PMS WIPS 006 - Brendan Ritchie - Dualism and the Limits of Conceivability

“Dualism and the Limits of Conceivability” by Brendan Ritchie, University of Manitoba.

One of the most widely discussed arguments against physicalism has been the Zombie Argument (or ZA) of David Chalmers (1996). There is, however, more metaphysical baggage in this argument than is usually assumed; in fact, Chalmers faces a trilemma: either (1) zombies are not ideally conceivable, which ZA requires; (2) zombies are ideally conceivable, but this can only be shown a posteriori, which undermines the a priori nature of ZA; or (3) Chalmers must abandon the particular brand of functionalism he espouses and adopt another one which one must hold in order for ZA to succeed: limiting the scope of ZA as an argument against physicalism.

[Link to full text of article]

[Link to further info on PMS WIPS]

8 Responses to “PMS WIPS 006 - Brendan Ritchie - Dualism and the Limits of Conceivability”

  1. Pete Mandik says:

    Hi Brendan,

    I’ve enjoyed your paper, but wanted to raise the following concern.

    For a given “world” there are two issues that can be decided separately (i) whether the world is ideally conceivable and (ii) how similar it is to the actual world. The former looks like the sort of thing that people like Chalmers would try to settle a priori where the latter is going to be largely a posteriori.

    Calling a world a zombie world combines claims of type (i) and (ii) and thus makes things potentially confusing. So, when I say that there is a world that has my zombie twin in it, I’m asserting (among other things) that (i) it is apriori possible for there to be a world with a being with such and such respects and (ii) that I have such and such respects. But whether I have such and such respects, like this functional organization rather than that, is known a posteriori.

    Similar remarks are going to apply to what you call inter-zombie-worlds and epi-zombie-worlds. That the actual world has such and such physical interactions in it is a posteriori. I suppose it is also a posteriori whether interactionism is true: it shouldn’t be too hard to design an experiment to see if a physical event happened as the result of a purely mental cause. Whether some world, W, that has creatures with such and such functional organization but no consciousness is ideally conceivable is a priori. Whether that world contains creatures functionally identical to the creatures in the actual world is a posteriori.

    So, I think Chalmers can say that the worlds picked out by you as “inter zombie” and “epi zombie” worlds are both ideally conceivable and thus their existence is knowable a priori, but whether they count as containing genuine functional replicas is a posteriori. And further, Chalmers can grant that you are right that both worlds cannot both contain functional replicas of the beings in the actual world.

    Here’s another way of putting the point. We can stipulate that “zombie” means simply “being that lacks consciousness” and thus identify things as zombies without thereby committing to their being anyone’s zombie twin. A zombie twin, Y, of X is someone just like X in all respects except that Y lacks consciousness. So, the worlds you label “inter zombie worlds” and “epizomibe worlds” both contain zombies and whether they contain zombies is knowable a priori. But those worlds cannot both contain zombie twins of the actual world. Further, which one is the one that contains our zombie twins is a posteriori insofar as it is a posteriori what the functional organizations in the actual world are.

  2. Pete Mandik says:

    [[A glitch in the matrix forced Brendan to submit the following via email:]]


    Thanks a lot for the comments. Very similar things were suggested to me by Justin Fisher, so thanks for bringing this up.

    So I think you might be right, but I do not think this helps Chalmers, since it amounts to accepting the second horn of my trilemma.

    1. First, let me challenge breaking worlds into (i) and (ii). if ideal conceivability applies to worlds independent of how similar they are to the actual world, then it ends up being that zombie twins and worlds are not ideally conceivable, when similarity relations are built into the sentence describing them. So my zombie twin is not ideally conceivable for instance. Afterall, if you ask me to conceive of a being physically identical to myself without consciousness obviously I am building in similarity.

    Now here is a reason why you might resist this. It seems I can conceive of a world where I have dyed my hair. When I do this, it does not seem like I conceive of a person with certain properties, then check to see if I have them. That I am conceiving of MYSELF with such hair is built into what I am supposed to be conceiving of. Likewise, it seems I can intellibly conceive of my zombie twin, and it be part of the conceiving that the zombie is me in another world, so to speak. If this is right, then my zombie twin is conceivable a priori, and the similarity relation does not require some further a posteriori move. One last point that is pretty weak: I have to confess, I am not sure what I am conceiving of if I try and conceive of someone with dyed hair and then “check” to see if it is me. I wonder if anyone shares this intuition.

    2. Suppose that zombie worlds require (i) and (ii). Then this is first, a different argument than Chalmers gives, and second, takes the second horn of my trilemma, and makes ZA a posteriori.

    2.1. Usually we can give the zombie argument like this

    1. Zombie worlds are ideally conceivable

    2. If a, then zombies worlds are possible.

    3. If they are possible, then consciousness does not logically supervene on the physical.

    C. Therefore, physicalism is false.

    But if zombie worlds require an a priori premise then 1. is no longer shown a priori. In fact, zombie worlds of the kind that builds in similarity are not ideally conceivable. Instead ZA relies of the following:

    0.a. worlds with physical beings without consciousness are ideally conceivable

    0.b. Such worlds are physically, functionally and psychologically identical to this one (shown a posteriori)

    1*. Therefore, zombie worlds are ideally conceivable.

    This makes ZA a different argument than the one Chalmers gives. But this you recognize.

    2.2 But, more importantly, it requires an a posteriori premise. If only 0.a is true, then we do not have a failure of logical supervenience of the phenomenal on the physical. After all, I can conceive of a worlds with only protozoa, with certain functional organizations etc. Such worlds tell us nothing about supervenience of phenomenal states since no one claims (who isn’t all ready say a panpsychist) that such things have consciousness.

    So you need 0.b., and this is a posteriori, and makes ZA rely heavily on a premise that the physicalist is gonna have lots of a posteriori reasons for denying.

    to respond to the last way you put the point: you are right, we can stipulate that ‘zombie’ refers to beings that lack consciousness, and identify things as zombies without holding that they are zombie twins. And we can expand this for dealing with my epi and inter-zombie worlds. Chalmers no longer risks contradiction. But this amounts to taking the second horn of my trilemma, and reduces ZA to an a posteriori argument. Which the physicalist should be happy with.

  3. djc says:

    Thanks for this. I think I more or less agree with Pete. One might see two of the tacit premises of the zombie argument as (i) P is the complete microphysical truth about the actual world and (ii) P&~Q is ideally conceivable. This is in effect an argument schema, where ‘P’ is to be filled in by a long microphysical sentence. Of course, one needs to use a P for which (i) is true, and the corresponding filled-in version of (i), once spelled out, will be a posteriori. But it’s uncontroversial that there is some P for which (i) is true, so this doesn’t raise any problems. Maybe your worry is that it is then unclear that the corresponding filled-in version of (ii) is true (or a priori). Here the response (see the section on type-C materialism in “Consciousness and its Place in Nature”) is that the considerations supporting (ii) are sufficiently general that they make the case that (ii) is true (and a priori) for any P that is a candidate to be the actual microphysical truth. I take it that you haven’t cast doubt on these considerations here.

    It’s also worth noting that the argument requires that the relevant zombie twins are our microphysical twins, but not that they are our functional twins. I think you are right that if interactionism is true, our microphysical zombie twins (inter-zombies) need not be our functional twins. But microphysical zombie twins are all that the argument requires.

  4. Pete Mandik says:

    [[Again, the gods of the matirx exhibit their scorn for Brendan, and again he must post his comments via email.]]


    Thank you for the comments. First to try and fit my argument into what you have said. My argument rests on the idea that:

    (i) P is the complete microphysical truth about the actual world

    Should actually be,

    (i*) P is the complete microphysical, functional and psychological(in your causal-functional sense) truth abou the actual world.

    This I got just from how Kirk motivates zombies, and from what you say in your book. You state at several points that zombie twins have functional organizations (FOs) identical to us at a level that makes us behave the same way. More of a defense below.
    Now it we take (i*) the epiphenomenalist and the interactionist are going to disagree. An interactionist and an epiphenomenalist will suggest I have different FOs. In effect the argument states that the considerations supporting (ii) are so general that it allows for (ii) to hold true for epi and inter-zombies (P&~Q is ideally conceivable whether we use the interactionist or epiphenomenalist P). The problem is that this means that we have ended up with a contradiction: I cannot have the FOs posited by both the interactionist and the epiphenomenalist in their stories of (i*). If this is right, then something must be done to make the considerations supporting (ii) less general.

    To get out of this one might think you have to determine (i) before (ii), in which case we can rule out one or the other of epi and inter zombies by showing we do not have a particuler FO. But this means versions of(ii) can be ruled out a posteriori. This the physicalist should be happy with. The other two options are in the paper.

    But you deny (i*). In which case 1. it seems the physicalist just needs to stipulate that consciousness supervenes on the microphysical WITH a certain functional organization, and he avoids the argument (basically, be a sort of restricted functionalist). Thus, microphysical zombies are not enough, and we need to include functional zombies inorder to block such a move. Which puts one in line for the argument above. This I suggested in the paper. 2. If (i*) is not needed, I am curious why you thought you needed functional identity of zombie twins in the book?

    I hope the above is clear.

  5. djc says:

    Again, for the argument, (i) rather than (i*) is needed. In my book, the definition of zombie is in terms of physical duplication. I go on to suggest that they are also functional duplicates, but this isn’t definitional, or even “needed”. At that point, I was simply assuming causal closure and the falsity of interactionism. Obviously there’s not much harm in doing this if the purpose is an argument against materialism (if interactionism is true, then materialism is false automatically), though I wouldn’t do this today. My recent formulations of the argument (e.g. in “The 2D argument against materialism”) do things in terms of physical duplication throughout.

    For related reasons I don’t think that your suggestion of moving to physical/functionalism will help the physicalist. The move will only make a difference if interactionism is true, and if interactionism is true, the physicalist loses.

    Re the “contradiction”: there’s no contradiction in simultaneously holding different versions of (ii) for both an “interactionist” and an “epiphenomenalist” P. Of course there would be a contradiction in simultaneously holding both versions of (i). But no proponent of the argument needs to do the latter. One just picks the P for which (i) is actually true, and combines it with the corresponding version of (ii).

  6. Pete Mandik says:

    [[from Brendan, via email:]]


    Thank you for the reply.

    I am not quite clear on some of your points:

    “I was simply assuming causal closure and the falsity of interactionism”

    I am guessing you were assuming interactionism was false because it claims that consciousness causally interacts with the physical, not because it claims that consciousness is non physical? In which case, why would you need any other argument beyond these assumptions: if you are assuming closure AND that there is no mental-physical causation, then it seems you already rule out physicalism.


    “For related reasons I don’t think that your suggestion of moving to physical/functionalism will help the physicalist. The move will only make a difference if interactionism is true, and if interactionism is true, the physicalist loses.”

    I am not sure why holding physical/functionalism only makes a difference concerning ZA if interactionism is true. It seems true that the physical/functionalist is going to agree with the interactionist that consciousness causally interacts with the physical, but this does not require or entail the truth of interactionism. Could you clarify?

    the contradiction: your right, using (i), not (i*), there is no contradiction. And my argument uses (i*). Though, to clarify: if there is a contradiction between say, epi(i) and inter(i), and we have corresponding epi(ii) and inter(ii), we just pick between epi(i) and inter(i) based on which P is true of the actual world. This we do a posteriori. Suppose that means epi(i) is true of the actual world. Thus, we combine it with epi(ii). This would mean that inter(i) is false. Given this, what is the status of inter(ii)? Is it still ideally conceivable? If it isn’t, it seems we can rule out the ideal conceivability of versions of (ii) (or rather certain formulations of P&~Q) a posteriori. In which case the physicalist will think we can do this for any version of (ii). If it is, then what role was (i) playing in the argument, since inter(ii) is true whether or not inter(i) is true? I feel I am missing something. Again, I hope this is clear.

  7. djc says:

    1. I was assuming causal closure (i) because I took it that physics supported it (not because of any premises about the mental), and (b) because if interactionism is true, physicalism is already false, so dialectically there’s no harm in ruling out interactionism for the purpose of the anti-materialist argument.

    2. My point was not that p/f entails interactionism (it doesn’t), but that the only way that p/f can be true while microphysicalism is false is for interactionism to be true.

    3. Yes, I think inter(ii) is conceivable even if inter(i) is false. Re the need for (i): To get from the conceivability/possibility of P&~Q to the falsity of physicalism in the actual world, one needs the claim that P is the microphysical truth about the actual world (as physicalism is roughly the thesis that the microphysical truth about the actual world entails all truths).

  8. Pete Mandik says:

    [[from Brendan, via email:]]

    Thanks djc,

    That was really helpful. 1 I think I get. Concerning 2 and 3:

    2. I was not claiming that if p/f was true it meant microphysicalism was false. Rather, I thought of p/f as being a kind of microphysicalism: consciousness supervenes on the microphysical when it is functional organized in a certain kind of way. This amounts to a the mental supervening on the physical WHEN the physical is arranged in the right sort of way. Also, it seems p/f is incompatible with interactionism since the later denies closure.

    3. I suspect you will disagree with the following:

    This makes the zombie argument rely on an a posteriori premise. Thus, the argument is no longer a priori. This seems like a victory for the physicalist. Afterall, the physicalist is going to deny any (i) that fits with any version if (ii). This seems to follow from the fact that they are going to think that we have good (or are getting better) empirical reasons for a P that includes closure and mental-physical causation being true of the actual world. In which case, the truth of such a (i) seems plainly incompatible with any (ii). So, ZA only refutes physicalism once we can show that there is some (ii) for which P is true of the actual world: i.e., show the physicalist is wrong about closure and/or mental-physical causation on empirical grounds. Since you think we have good reasons for closure, that means empirical reasons for denying mental-physical causation.

    Also, at this point it seems we might run the risk of (ii), and hence ZA, becoming redundant for proving the falsity of physicalism: if we can prove closure AND deny mental-physical causation on empirical grounds (i.e ., disprove the physicalist (i)) it seems the dualist has already won. So: it seems adding (i) as a hidden premise makes ZA a posteriori and potentially redundant since disproving the truth of a physicalist (i), which ZA seems to require, is enough to disprove physicalism. The later of these two claims I am far less sure of.