguments, the knowledge argument, and so on. but i never argue in this way. in my 1995 paper on the hard problem of consciousness i introduce the problem without ever mentioning zombies or the knowledge argument. in my 1996 book i use zombies and the knowledge argument to argue against materialism in chapters 3 and 4, at a point where i’m already taking for granted that there is a hard problem. (the “hard problem” makes only a very brief appearance under that name in the book, in the introduction, but in effect the problem is at the heart of the introduction and chapter 1.)

    so when glenn and liz say that these arguments against materialism don’t support the claim that there is a hard problem, i’m more or less happy to agree. i think it is more obvious that there is a hard problem of consciousness than that zombies are conceivable and the like, so i wouldn’t argue for the former using the latter. insofar as i have an “argument” for there being a hard problem at all, it is the argument in “facing up to the problem of consciousness” about there being a further question after explaining functions. but if someone wants to say that this argument too is circular, that’s ok with me.

    i think it’s obvious to most people that there is at least a distinctive problem of consciousness. i say in the introduction to my 1996 book that if someone doesn’t agree, then after a point there’s not much i can do to persuade them. but even dan dennett in some moods allows that there is a hard problem. he just thinks that it can be solved in a deflationary way. the real action comes when we argue over how and whether it might be solved — for example, whether there might be a neurobiological or some other materialist solution to the hard problem. the role of zombie arguments and the like is to make the case that there cannot be.

    now, i’m sure that glenn and liz object to zombie arguments and the like even construed that way. if the objection is just that the arguments presuppose that there is a hard problem, then i won’t argue too hard (though i think the arguments can be construed in a way that doesn’t require an explicit presupposition about this). on the other hand, if the objection is that the arguments presuppose that materialism is false, or that there is no physical explanation of consciousness, and so on, then i think this is obviously incorrect. its incorrectness is brought out by the legions of people who (like janet and many respondents to the philpapers survey) hold both that zombies are conceivable and that materialism is true. maybe these people are wrong about zombies, but if so it presumably isn’t because they are presupposing the falsity of materialism.

    i also note that one can argue against materialist explanations of consciousness without ever mentioning zombies, mary, and the like. my 1995 paper adopts an argument strategy of that sort. those arguments might be a useful focal point for those (like ellen) who are skeptical of an overly strong link between conceivability/possibility and explanation.

    a smaller point regarding the discussion of conceivability. there are lots of things that one might mean by “conceivable”, but i’m pretty explicit that the sense relevant to my own argument is tied to apriority — roughly, p is conceivable when ~p is not a priori. in this sense, ‘water is not H2O’ and ‘light travels faster than one bill meters per second’ are pretty clearly conceivable — there are other senses in which it is not but these aren’t the senses relevant to my argument. of course glenn and liz can then feel free to argue that one can’t draw conclusions about metaphysical possibility from this sort of conceivability (contra my claims in e.g. “does conceivability entail possibility”). but as far as i can tell this would require arguments distinct from any they give in their paper.

  9. Thanks David for your thoughts, I have a few things to say by way of reply and I hope Liz will join in later too.

    A clarification first:

    Yes, the claim is just against the use of these arguments in support of a hard problem, not as arguments against materialism (not that these must be independent concerns, but we can only be taken to be saying something about the later if it is dependent on the former, which we havn’t tried to show). I do think there are premises which need further justification in order to argue against materialism, but that’s neither here nor there for minute.

    I don’t think you need to convince either of us that there is a distinctive and exciting problem of consciousness. Or indeed many of them. But, I take it that the hard problem is more than this. The hard problem seems to me to be the problem that a certain class of potential solutions to this distinctive problem (those on which consciousness is dependent on structure and function of mental states) are not open. Argument is needed to close off these options and it certainly won’t do to show that particular structures and functions of mental states are insufficient a very general argument is needed.

    I am concerned that David claims never to have used the thought experiments as arguments for a hard problem. Isn’t this the strategy in section 2 of “Chalmers, D. J. (2002). Consciousness and Its Place in Nature. Philosophy of Mind: Classic and Contemporary Readings. D. J. Chalmers, Oxford.”? We should take the authors word that it is not. So then what is the strategy of this section? It seems to push beyond a rehearsal of some familiar arguments against materialism. Another possible reading is that this is intended merely as a characterisation of the problem and the thought experiments are illustrations to aid understanding- in which case there is no argument put forward that consciousness is independent of the structure and function of mental states. David I think your comments above suggest that this was your intention?

    So, then, why should we believe that there is a hard problem? As David notes he has given some further reason in “facing up…” where he tries to highlight that there is something left over after an account of the function of mental states, using the 40Hz hypothesis as the main example of a functional account. There is some worry here that this doesn’t speak to all of the hard problem- which seems to involve the claim that consciousness is independent of the function AND structure of mental states. Perhaps we could take the 40Hz hypothesis as a structural or identity claim. This would be reasonable, but then we are still left with the problem that the reasons for believing there is a hard problem, at least as they are presented in “facing up…” are centred around a particular (and not very good- indeed much weaker than contemporary accounts, like Austen Clark’s quality spaces) explanation. Of course if we look at the 40Hz hypothesis we don’t have an account of consciousness- it hasn’t identified that right structures and functions!

    The advantage of reading the thought experiments as arguments for a hard problem is that they are properly general arguments. Perhaps we have taken this consideration too seriously and so misinterpreted David’s claims. If this is so then the worry is there is no argument for the existence of a hard problem, i.e. no reason to think that consciousness is independent of the structure and function of mental states.

    In closing I am tempted to ask the question again. What reason can be given to rule out an entire class of potential solutions to the distinctive problem?

  10. hi glenn. section 2 of that paper (http://consc.net/papers/nature.html) does exactly what it seems to — introduces the hard problem and says why it is hard. the thought-experiments are not even mentioned in that section. they’re discussed in section 3, which contains arguments against materialism. as for 40hz (in http://consc.net/papers/facing.html), this is supposed to be a specific instance of a general point (the general point comes in the previous section) — that for any cognitive function one might explain, or indeed for any brain structure one might explain, one can always raise the further question: why is that function (or that structure) accompaned by consciousness?

  11. Thanks for your reply Glenn and thank you both for your paper. Read ‘physicalism’ in the line, “But in fact other hypotheses have been offered that would explain the key intuition that Mary learns something new upon seeing red that are also compatible with physicalism” as referring to the epistemological thesis that its all explainable in physical and functional terms, rather than the metaphysical thesis. I don’t see that the ability hypothesis necessarily suggests a hard problem. Think of the knowledge that Mary gains in experiencing red as analogous to the knowledge possessed by a Kung Fu master. Mary may have perfect knowledge of how the body moves and of all Kung Fu techniques but we may have the intuition that upon leaving her house and studying with a teacher she gains a kind of knowledge that was not previously possessed; the ability to do Kung Fu. You might not expect her to get the ability by just having the knowledge about human kinesthetics, martial art technique, neuroscience etc. But I do not see why all of her ability couldn’t be explained in physical/functional terms.

  12. Hi everyone really interesting session and discussion!

    I am running around out here like a bat with its head cut off but I couldn’t resist jumping in at this point.

    What is the Dialectic?
    It seems to me that perhaps Glenn and Dave are slightly talking past each other here. Glenn thinks that the zombie argument is used as an argument for the existence of the hard problem, Dave says it isn’t. But isn’t it just the case that, as Dave says, the zombie argument is invoked to try to show that any given physicalist attempt to solve the hard problem will fail? If so then it is used as an argument for the hard problem in the sense that it is an argument that there cannot be a physicalist solution to the hard problem…if one were a physicalist then one could rightly regard that as the challenge that needed responding to. How does one show that there isn’t a hard problem? By showing that the intuitions that lead to it are suspect.

    Another way to put it might be as follows.

    1. [most? some?] people have an intuitive sense that consciousness isn’t explainable by appeal to function or structure
    2. But why think this shows there is a hard problem? Maybe it can be explained by appeal to Clark’s notion of quality spaces (or whatever). Well, because [insert zombie argument; i.e. because in principle any physicalist argument will fail]

    Without (2) all you have is a claim that it seems to you (or to one, or it seems to us now) that there is a hard problem, not that there actually is one. So, don’t you (Dave) need the zombie arguments to establish that there really is a hard problem as opposed to there merely seeming to be one?

    Explanatory Success and the Conceivability of Zombies

    Dave says,

    if the objection is that the arguments presuppose that materialism is false, or that there is no physical explanation of consciousness, and so on, then i think this is obviously incorrect. its incorrectness is brought out by the legions of people who (like janet and many respondents to the philpapers survey) hold both that zombies are conceivable and that materialism is true. maybe these people are wrong about zombies, but if so it presumably isn’t because they are presupposing the falsity of materialism.

    Isn't this too quick? Doesn't Janet think that 'there is no physical explanation of consciousness' even though she is a materialist? If so then it could still be argued that conceiving of zombies depends on thinking of consciousness as already failing to be susceptible of explanation via physical routes.

    If so then we can take Glenn and Liz as making the argument that those who say that zombies are conceivable are already assuming that no physicalist theory could ever suffice as an explanation of consciousness (and not merely that it seems now to be the case) as opposed to those who think that the conceivability of zombies shows that no physicalist theory could suffice. Dave, you wouldn't accept this would you?

  13. I wonder how many here would believe that the hard problem was solved if it were successfully demonstrated that the biophysical structure and dynamics of a theoretical brain model could predict novel phenomenal events as well as previously inexplicable conscious content.

  14. Glenn,

    Would you please be able to clarify for me what you mean by “mental” in your phrase “consciousness is dependent on structure and function of mental states” above? How does “mental” differ from “conscious” for you here? Would you give an example of a structure of a mental state in the sense you mean?

  15. For Richard: a comment and a question:

    (1) I endorse the claim that ‘there’s no physical explanation of consciousness’ only when ‘physical explanation’ is understood as requiring an priori entailment of the ‘what it’s like’ of experience, so conceived, from a physical description of the world (and not that there’s no necessary connection between conscious experiences and physical states). Perhaps that’s what you meant, but the conceivability of zombies and the like, e.g. disembodied minds (in my view) motivates the claim about explanation, understood in the first way.

    (2) What is it like to be a bat with its head cut off?

  16. richard: it’s not true by definition that the hard problem (if it exists) doesn’t have a materialist solution. lots of people think there can be such a solution to the problem — i think they’re wrong, but they’re not misusing the term “hard problem”. still, we can set aside this verbal issue by defining “very hard problem” (VHP) as “physically unsolvable hard problem”. in these terms, you’re right that the conceivability of zombies can be used as one argument for a VHP (in fact i use it as such in chapter 3 of my book), though there are other arguments for a VHP that don’t go via zombies (e.g. in the 1995 paper).

    on your main point, while i and some materialists think that the conceivability of zombies entails that there is no physical explanation of consciousness, plenty of materialists don’t, holding both that zombies are conceivable and that consciousness can be physically explained (i.e. denying the connection between explanation and a priori entailment). i think that’s enough to bring out that the conceivability of zombies can’t wholly turn on the presupposition that consciousness can’t be physically explained.

  17. Hi all thanks again for your thoughts. I must apologise for missing yesterday, Liz’s ankle injury is worse than initially feared (she has torn a ligament) so we spent the day in doctor’s offices.

    Arnold asks an interesting question that goes to the nature of the hard problem and I think there are two things to say about it. If we can predict new phenomenal states does that solve the hard problem? Whilst being able to predict new phenomenal states, or just what state one will be in given a certain stimulus, is an important part of a theory of consciousness, this kind of prediction isn’t sufficient for explanation. Good predictions can be formed from good descriptions of phenomena.

    Second even if there is an explanation of consciousness, does that qualify as a solution to the hard problem. In one sense of course it does if the hard problem is just the problem of explaining consciousness. However, in consciousness and it’s place in nature, David says this:

    “What makes the hard problem hard? Here, the task is not to explain behavioral and cognitive functions: even once one has an explanation of all the relevant functions in the vicinity of consciousness — discrimination, integration, access, report, control — there may still remain a further question: why is the performance of these functions accompanied by experience? Because of this, the hard problem seems to be a different sort of problem, requiring a different sort of solution.”

    Now despite my rather embarrassing tendency to get the labels of the sections wrong, I read this as suggesting that the hard problem isn’t just the problem of consciousness but the conjunction of the problem of consciousness and the claim that a certain class of possible solutions are not open.

    In this sense then no materialist explanation of consciousness would count as a solution to the hard problem, but rather a denial of the existence of a hard problem. If this is what the hard problem is then Richard is right to characterise the debate as one between those who think there merely seems to be a hard problem (e.g. Liz and I) and those who think there really is one (David). I think this is a very helpful characterisation. But then David says in response to Richard that:

    “Richard: it’s not true by definition that the hard problem (if it exists) doesn’t have a materialist solution. lots of people think there can be such a solution to the problem — i think they’re wrong, but they’re not misusing the term “hard problem”.”

    which suggests that the hard problem just is the problem of consciousness and the claims about what makes the hard problem hard are not essential to its formulation. Alright, so what’s the upshot of this. Perhaps we have only been talking about what David has now called “the very hard problem”. This, I think, is ok except that it’s not so easy for Liz and I to say we’re not talking about materialism, rather in talking about the hard problem we are talking about arguments against materialism. However, whilst it was a rather stupid mistake on my part to say, above, that we weren’t interested in such things it is still true that we’re not talking about the use of these arguments is arguing for any particular form of dualism.

  18. @drbillh

    So by mental I mean the states (representing vehicles) and process (computations) that perform tasks typically classified as mental, e.g. learning, perceiving, inhibition of inappropriate responses. These can be either conscious or unconscious.

    Some of the functions of a state that might make them conscious rather than unconscious are global access/being a dominant draft (Baars/Dennett) or being disposed to cause a higher order thought to be formed (P. Carruthers).

    Structural features that might make a state conscious rather than unconscious might be stability of activation patterns (O’Brien and Opie) or isomorphism between an activation space (each dimension is the activity of a particular unit/neuron) and a quality space (each dimension is a way in which stimuli are discriminable). So by and large you might conclude from this that I suspect that the relevant structures are those which compose particular mental states/representing vehicles.

  19. @Jennifer

    ok so I need to think this through again. I shouldn’t have said in my initial response that:

    “However, my reading of these replies (and it may be more useful for Liz to chime in here) is that these replies accept the existence of a hard problem, but deny that this leads to dualism as the hard problem is soluble.”

    Whilst I think it is true that the main point of these replies is to block the inference to dualism, given my above thinking they do so by denying that there really is a hard problem, but of course accept that there seems to be one to put it in Richard’s phrasing. To use David’s phrasing they would accept there is a hard problem, but deny that it is very hard.

    perhaps then I should have said:

    “However, my reading of these replies (and it may be more useful for Liz to chime in here) is that these replies accept that there appears to be a hard problem, but deny that this leads to dualism as there isn’t really a hard problem.”

    Jennifer, do you think that would more accurately characterise the ability reply?

  20. Hi,
    Sorry for my absence, turns out I am a total clutz. Thanks everyone for the great discussion. I will reply to Jennifer first on the knowledge argument and get back to you in a bit on the general discussion re the hard problem.
    I think that the key to whether the Mary thought experiment leads to the claim that structure and function cannot explain consciousness depends on how the intuition is characterised. If the intuition is simply that she gains new knowledge, then Jennifer is right our worry does not arise because this new knowledge need not be knowledge of a new fact (because as Jennifer points out it is non-factual ability knowledge or because it is knowledge of an old fact). This is why we were careful, at least in our initial presentation, to characterise the intuition as being that Mary learns a new fact.
    Now of course it is an open question as to how to best characterise the Mary intuition and I think that Jennifer is right in that the intuition is usually characterised as being simply that Mary gains new knowledge.
    However I think it is now widely accepted that if the thought experiment is to be used e.g. in an argument for dualism then the stronger claim (i.e. that Mary gains knowledge of a new fact) is required. So I don’t think it is entirely unfair of us to use this stronger claim in mounting our argument.
    But thank you Jennifer for pointing out that we at least need to make this clarification re what the intuition about Mary is.
    Does this satisfy at least some of your worries?

  21. I want to say that I suspect (and hope) that we are not the only people who have made some errors in understanding your position David, so I trust this is of general interest and that you will bear with me while I try and see how the problem hasn’t just melted away.

    I want to push home Richard’s point: “Without (2) all you have is a claim that it seems to you (or to one, or it seems to us now) that there is a hard problem, not that there actually is one. So, don’t you (Dave) need the zombie arguments to establish that there really is a hard problem as opposed to there merely seeming to be one?”

    The thing I am left wondering after going back and looking at your works again David, is what is the hard problem? Perhaps one way to home in more precisely on what my worry is is this: I believe that consciousness can be reductively (but not eliminatively) explained in terms of the structure and function of mental states (in fact I would make the stronger claim that we can already see how such an explanation could work). Can I believe this and also believe that there is a hard problem of consciousness in your sense? In your response to Richard this is what you seem to be saying.
    But if I can hold both beliefs then to be honest I (like many people before me) don’t see what all the fuss is about. Because if I can hold both then it seems to me that the claim that there is a hard problem is nothing more than the claim that we don’t at the moment understand how an explanation of consciousness in terms of structure and function would work. If so then isn’t an appropriate response simply: “give us some time, we are working on it”.

    One reason why I had always thought that you were a supporter of a stronger reading (along the lines of consciousness cannot ever be explained in terms of structure and function) is that it seems to me that the weaker claim (i.e. that we can’t now see how an account in terms of structure and function could work) is insufficient to support the second premise in the explanatory argument in Consciousness and its Place in Nature. If the hard problem is simply that we don’t understand at the moment then how can we, from this ignorance, draw the strong conclusion that it can never be understood in that manner? This is a point the Churchlands in particular have often made in response to your work (from our ignorance all we can conclude is that we are ignorant) and I had always thought they were slightly missing the point because I had thought you were making the stronger claim. Now it seems that they might have been right all along.

    [Just to be (painfully) clear I am not saying you haven’t provided other good arguments for the strong claim that a materialistic account of consciousness is insufficient. I am now just worried that the hard problem is not a good premise in any such argument.]

    I suppose to sum up I am now beginning to wonder whether in trying to take the claim that there is a hard problem of consciousness seriously and pin down an argument to support it we may have in fact uncovered how trivial/uncontroversial the claim really is: We don’t have an explanation so we can’t understand how an explanation could work. So?

  22. thanks, liz and glenn. i’d thought my 1995 and 2002 articles were clear about these things, but obviously they aren’t as clear as they could have been, so i’ll try to be really explicit.
    the hard problem as initially characterized is just the problem “explain experience” or in more detail, “explain how physical processes give rise to phenomenal consciousness”. but what does the key work in the argument is the claim that the hard problem is distinct from the easy problems, which are defined as the problems of explaining various (behavioral/cognitive) functions: discrimination, integration, report, and so on. if you want to see this as built into the definition of the hard problem — that the hard problem is by definition distinct from the easy problems — then i don’t mind going along with that. we might call the notion so defined the distinctive hard problem (DHP) — i.e. the hard problem distinct from the easy problems. if you like we could also allow here that the easy problems include the problems of explaining various physical structures (e.g. explaining neural architecture), so that the DHP is by definition a problem distinct from those of explaining functions and explaining structures.

    so: if someone says that there is no problem of explaining consciousness distinct from the problems of explaining discrimination, integration, and report, and so on, then we can take them to be denying the existence of DHP. for example, if someone says: to explain consciousness, we just need to explain discrimination — the problem of consciousness just is the problem of explaining discrimination and there is no interesting further problem — then they deny the existence of a DHP. for present purposes we can also stipulate that the claim that there is a DHP builds in the claim that the phenomenon to be explained exists — so if someone says “sure, there is an apparent further *problem* of explaining experience, but the solution is to deny that experience exists”, then they deny that there is a DHP.

    the claim that there is a DHP does *not* rule out many nearby physicalist views. for example:

    (i) the view that consciousness is identical (a posteriori) to some physical structure or to some cognitive function. it’s quite consistent to say that consciousness = F (a posteriori) while holding that the problem of explaining consciousness is distinct from the problem of explaining F. compare: even if heat = molecular motion, the problem of explaining heat is distinct from the problem of explaining molecular motion — to do the former one has to do a lot of thermodynamics, to do the latter one just needs low-level statistical mechanics. problems are individuated epistemologically, not ontologically.

    (ii) the view that consciousness can be explained in terms of structure and function. someone might say: i agree that the problem of explaining conscious experience is distinct from the problem of explaining the various structures and functions, but i think one can explain it in terms of structure and function all the same. i think those people are wrong, but they aren’t contradicting themselves.

    (iii) various other views that connect consciousness and function — e.g. the claim that consciousness has a function, or the claim that explaining functions will play a central role in explaining consciousness. those views are quite compatible with the claim that there’s a DHP.

    what does the key work for me, then? there are really three premises (as laid out in the 2002 paper). first, the claim that there is a DHP. second, the claim that all physical explanations are explanations in terms of structure and function. and third, the claim that structure and function explain only structure and function. it’s the first and third claims together (and not the first claim alone) that rule out the views like (ii) above according to which one can explain consciousness wholly in terms of structure and function.

    the most fundamental premise here is the first, that there is a DHP. i’ve always taken this premise to be obvious (which is not to say that no-one will deny it). the catalogs of phenomena in the mind that need explaining include discrimination, integration, report, various other cognitive/behavioral functions, *and* conscious experience. i think that to deny that there’s something else that needs explaining here is just to deny the manifest. but i’ve always allowed that i don’t have a great deal of *argument* for the premise — i can motivate it by rehearsing various familiar ways of pointing to the further explanandum, but if someone isn’t moved by any of this, there’s not much i can do to convince them. that’s the point of the “great divide” discussed in the introduction to my 1996 book. still, i’m reassured by the fact that a pretty significant majority of people inside and outside philosophy seem to accept that there is a DHP. and from there (with the ancillary premises above) i think the rest of the claim about explanation follow.

  23. Thanks Liz, yes that helps. But I’m not sure that we need to have the intuition that Mary learns a new fact in order to for the thought experiment to be used to support the hard problem. This is what I had in mind in my second comment on the PDF where I present the argument for the hard problem which I characterize as either incomplete or weak but not question begging. That argument could be strengthened or completed by adding a premise that links Mary’s new knowledge to knowledge of a new fact. Could that premise be satisfactorily motivated? I don’t know.

  24. Dave you say,

    on your main point, while i and some materialists think that the conceivability of zombies entails that there is no physical explanation of consciousness, plenty of materialists don’t, holding both that zombies are conceivable and that consciousness can be physically explained (i.e. denying the connection between explanation and a priori entailment). i think that’s enough to bring out that the conceivability of zombies can’t wholly turn on the presupposition that consciousness can’t be physically explained.

    Is this all it takes? I would have thought it would need to be something like that they can hold that even on idealized reflection without contradiction. Or do you think this kind of variation would be around even in idealized conditions?

  25. David: “… for any cognitive function one might explain, or indeed for any brain structure one might explain, one can always raise the further question: why is that function (or that structure) accompaned by consciousness?”

    Could this question be validly raised if activation of that brain structure and its intrinsic dynamics were to *constitute* consciousness?

  26. David wrote:
    even if heat = molecular motion, the problem of explaining heat is distinct from the problem of explaining molecular motion — to do the former one has to do a lot of thermodynamics, to do the latter one just needs low-level statistical mechanics. problems are individuated epistemologically, not ontologically.

    Are you saying that ‘explains’ induces a nonextensional context? I would resist that. If I have explained how a gases temperature has gone up, I have also explained how the energy of the molecules in the gas has gone up (even if I don’t know I have done this, it follows from the identity, no?).

  27. I may be missing something here, Dave, but the analogy between heat=molecular motion and qualitative states=structure/function of mental states doesn’t really seem to hold. After all, I’ve always taken the hard problem to arise because even after we explain everything easy, we can still ask, “why?!” “Why do these easy states feel the way they do, or any way at all?” That would be a question about the *connection* between qualitative states and their functional descriptions. I take it that for heat and molecular motion the point is that each concept requires independent explanations. But one wouldn’t be left with a hard problem. Just two different explanations related to one another through identity. It seems to me perfectly plausible to do the same with consciousness: our explanations of phenomenological, qualitative states need not be the same (in language or method) as explanations of the structure and function of brain states, but when we ask of the functional state, “why does it feel like this?” We can say because it is identical to some qualitative state (in the actual world). Again, I think of isomorphism here. Where’s the (very) hard problem? Wouldn’t the hard problem be analogous to demanding that we have another explanation about WHY heat = molecular motion? And who thinks we need *this* kind of explanation?

  28. Dave, I guess if it really is possible for an ideal agent to deny the connection between explanation and a priori entailment then I would admit that the conceivability of zombies doesn’t wholly turn on assuming that consciousness cannot be explained physically (though it would pain me greatly 🙂

    Even so it seems like we can rephrase the main idea behind the argument. Instead of arguing that the conceivability of zombies depends on thinking that there is a Very Hard Problem they can argue that given certain theoretical commitments conceiving of zombies depends on assuming that there is a Very Hard Problem.

    On a related note do you think the same holds for the case of possibility? Glenn and Liz argue that zombies will only be possible on the assumption that one takes consciousness to have no physical explanation. I take it that their idea is that if the zombie world is really a way our world could be then that must be because consciousness isn’t physical (and hence can’t be explained in physical terms; I take these to be synonymous but I know some people don’t). For instance someone could hold that zombies are conceivable and possible but still be a physicalist because they hold an identity theory and deny the necessity of identity. Is that enough to show, in your view, that zombies being possible doesn’t wholly turn on consciousness being unexplainable/non-physical?

    Ellen, the basic difference is that in the case of heat we start of with something that looks like it can be analyzed in terms of structure and function and then just do the relevant empirical work to figure out what it is but in the case of consciousness it seems like we do not start out with structure and function (this is what the Distinctive Hard Problem really amounts to).

  29. eric: yes, that’s exactly what i’m saying. ‘explain’ is an epistemological verb and epistemological contexts are non-extensional contexts. doing statistical mechanics explains the motion of molecules but doesn’t yet explain heat — to do the latter one has to give some further explanation linking the statistical mechanics to the thermodynamics.

    ellen: i agree that the heat case is disanalogous to the consciousness case in many respects. i was just using it to bring out the limited point that problems and explanatory challenges are individuated epistemologically, not ontologically. of course in the heat case, it’s not all that hard to give the further explanation that links statistical mechanics to thermodynamics — but the further explanation certainly needs to be given.

    people sometimes say that identities don’t need explanation, but i think this is just false. one can give a perfectly good explanation of the identity between heat and the motion of molecules (roughly, by showing how the motion of molecules plays the heat role), between genes and DNA (show how DNA plays the gene role), and so on. in fact those explanations are at the heart of most interlevel reductions. it’s a familiar semantic point that epistemological contexts invoke not just referents but modes of presentation: so to explain X=Y, one has to explain not just how it is that the referent is self-identical (that’s trivial), but how it is that one thing satisfies the modes of presentation associated with both X and Y (that’s often nontrivial, and especially so in the consciousness case).

  30. richard: it’s true that given certain theoretical commitments, the conceivability of zombies *entails* that there is a VHP. in fact i argue for that in my 1996 book. but i don’t think it’s right to say that it presupposes that there is a VHP. after all, many people find it intuitively obvious that zombies are conceivable, whether or not they have antecedent commitments on whether there is a VHP. some people in fact become convinced that there is a VHP by this route. of course others might move in the other direction. but i think one has to be very careful not to slide too quickly from entailment to presupposition here; otherwise every valid argument presupposes its conclusion.

    as for possibility: yes, i think arguments for the possibility of zombies can go via the conceivability of zombies without bringing in any presuppositions about explanation or about ontology along the way. and i think the premises of the argument (zombies are conceivable, conceivability entails possibility) have support that does not require such presuppositions, e.g. that there is a VHP. one might try to argue that the conceivability premise requires the presupposition that there is a DHP — i don’t think that even this is quite right, but in any case that’s a much weaker presupposition.

  31. Thanks David. Can we treat ‘ideal conceivers’ as extensional contexts?

    You wrote:
    people sometimes say that identities don’t need explanation, but i think this is just false. one can give a perfectly good explanation of the identity between heat and the motion of molecules (roughly, by showing how the motion of molecules plays the heat role),.

    Is it necessary to explain that something is identical to itself? It seems not. It isn’t even clear it is possible to explain, as much as hope your interlocutor understands the meaning of identity.

    Sure, you might need to do a lot of work to show that some identity holds in the first place, but that is not the same thing.

  32. David, I’ve proposed that phenomenal events can be understood in terms of brain events if we adopt a bridging principle of corresponding analogs between 1st-person descriptions of phenomenal events (generated within the brain of a subject) and 3rd-person descriptions of brain events (generated by observers other brains). These two kinds of descriptions occupy different descriptive domains so we cannot assert mind-brain identity. However, if we can formulate biophysical models of brain mechanisms that generate good biophysical analogs of salient phenomenal events (matching features), I think we are justified in claiming (within the norms of science) that we can explain conscious content. What are the counter-arguments?

  33. eric: “it is ideally conceivable that S” is still epistmological and non-extensional, just as “it is knowable a priori that S” is. regarding “is it necessary to explain that something is identical to itself?”, see the last sentence of my previous note.

    arnold: all that is fine. it’s precisely the strategy that i adopt myself (see e.g. http://consc.net/papers/scicon.pdf). but it is not a wholly reductive explanation, precisely because you have to appeal to a primitive bridging principle that is not itself explained.

  34. Richard: The seeming difficulty for explaining consciousness may be much more a matter of our historical situation than a reason for why the hard problem can never, in principle, be solved. I thought zombies were supposed to convince me that this seeming difficulty really is a difficulty. Maybe that’s not exactly right, but that’s how I’ve always understood it. We could imagine, however, with not too much difficulty, I think, that people in the middles ages would find explanations of heat or light or water in terms of structure and function largely unimaginable (people were very stupid, you know!). Of course, they were just wrong. So what? Why should we believe consciousness is any different? As I say in my comments, our common sense understanding of the world is often (perhaps even always) in conflict with the best scientific explanations we have. That doesn’t make the explanations wrong. It makes our intuitions wrong.

    Dave: I agree that some explanation of identity is necessary for reduction. But your examples give explanation in terms of functional role. I don’t see why such an explanation would be unavailable (or, indeed, in principle impossible) for consciousness.

  35. Excellent discussion here! Thanks to everyone and especially to David Charmers for rehearsing the main points of his position. I’d like to follow up on certain things he said.

    It concerns the modes of presentation, physical explanations and explicability of identities.

    I agree that there is a distinct hard problem of consciousness. It only seems that it emerges because we can conceive of zombies, i.e. we can conceive of the psychophysical identity falling apart. But that’s not the main reason why it emerges, because we can also conceive other identities of natural kinds falling apart, like heat and molecular motion, but still in that case the argument wouldn’t go through because although conceivable it’s not metaphysically possible that they fall apart. The thing is we can conceive of identities falling apart in both cases because they are by assumption a posteriori identities.

    However, the main reason why the argument goes through in the case of psychophysical identity is because of the peculiarity of modes of presentation that are involved. Namely, we use “phenomenal concepts” to refer to the unique, ineffable qualities of consciousness, i.e. to qualia. Phenomenal concepts use direct modes of presentation, Levine (2001), who proposes another account of the hard problem, namely the explanatory gap, calls them “thick” modes of presentation.

    Now, the fact that phenomenal concepts use direct or “thick” modes of presentation instead of descriptions of the causal or functional roles (the “thin” modes of presentation) to connect to their referents, prevents their integration with physical concepts and thus creates the explanatory gap.

    David says that identities do require explanations, i.e. that they are not inexplicable. But I think Block and Stalnaker (1999), Levine (2001) and Papineau (2002) agree with this to a certain extent. They say that it’s not a further explanation, it’s merely a justification of the identity. Once it’s established that X and Y are identical then it doesn’t make sense anymore to ask further why they are identical. Which is in my opinion a bit different from the claim that they are inexplicable. And more to the epistemological point that David makes, the purpose of identities according to these authors is to transfer explanatory power onto some further or some new properties and in that way integrate properties into a single scheme.

    In my opinion what seems to give rise to a feeling of something left unexplained here is the difference in modes of presentation that prevents integration of properties under a single scheme. For example, we don’t have a hard problem with natural kinds, it doesn’t emerge between micro and macro concepts in physics because we can see that if the concepts of heat and molecular motion use descriptions of the causal or functional roles they can both be integrated in a more global scheme of the causal or functional roles. SInce phenomenal concepts don’t use the descriptions of causal or functional roles as modes of presentation they can’t be integrated into such a scheme. And this creates the explanatory gap or the hard problem. That is, it doesn’t seem intelligible how they can be identical.

    But why assume that descriptions of the causal or functional roles alone constitute intelligible explanations?

    Why not use topological models instead of causal descriptions as a basis of intelligibility of explanations? After all, they are widely used in physics, chemistry or in evolutionary biology to explain certain phenomena or to integrate certain phenomena that are at different level levels of explanations. For example, they are used to explain the relation between genotype and phenotype spaces, or heritability, etc.

    And they don’t have to be strictly speaking opposed to casual or functional explanations. They can be perhaps contrasted with mechanistic explanations a la Bechtel, Richardson and Craver. However, they do a great job at explaining and integrating phenomena or domains regardless of the causal or functional roles the properties play on any standard view.

    I was wondering what David Chalmers thinks about this proposal? Or in general what he thinks about other bases of intelligibility of explanations apart from the descriptions of the causal or functional roles since there is no argument for using exclusively descriptions of the causal roles as a basis of intelligibility?

  36. ellen: again, the gene-DNA and heat-molecule examples were just used to make a limited point: that explanatory contexts are non-extensional and that identities can be explained. of course a central aspect of the DHP is that the sort of explanatory strategy that works for life, genes, heat, and so on doesn’t work for consciousness. as for the point about heat/light/water and the middle ages, there’s a discussion of these cases early in http://consc.net/papers/moving.pdf.

    daniel: actually my version of the anti-materialist argument doesn’t require the premise that phenomenal concepts are direct (or that they have the same primary and secondary intensions) — that’s discussed in http://consc.net/papers/2dargument.pdf, pp. 12-13. as for topological explanations: i’d be more than happy to use these if they worked. but as far as i can tell there is the same explanatory gap between topology and consciousness that there is between causal role and consciousness and between physical structure and consciousness (in fact i’d count physical topology as an aspect of physical structure in the arguments discussed above).

  37. But David, every identity is something being identical with itself. That’s my point. That’s identity.

    The concern with ‘modes of presentation’ I would argue is another way to say that providing evidence for an identity can be nontrivial. But again, that is a different question. I think this is the point of Block and others (as Daniel just pointed out).

    Thanks for clarifying what you mean by ‘ideally conceivable’ as still inducing a nonextensional context. So it doesn’t mean they cannot make mistakes (e.g., even an ideal conceiver could still think that Mark Twain is not Samuel Clemens, if given misleading evidence?). This is what I was leading up to with all these questions about nonextensional contexts: can ideal conceivers make mistakes of the Clemens/Twain variety, or does ‘ideal’ include the extension being transparent?

    For the philosophers in the house, is there a better way of talking than Fregean ‘modes of presentation’? Has anyone developed a more general and literal and nature-joint-carving way of discussing this stuff than ‘modes of presentation’?

  38. why is the morning star the evening star? we don’t just say: because it is identical to itself. we say: it’s because there’s a planet closer to the sun than us that stays near the sun relaiove to us and is usually visible around sunrise or sunset. this isn’t just evidence — it’s explanation.

    ‘it’s ideally conceivable that S’ for present purposes comes to roughly ‘It is not knowable a priori that ~S’ (strictly this is the notion of ideal negative conceivability). It’s not knowable a priori that clemens is twain, so ‘twain is not clemens’ is ideally conceivable. the relevant idealization here is just one of a priori reasoning, not empirical knowledge. see “does conceivability entail possibility?” for much more on this.

    speaking for myself i think the 2D framework (and in particular primary intensions) provides a good formal way of understanding modes of presentation.

  39. David: I think the main point is that it doesn’t make sense to ask why a is itself. If you agree with that, then I think you have agreed with the key idea.

    In nonextensional contexts, things get interesting when you have the same thing with two names.

    But in an extensional context this doesn’t come up as an issue, and there is an extensional sense of ‘explains’ that I think many people use (I know I do). That is, I use ‘X explains Y’ extensionally, to describe an objective, counterfactual-supporting, feature of the relation between X and Y (the things out in the world).

    It is like asserting ‘Dogs are mammals’ rather than ‘I believe that dogs are mammals.’ It is not making any reference to concepts about dogs, but is just stating a fact about dogs. When I say that ‘This animal’s history of reinforcement explains why it pressed the lever’ I am talking about the world, not my beliefs about the world.

  40. Note I am just trying to unpack what I mean when I say identities don’t have explanations. Since you use ‘explains’ in a different way, it becomes more interesting for you. When you write ‘explains’ it requires something like a ‘believes that’ operator to indicate a nonextensional context has been set up, so all bets are off regarding the standard rules of inference.

  41. Thanks for the response Dave!

    I wouldn’t say that every valid argument presupposes its conclusion but I definitely tend to think that every argument that has a conceivability claim as one of its premises does (because in order for that premise to be true you must presuppose the conclusion, which is what I take Glenn and Liz to be arguing for here). It is true that some people find zombies intuitively conceivable, yet others don’t (as you know!)…it is also true that some people have become convinced that there is a VHP because of those intuitions and others have become convinced that there is no link between conceivability and possibility but this is consistent with those people already holding the view about the unexplainability of consciousness implicitly or unconsciously and the intuition or feeling that zombies are conceivable bring that theoretical commitment to light, so to speak.

  42. Ellen, thanks for the follow up!

    You don’t have to convince me that what you say could be true! I was only trying to get clear about what the argumentative landscape looks like. You objected that we could get an analogous hard problem for water-h2O or whatever and I was just trying to say how the opponent would respond (not endorsing the response). The key idea is that in all other cases identities can be deduced because we can come to see that a certain physical bit of space-time plays a certain functional role but with consciousness we start from the position that there is/seems to be a DHP which just is the claim that there doesn’t seem to be any functional role for consciousness to play…now of course you and I disagree about that and think that there only seems to us now to be a DHP (and if we ever did come to see that there wasn’t then I agree that we could deduce the identity just like everything else…in fact I find myself believing that at least every other day 🙂 ).

    Dave objects that consciousness still seems different because in the other cases we could at least see that there was some functional role in the area and just could not see how something could instantiate that role. So, when Locke says that it is inconceivable how we could explain the macro properties of water in terms of microstructure it looks what he means is that there is no way that we could account for the causal properties of water in terms of microstructure and if so then he is just stumped about a particular functional explanation not stumped because he thinks there are non-functional things (or things that seem non-functional) that need to be explained.

    I really am not sure what the best response to this is…I have some thoughts but I have to run out to the ‘conference reception’! I’ll try to get back to it tomorrow 🙂

  43. It seems to me that those who believe that consciousness plays no functional role must believe that we get our phenomenal world for free. They have the explanatory burden of showing how we could possibly experience the world around us without being conscious.

  44. eric: it doesn’t make much sense to ask why hesperus is itself, but it makes a lot of sense to ask why hesperus is phosphorus and to give an answer along the lines above.

    richard: conceivability arguments typically have possibility claims as their conclusion, but i think it’s obvious that the conceivability premise often doesn’t presuppose the possibility claim — as many accept the former but reject the latter. maybe accepting the conceivability claim brings out some sort of “theoretical commitment”, but only in the sense that accepting the premises of any argument brings out a “theoretical commitment” to the conclusion.

    this is glenn and liz’s show, so i don’t want this discussion to devolve into a back and forth with me — how about getting glenn and liz back in here!

  45. David: you are (arguably) right if you use ‘explains’ in a nonextensional way, which many do not. It seems Block/Stalnaker are not. This would explain their claim:
    “Identities don’t have explanations (though of course there are explanations of how the two terms can denote the same thing).”
    At any rate, you are right this is getting outside the range of the target article.

  46. An interesting debate but, following drbillh, I’m still confused as to what’s being talked about in regard to ‘structure and function’, particularly because of Glenn’s statement that drbillh comments on:- “consciousness is dependent on structure and function of mental states”. I’ve read Glenn’s response to drbillh, but it’s still unclear to me, so perhaps if I give an example someone could clarify if I’m on the right lines.
    If I stare at a bright green image for a minute or so, then look away at a white background, I will see a cyan after-image. When people have been talking about ‘structure and function’, in this example that would refer to the bleaching of my retina, or the neural activity in my brain, say. These are what Glenn means by “mental states”, or “representing vehicles”.
    What it wouldn’t refer to is the cyan after-image that’s induced by my retinal/neural activity, principally because this can’t count as evidence since no-one else can see it (though they can of course see a cyan after-image induced by their own retinal/neural activity). So my particular cyan after-image isn’t what Glenn would call a “mental state”, or “representing vehicle”. But it would count as “consciousness”.
    Am I on the right lines, or am I missing a point?
    Thanks

  47. Hi Tony
    thanks for your thoughts, but I’m a bit confused about what you’re getting at. I think the cyan after image would be a mental state- conscious experiences are the paradigmatic examples of mental states (indeed I think this so strongly that I’m inclined to say there’s no point in studying the mind without studying consciousness- but then maybe that’s something I say to bug people). At any rate I certainly never intended to deny that any conscious experience is a mental state- is that what you’re worried about or have I misunderstood?

  48. Hi Tony,

    This is a hunch because I am also a bit confused by where you are coming from but I think it might help to get a bit clearer about representation and the relations between representation and consciousness. Representation involves (at least) two things: the representing vehicle which is the thing that does the representing (e.g. the words on the page or your mental states) and; the represented object which is the thing that the vehicle is about (e.g. ‘dog’ is about those (usually) four-legged hairy things that bark and are man’s best friend).
    What I (and I think most people who don’t believe in sense data) would say is that when you have an after-image you are having a representational mental state which is incorrect i.e. it is telling you that there is a cyan square when in fact there is none. Now it is an open question as to how to flesh out this representational part of the story, but this aspect does not concern consciousness.
    Where consciousness comes into the picture is when we try to find out what the bearer of phenomenal qualities is within this representational picture. Glenn and I are phenomenal internalists (i.e. we think phenomenal properties are in the head) and we prefer a picture where the phenomenal qualities are the medium of mental representation i.e. they are what we use to build our representations out of, they are akin to mental paint or clay. In a similar vein Dan Lloyd talks about consciousness as being a virtual reality we create in our heads. So we would say that the phenomenal redness is a property of your mental state that is being used to stand-in for some (probably complex) property of objects in the world (see http://mq.academia.edu/ElizabethSchier/Papers/1470529/The_represented_object_of_color_experience for my take on what color experiences represent). In veridical representation the object does have the property you are representing it to have, in mis-representation it doesn’t. In the case of after images this is because there is in fact no object (or surface) there at all. So the phenomenal redness is something you correctly or incorrectly use to represent the world.
    Phenomenal externalists locate phenomenal qualities in the external objects (e.g. the surfaces of objects for colors). To be honest I am not up with the details of the story they would tell here but my guess would be that they would agree with the general misrepresentation story but say that an after-image is like imagining say a square in that you can represent the property without it being present.
    So to answer your question we would say that the cyan is a property of your mental state which you use to represent something about the world; it is the clay with which you build your representational model.
    As an aside, and to use a useful distinction Dave has drawn, we are not representationalists in the traditional sense. Representationalists usually explain consciousness in terms of representation (e.g. according to Dennett, those representations which are conscious are those which are famous i.e. which the rest of the system actually gets to know about), in contrast we (and most noteably also Gerard O’Brien and Jon Opie as well as Dan Lloyd [and I would argue Paul Churchland]) explain representation in terms of consciousness i.e. we see phenomenal qualities as coming first and representation as emerging out of the use that they are put to in the cognitive system (you need to have the paint before you can have a painting but paint by itself is not a painting)
    Sorry that was long-winded but I think you are touching on some complicated but very important issues.

  49. Dave,
    Thank you so much for clarifying things. Thanks to that we now have a much clearer idea of what our worry is.

    What continues to worry Glenn and I, Dave, is that we feel that you are vacillating on what the hard problem, or the distinctive hard problem, is and that when the ambiguity is cleared up your argument is no longer convincing.
    It seems to me that you are saying that hard problem is a challenge i.e. to explain consciousness and also a description of the phenomenon i.e. that there is an aspect of consciousness (what it is like) that is not functional (or structural).
    The ambiguity that worries us is that sometimes (e.g. when you grant that what it is likeness might be explicable in functional terms) it seems that the DHP is to explain the aspect of consciousness that does not *seem* functional.
    Now we are happy to grant that this is (at least at this point in time) obvious enough that you can have it as a premise without further argument.
    But this “seeming” version of the DHP is inadequate to do the argumentative work you put it to when you use it as a premise in the explanatory argument for dualism. It might be the case that function and structure can only explain function and structure. But function and structure (as you acknowledge) can explain something that does not *appear to be* functional or structural.
    Because of this we had always (in what was intended to be a charitable way) read the DHP as being the stronger claim that there is an aspect of consciousness that cannot be functional (or structural). I hope you agree that this is not a claim that is obviously true and is something that requires further support. If this is the case, what is the further support that you have provided or could provide? The point of our paper was to point out, and undermine, one source of support.
    I just want to note that I am aware that you give other arguments against the explicability of consciousness in terms of structure and function. We are focusing on the use of the DHP as a premise in an argument against physicalism/materialism.
    To sum up: it seems that the version of the DHP you can get for free is insufficient to do the argumentative work you put it to, and the version of the DHP that can do the argumentative work hasn’t been sufficiently justified.

    p.s. sorry about the delay, turns out I have torn ligaments in one ankle and probably the other foot so I have been busy with Drs appointments and scans. Thankfully doing philosophy is compatible with lying in bed with my legs up

  50. To get back to Glenn and Liz’s paper, and to tie in some of the comments above, I wonder if this isn’t an appropriate question: According to Chalmers, which intuition is more fundamental, the intuition that there is a DHP or the intuition that motivates the conceivability of zombies? If we have a sense that there is a DHP but not zombies, what kind of explanation can we offer in order to prevent the VHP? But if we can conceive of zombies, but don’t think there’s a DHP, where are we left?

    Perhaps this will help us to figure out where we are left. Or, better yet, where Glenn and Liz’s paper leaves us.

  51. Also, I just came across this and thought it might be of interest to some who have been participating in the discussion:

    Science Weekly podcast: Can science ever explain consciousness? w/Anil Seth, Chris Frith, and Barry Smith (Monday 27 February 2012)

    http://www.guardian.co.uk/science/audio/2012/feb/27/science-weekly-podcast-consciousness

    I especially like the last point about identifying what kind of explanation we would count as an adequate response to the hard problem of consciousness. And that if we don’t do this, we risk sounding like 5-year-olds, asking the “why” question.

  52. hi liz,

    thanks for this, and sorry to hear about your ankle! i hope you’re better soon.
    i’m not quite sure i understand the objection. the claim that there’s a DHP is quite compatible with the claim that consciousness is (a posteriori) identical to a functional property, as i discuss above. so my DHP claim is certainly not that consciousness is not a functional property. it’s also not my DHP claim that consciousness does not seem to be a functional property. so neither of your two proferred premises are premises that i invoke.

    instead, my DHP claim is that the problem of explaining consciousness is distinct from the problem of explaining functions — in the sense that there’s no function F such that an explanation of F is ipso facto an explanation of consciousness. i suppose you must deny this claim, but i’m not sure why. what’s your candidate for F? as soon as you name an F, i think it will be pretty clear that to explain F need not be to explain consciousness.

    maybe your view is something like: there’s no function F such that to explain F is obviously to explain consciousness, but there’s some function F such that to explain F is unobviously to explain consciousness. or perhaps better: explaining F combined with a few further unobvious steps will yield an explanation of consciousness. these are interesting views, but i’d still like to see you make a suggestion about F and about how the extra unobvious steps might work. these questions are especially pressing in light of the apparent residual question that arises for any such F: namely, why is performing F associated with consciousness?

    i’m also interested to hear just what examples you have in mind when you say function and structure (as you acknowledge) can explain something that does not *appear to be* functional or structural. if it’s things like heat, light, color, or life, i’m curious about how you address the disanalogies between these cases and the case of consciousness discussed in “moving forward on the problem of consciousness”.

  53. David, the ‘life’ analogy seems a good one. Some vitalists thought that there was an intrinsic teleology in biological processes, and this teleology could not be characterized in functional or structural terms, but in terms of their aim/purpose. (Note they thought this distinct from consciousness, as they typically did not think vital fluids were conscious, because plants and such are alive, but not conscious (though there was admittedly some disagreement on this among vitalists)).

    I am presently writing up the results of my survey of the vitalist’s primary literature, and its bearing on arguments about consciousness. The parallels, once you really dig in, are more striking than I had originally predicted.

    Also, I agree with the authors that you do seem to equivocate, as you sometimes characterize consciousness as an additional phenomenon in need of explanation (something ‘over and above’ the physical/chemical/biological facts of the situation), and other times characterize simply the concept of consciousness as having a nonstructural or nonfunctional aspect that cannot be explained.

    This is one reason I kept bringing up your use of ‘explains’ in a nonextensional way, as I think it adds to the confusion. It is not universal usage and invites misreadings. For instance, when you say ‘there is a further sort of explanandum: consciousness itself’, you are inviting the (mis?)interpretation that you think there is a novel phenomenon out there, not simply a unique concept ‘in our minds’ that may or may not have the same extension as some complex description of a brain state (even if it doesn’t have the same meaning).

  54. Eric, I would be really interested to read your historical work on vitalism. It is great that you have actually gone and done the research, I had always hoped that someone would do it. If/when you have a draft I would love to see it

  55. Dave, I have two general points to make. First of all I think it is fairly apparent that we are on opposite sides of the great divide. But I still think that there is an argument to be made that you are the one with the explanatory burden. Secondly I want to do some work to bridge the divide by thinking through the fact that consciousness has something to do with appearances.

    So, first, I don’t feel like you have addressed our worry, but particularly after re-reading your response to Dennett in “moving forward” I think this comes down to an issue of who has the burden of proof. I point this out because I suspect that this might be another way of saying it is a stand-off. So ultimately you “justify” the hard problem by appealing to the ubiquity of the intuition. I have three thoughts about why you nevertheless have the argumentative burden:
    1) I would be interested to see if the numbers changed if people were presented with the contrast between consciousness appearing not be structural or functional and consciousness not being structural or functional. As I have been saying we are happy to grant the former without further argument but we think you need the latter to use the hard problem as a premise in an argument against dualism and we think there is no simple move from the former to the latter (is this where you would invoke all your 2D work and point out that the primary and secondary intensions for consciousness are identical so the move from how it seems to how it is is justified here, unlike other cases?).
    2) if we take a broader view it is far from clear that you have the intuitive upper-hand if you are right that proto-panpsychism is the way to deal with the hard problem. If we are comparing type A materialism and proto-pansychism it is not clear to me that proto-panpsychism wins on intuitive grounds
    3) i don’t think it is entirely fair to shift the onus onto us to provide an example of a candidate structure or function so you can show it wont suffice to explain consciousness. You are the one presenting the positive argument and all we are asking is for you to provide evidence/justification for the truth of your premise.

    As for as my second general point, this is where I have some more positive thoughts that I sometimes hope in my more optimistic moments might help bridge the divide. Just a warning, what is to follow is very much a work in progress (I am only just starting on the book that will address many of these issues) so I appologise in advance for any lack of precision/details in what I present. The immature nature of these thoughts is why I have not raised them earlier. I hope (but am far from sure) that there are enough details for the position to be understood.
    I think one key bit of conceptual work that needs to be done for consciousness is to think more about what it means to say that consciousness is how things appear to us. It seems to me that we have two general options: either we accept that we can draw an appearance reality distinction for consciousness or we can’t.
    Let’s think first about what it means to say we can’t draw that distinction. It is here that I think that what Dennett once said to me is crucial: that it doesn’t make sense to talk about appearances unless they are appearing to someone. I take this to mean that the subject of an experience is necessary for something to be an appearance; if the subject is not aware then there is no appearance. This is I think why Dennett denies that there are real seemings that are independent of the subject’s awareness/judgements/access of them. But to accept that the subject being appeared to is necessary for phenomenal qualities is, it seems to me, to accept a fundamentally process/functional view of phenomenal qualities. Because how else are we to understand appearing to the subject except in functional, particularly access terms? It is here that I think you can just plug in Dennett’s functional account of phenomenal qualities. I suspect that the hard problem worry that would be raised here is that these functions seem independent of phenomenal qualities, to which I think the appropriate response is: that question is mistaken because it assumes that there are real seemings that the subject is accessing, but there can’t be because we can’t draw a reality/appearance distinction for consciousness. All there is is the subject’s “judements”.
    The alternative route (which I prefer) is to accept that you can draw an appearance reality distinction for consciousness. That how things appear to the subject is no more of a reliable guide here than anywhere else. I think that when going down this path, one accepts the division of the phenomenon to be explained along the lines of what Kriegel calls a division between what it is likeness and for the subjectness. To explain the former is to explain phenomenal qualities, to explain the latter is to explain awareness. The explanation I prefer here is one that identifies phenomenal qualities with connectionist patterns of activation, with the identity being motivated by the (ultimately) isomorphism between phenomenal space and activation space. I suspect that the hard problem worry that would be raised here is: why should that pattern of activation feel like that? The general response that can be given is question why we should trust how it appears to you? Given that we have accepted that we can draw a reality/appearance distinction for phenomenal qualities, how things appear to the subject is no more reliable here than in any other domain. I also think that there is a lot to be said for the sort of reply, that Austen Clark gives, that the relational structure of phenomenal qualities is defining of them. So to explain their relational structure is to explain them. But I think the appearance/reality move is the crucial one.

    Could that perhaps be the other non-obvious steps that can allow us to see how a function (Dennett) or structure (me, Clark) can explain consciousness?

    A final point, to answer your question, I don’t have anything to add to the historical examples (but I can’t wait to see Eirc’s work). The only, unhelpful, thing I can say is that it seems just obvious to me that something that does not appear to be structural or functional can nevertheless be explained in terms of structure or function.

  56. eric: i’d love to see the paper too. there was a paper on vitalism and the hard problem in PPR a while back, arguing that the vitalist nehemiah grew had views about life analogous to mine about consciousness — though on reading the article closely you see that the analogy isn’t all that close. and of course even if some vitalists did have views about life analogous to mine, that doesn’t show much — the issue isn’t so much what every vitalist said as what they should have said given their evidence and what they mainly said. after all there are plenty of dualists about consciousness who i think make bad arguments given their evidence!

    liz: (1) i agree that you’d get different answers to that question but i don’t think that would have all that much bearing on whether there’s a DHP. for there to be a DHP doesn’t need the ontological claim. yes, the 2D framework plays a key role in moving from the epistemological claim to ontological claims, though the claim that primary=secondary intension for ‘consciousness’ isn’t strictly required, for reasons i explain in “the 2D argument against materialism”.

    (2) sure, it’s not unreasonable to hold that various dualist views have just as many problems as many materialist views. but the claim that there is a DHP is much weaker than the claim that e.g. panpsychism is true or dualism is true — again, many materialists accept that there is a DHP. and even if one takes the view that all the options are bad, i’m not sure that does much to undermine the claim that there’s a DHP — it just makes the hard problem all the harder!

    (3) i give pretty general reasons to think that explaining functions doesn’t explain consciousness. on the face of it those reasons apply to any function someone might suggest. if you were left saying “maybe there’s a function you haven’t thought of yet for which they don’t apply, but i can’t tell you which it is”, then i’d need to have some inkling of a reason about why the general reasons don’t apply — and i’d also take it that anyone taking this line is acknowledging that there really is a distinctively difficult problem here!

    (4) the appearance/access strategy is one of the more interesting functionalist strategies. but as with the discussion of dennett on “seeming” in my 1996 book (basically the same issue), it seems to trade on an equivocation between phenomenal appearing and psychological appearing. to explain psychological appearing (i.e. the formation of judgments functionally construed) is not to explain phenomenal appearing (why there is something it is like to have the states in question. as for appearance/reality, i think one would still want to explain the phenomenal appearance, even if the psychological appearance can be misleading! still, both of these are strategies worth pursuing and i look forward to seeing your work on them!

  57. I’ll let you guys know when I am done with the vitalism ms. Right now it is too long (what a shock I need to trim my prose). It will be a couple of months.

    The analogies are strong, but there are clearly disanalogies. Vitalism parallels are not a knock-down argument, but a source of interesting ideas and part of an inductive cumulative case (the ‘cautionary tales’ chapter).

  58. Hi everyone, really enjoying the discussion in here!

    I hope you are feeling better Liz and thanks for keeping up the conversation!

    I just wanted to second Liz’s comment to the effect that it is unfair to ask for a specific function given that we admit that we don’t know how such an explanation would go. The question is whether it is possible for some functional explanation to explain something that does not appear to be functional and for that all one needs is to show that it is conceivable that it be the case. Certainly what Liz is claiming here is conceivable; it seems that way to me! In fact it seems like I can conceive of what she is talking about here even though I can’t conceive of how the explanation would go (it’s a bit like conceiving of an unknowable truth :).

  59. Hi everybody, it is a great pleasure to follow this discussion.
    My impression is that Glenn and Liz misinterpret somehow the zombie argument or at least its first premise. In Chalmers’ setting of the zombie argument, the burden of proof is on the proponents of type-A,C physicalism to show that zombies are ideally inconceivable, as well as on the proponents of “the phenomenal concepts strategy” (they appear later in the zombie argument) who should offer an account of how physical and phenomenal concepts could be independent of each other and yet co-refer to the same entities of the world. It seems to me that Glenn and Liz ignore that the burden is on their side.

  60. Hi you all. Sorry for steering back to an issue a bit upper in this thread.

    Thank you David Chalmers for your comments. They make me be clearer.

    Even if the topology was a part of the physical structure, a topological explanation doesn’t have to refer to that physical structure, or to its role in that physical structure for a topological explanation to work.

    My proposal is not just to mechanically swap physical structure for topology. If that were the case you would be obviously right. Maybe you are right about this anyways, but it’s not yet obvious to me. The idea is to use topological models of explanation to integrate properties at different levels of explanation. In that sense, we would be able to use the topological explanation even if the topology was not a part of the physical structure, i.e. even if there was no topology altogether.

    Given this, it’s not obvious to me how we have a hard problem between topology and consciousness, like we have it between causal descriptions and consciousness. Unless you meant that only in case we mechanically swap the topology for physical structure.

    The proposal here is to change a perspective about what constitute an intelligible explanation and to think about it in terms of features of topological spaces that are not necessarily the same as some natural topology, or the topology of the physical structure. We could think about some abstract topological spaces between genotype and phenotype spaces. Or we can think of one and the same system which when thought of in terms of descriptions of the causal roles can be seen only in one way. But we can think of particular casual roles in terms of their abstract relations that constitute topological spaces. In this sense given the same set of casual roles we could have many different topological spaces based on them. Some great papers about this are for example: Huneman, P., (2010), Bärbel M., Stadler, R. and Stadler, P. F. (2001), Benkö, G. et al, (2009), or Bullmore, E. and Sporns, S. (2009), or Jan Cupal, Peter Schuster and Peter F. Stadler-Topology in Phenotype Space (forthcoming). The idea of using topological explanation is widespread in biology, mathematics and physics. They don’t necessarily refer to the topology of the physical structure of the world. I think this kind of explanation (not structure) can be adapted for consciousness.

    I’m writing a paper about this, and I’m still formulating my ideas. This discussion helps tremendously. If you like I could send a synopsis (or the whole paper) when it’s done. It would mean a lot to me to get some feedback from you David and from all of you folks here.

    I have another question for David Chalmers. If your formulation of the DHP doesn’t require direct reference of phenomenal concepts or collapsing intensions, would you agree that the DHP has a different formulation than the explanatory gap (Levine 1983, 1993, 2001)? They could be even two different problems. But in any event, even if they are one and the same problem, their different formulations may require different solutions. people often use them interchangeably, but that might be wrong. Does this make any sense to you?

  61. richard: i suppose there are two sorts of type-A materialists — the gung-ho type-A materialists (analytic functionalists and the like) who think there’s some function F such that it’s already clear that explaining F will explain consciousness, and “mysterian” type-A materialists (ormaybe type A/C-materialists) who think that we will one day discover an F such that explaining F will explain consciousness. i’m not certain how to best understand the mysterian version — is the idea roughly that once we hit on the right F we’ll see that explaining F will automatically explain consciousness? or is it that we’ll find some deep unobvious connection between F and consciousness, so that explaining F and then carrying out a bunch of further bridging steps will explain consciousness?

    daniel: as an ex-topologist i’d love to see these ideas developed, though i don’t yet see how topology escapes the structure-function dilemma. as for the HP and the EG, at a very broad level of abstraction these are similar (it’s hard to explain phenomenal consciousness in physical terms!) — though the HP formulation focuses on the distinctive explanandum (contrasting it with the EP explananda), whereas the EG formulation doesn’t. (i think this is why the HP formulation has been much more influential among scientists, the audience it was always especially intended for.) at a closer level the associated arguments are pretty different — levine uses a conceivability argument to argue for the EG, whereas my case for an HP doesn’t involve conceivability arguments at all, but instead the considerations about structure and function as above. as for the material about collapsing intensions, i’d say this isn’t so much relevant to either the HP or the EG as to the further step from those things to the falsity of materialism.

  62. Daniel – I would be interested in reading your paper when it is done.
    Richard – it is reassuring that our position seems viable to at least one other person!
    Dusko – our problem is not with the zombie intuition but rather the intuition that there is a distinct hard problem. In fact we mistakenly thought that the zombie intuition was evidence for a DHP

    Dave, I will address each of your points in turn
    Re1if it is not an ontological claim, then how can you get anything other than a claim about how it seems to you? And if all you can get is a claim about how it seems to you, then I don’t see how this is sufficient to support the claim that any explanation in terms of structure and function must be insufficient. This is the ambiguity that worries us.

    Re 2 I wasn’t claiming that DHP includes an assumption of dualism, hence the talk of “taking a broader view”. Rather the point was that you seem to be inconsistent in your use of intuitiveness as a criterion for theory evaluation. It seems to me that your ultimate objection to type A materialism (which as I understand it is the position that denies the existence of a DHP) is that it is counterintuitive (see p.9 of “Moving Forward”). So the point of brining up the counterintuitiveness of dualism is simply to say that it is inappropriate to reject one position only because of its counterintuitivenss if you then go on to adopt what is arguably an equally counterintuitive position. I know that in the section I point out you use counterintuitiveness primarily to shift the burden of proof, but who has the burden is precisely what worries Glenn and I. I feel the great divide looms again.
    On a related note, and again this is a work in progress, It seems to both Glenn and I that all the current options re consciousness have counterinutive components. This has led us to play around with the idea that in fact our intuitions about consciousness are inconsistent in that they cannot all be true at once. If this is the case then maybe the great divide is created by the fact that different people take different intuitions to be more basic/important.

    Re 3 see (1) we don’t think you have met this burden because ultimately you appeal to an intuition which we think can only justify a claim about how consciousness appears and that such a claim cannot do the argumentative work you put it to.

    Re 4 Sorry, I should have known you would bring up psychological vs phenomenal seeming. I avoided the issue because I thought it would make things more complicated. But I do not think my point relies on a conflation of the two, which is why I talked about “plugging in” Dennett’s position i.e. to suggest that his equation of psychological and phenomenal seemings was only one route one could go.
    I think what is crucial is accepting that phenomenal appearing is relational in that it involves something (maybe in the head, maybe not) being in an appearing relation to the subject. I make the claim that any other way of understanding appearing is nonsensical. So the question then becomes is the thing that appears phenomenal independent of its appearing relation to the subject?
    If it is then introspection and the first-person point of view are no more privileged for phenomenal qualities than any other area and so we have opened the door to “objective science” (with respect to the DHP I would say that the fact that it does not phenomenally appear to you that consciousness is structural/functional is no more of a problem than the fact that the earth does not phenomenally appear to be moving).
    If on the other hand appearing to the subject is necessary for phenomenal qualities then it seems that being in the phenomenal appearing relation to the subject somehow makes something phenomenal. So to complain that explaining phenomenal qualities in terms of some relation to the subject is to leave out the real intrinsically phenomenal things is to miss that being related to the subject is (at least part of) what makes something phenomenal in the first place. Of course this does not show what that relation is and that it is physical, but my claim is that there is no reason to think that it cannot be, beyond a belief in real seemings.

  63. A bit more on vitalism, on how to explain in materialism-friendly terms something that is not initially characterized in such terms.

    I can conceive of organismal development teleologically. E.g., there is an intrinsically teleological force that pervades an entire organism from birth, guiding development toward the end product. The easy problems of development would be the description of how this force interacts with various molecules and such to produce changes at a particular point in time. The hard problem of development is purpose or teleology itself, which I conceive of as an intrinic feature of the vital force, something upon which its activity depends, but which is not reducible to said activity.

    It isn’t that hard to imagine how this apparent teleology could be explained in materialism-friendly terms, even if it was once considered impossible. Using the language in this thread, this seems to be a case of explaining something that is not structure/function in terms of structure/function.

    People today don’t seem to appreciate how pervasive the notion of teleology used to be in thinking about biological systems, even basic physiology. Kant suggested it was ineliminable. He wrote (Critique of Judgment):

    “Absolutely no human Reason (in fact no finite Reason like ours in quality, however much it may surpass it in degree) can hope to understand the production of even a blade of grass by mere mechanical causes. 1790/1914 p326

    Why was this? Because there will be a teleological conceptual residue:

    “[I]t may be that in an animal body many parts can be conceived as concretions according to mere mechanical laws (as the hide, the bones, the hair). And yet the cause which brings together the required matter, modifies it, forms it, and puts it in its appropriate place, must always be judged of teleologically… ” ibid p 282

    Whewell reinforced this (in talking of ‘organization’ and ‘organized bodies’ he is just talking about living things):
    “Thus we necessarily include, in our Idea of Organization, the notion of an End, a Purpose, a Design; or, to use another phrase which has been peculiarly appropriated in this case, a Final Cause. This idea of a Final Cause is an essential condition in order to the pursuing our researches respecting organized bodies.” (1858, p 240).

    The parallels seem pretty clear. There is an ineliminable conceptual tarnish for the mechanical philosophers: the idea of an intrinsic end/purpose/teleology. The vital force was conceived, by the vast majority of vitalists, as a cause of the structural and functional facts, and not constituted by such facts.

    For Kant, we were forced to think teleologically about physiological systems, just as we are constrained to think in terms of the categories of the understanding about physical systems. At least that’s how I’m reading him. I’m no Critique of Judgment expert.

  64. Chalmers, discussing vitalism, wrote (in ‘Moving Forward’):
    When it comes to the problem of life, for example, it is just obvious that what needs explaining is structure and function: How does a living system self-organize? How does it adapt to its environment? How does it reproduce? Even the vitalists recognized this central point: their driving question was always “How could a mere physical system perform these complex functions?”, not “Why are these functions accompanied by life?”

    This is the best response to give, I think. But to point out a disanalogy does not kill the analogy altogether.

    The conceptual point remains, that they believed that there existed this stuff with irreducibly teleological properties (and really, they were just following Aristotle). Kant and others argued that they literally could not think about them any other way. It is true that the less armchair-focused vitalists said they had really good evidence for the existence of this teleological force.

    Chalmers could argue about who had better evidence and arguments to take the two respective nonfunctional conceptions seriously. But that is a different question, it seems, and one the materialist about consciousness should be happy to take up once the basic conceptual points is granted.

  65. Note: for vitalists the teleology was not just a theoretical construct. They observed teleology in things like the plant moving toward the sun. Yes, at some level, that is just movement, but it is movement toward an end. To simply describe it mechanically is to miss the whole point, is to not do biology, but physics (that is the point, I take it, of Kant’s views).

    So Chalmers is partly wrong to say they simply wanted to explain structure and function, as if that is all they thought they were observing. They observed life itself, which was partly constituted by its teleological nature.

  66. liz: (1) it’s a claim about explanation, not about ontology or about seeming. (2) i think the basic objection is that denying the DHP denies a manifest explanandum — the basic objection isn’t really that it is counterintuitive, though that’s a consequence. (3) see (1). (4) i’m happy with the idea that some relation to the subject makes things phenomenal, but i don’t see that this does anything at all to dissolve the DHP.

    eric: i think for the majority of vitalists who appeal to nonmechanical teleology, it serves as part of the best explanation of the functional data (an entirely reasonable explanation at the time), and not as part of the data. some vitalists may have treated nonmechanical teleology as a datum, but to them the right and reasonable response even at the time would have been “no it’s not — the data to be explained concern complex structure and function”. some were probably in between, taking the step from complex function to teleology to be a really obvious and immediate one, and thereby talking of perceiving teleology — but i think on reflection these people would/should allow that if (improbably!) there were to be a mechanical explanation of the functional stuff, that would explain the data.

  67. David: it is actually hard to tell how “observational” the teleological judgments were sometimes, as teleological thinking had an obligatory flavor to so many people.

    Even for those who used teleological concepts as part of an explanans rather than explanandum, the conceptual parallels remain and can be applied at the level of the explanans (this is why I said the discussion would then move to concerns about evidence for applying these concepts, which the materialist should be OK with once the conceptual point is made). It would be a little odd to ignore this conceptual parallel, given that your arguments are almost purely conceptual.

    That said, there were also those that described their observations of plants moving and such in teleological terms. Here the parallel is stronger. They had this teleological concept they applied to plants moving and such, a conception that some couldn’t imagine getting rid of, but now we explain this apparent teleology using modern concepts from biology (concepts they simply did not have at their disposal). Again, you could argue about how reasonable they were being and such, but the conceptual parallels are pretty clear.

    I would caution against being too cocky, with our modern biological notions, of how easy it would be to recast things in nonteleological terms. Even today, it is often very hard for people to cast descriptions of biological processes in nonteleological terms. It isn’t hard to imagine a time when such language was taken literally.

  68. Thanks David Chalmers for the comments. And Liz get well soon! I’ll send you the paper when it’s done. Cheers! Eric could you please put me on some mailing list for that paper about analogy with vitalism. It sounds very interesting.

  69. To Liz’s point, David wrote:
    denying the DHP denies a manifest explanandum

    Let’s stipulate for argument that there is a conception of consciousness that picks out experiences, and this conception doesn’t characterize said experiences in functional/structural terms. Note that this conception doesn’t explicitly exclude any explanatory or ontological resources. It is ‘topic neutral’ in that sense.

    It should take a lot more than that to convince you to take the giant leap of dismissing the explanatory armamentarium of the natural sciences. As Liz suggested, I think the burden of proof should be on the person making this primae facie radical claim. This thin, effectively ostensive conception of experiences seems nowhere near enough.

    The materialist ought to just say they are satisfied that this phenomenon, picked out partly with the help of this topic-neutral ostensive definition, is being empirically revealed to be a complex brain state (or whatever). This theory also has the nice consequence of not implying that thermometers are conscious.

    The upshot is that while I deny the DHP, I do not deny the experience-focused, topic-neutral, conception of consciousness. I just would resist using such a conception to undergird the radical claims about what types of explanantia are off limits. It just smells like a misframing of the issue, because the previous paragraph, written in the materialist frame, just seems so obviously reasonable.

    Thanks to David Chalmers for being patient here and engaging with all of us.

  70. I think anti-physicalist arguments from the conceivability of a ‘ghost’, i.e. a subject of experience whose nature is exhausted by its conscious experience, put the onus on the physicalist more so than arguments from the conceivability of a zombie. It is extremely plausible that when one reaches the end of Cartesian doubt one finds oneself conceiving of oneself as a ghost.

    It’s not a knock down argument (what is in philosophy?), but I think the onus is on the physicalist to explain where they end up when they engage in Cartesian doubt, and what mistake the anti-physicalist is making which results in her conviction that she can concieve of the non-existence of everything except herself and her conscious experience.

    It’s not quite as straightforward to get from the conceivability of ghosts to the falsity of physicalism as it is to get from the conceivability of zombies to the falsity of physicalism. But it looks like the conceivability of ghosts is going to be inconsistent with analytic functionalism (there are no non-phenomenal states in a ghost’s world to realise its phenomenal states) and analytic information theoretic representationalism (there is no external world with which a ghost interacts).

  71. Philip is anyone an analytical functionalist these days?

    I think my description above explains why we can conceive of ghosts/zombies. It’s because when you use the topic-neutral characterization of experiences, it leaves things completely unconstrained regarding the basis of the phenomenon. So you can conceive of it attached to pretty much anything, or even completely detached from everything, in Disney-esque fashion (e.g., conscious goblets and carpets).

    But I do think zombie/ghost arguments are orthogonal to the questions here, specifically focused in on the DHP, which doesn’t depend on such things (as David pointed out).

  72. eric: this sounds to me like a version of the “phenomenal concept strategy” (itself a sort of type-B materialism) and raises somewhat different issues from the ones we’ve been discussing above. see my “phenomenal concepts and the explanatory gap”.

  73. Eric: Well, I think Frank Jackson is some kind of information-theoretic analytic functionalist. Sure, I was just thinking of conceivability, which is only a threat to a priori physicalism. Regarding your approach, it seems to me to imply that phenomenal concepts are ‘opaque’: that they reveal nothing of the nature of the states they denote. This seems to me a very implausible account of phenomenal concepts. No more plausible indeed that analytic functionalism.

  74. Hi all
    Glenn and I just wanted to say thanks to everyone for all their great comments and particularly to Dave for his patience and interest. It has been great to hear all of the reactions to our ideas.
    Cheers
    Liz and Glenn

Comments are closed.