Levin on Brown

My paper Deprioritizing the A Priori Arguments Against Physicalism, which was a product of the online consciousness conference, directly grew out of blog discussions I had around here shortly after I started this blog in May of 2007 (which, by the way, I just noticed, means that the 10 year anniversary of Philosophy Sucks! is coming up soon!!)…at that time I had been interested in modal arguments against physicalism but had no plans at all of writing a paper on zombies. At any rate this paper has become my second most cited paper (and is even cited by the Stanford Encyclopedia of Philosophy’s entry on Zombies (and the Wikipedia one too!)) but it is usually cited en passant, so to speak, so it is nice to see some actual discussion of the argument.

Janet Levin discusses it in her paper Do Conceivability Arguments Beg the Question Against Physicalism? which was published in the 2014 issue of Philosophical Topics that I edited as a result of the 4th online consciousness conference. At the time I was putting the issue together I contemplated writing something about it in a brief response but decided to wait. ‘Better late than never’ is quickly becoming my motto!

Levin starts with Perry’s response to the zombie argument in his 2001 book Knowledge, Possibility, and Consciousness and I think this is a good place to start. As a bit of an aside Perry’s book  has been hugely influential on me. I read it in my philosophy of mind course, with Kent Bach, in the fall of 2001 and at the time I remember feeling that the identity theory was not really given the due that it deserved and then I saw Perry defending it and it gave me hope. I even invited him out to SFSU to give a talk to the philosophy club as a result and he did. This may have been my first attempt at organizing an academic event! The paper I wrote for that class, “Sticking to the Subject: My Response to Chalmers’ Response to Perry” became my writing sample when I applied to PhD programs. I rewrote it with feedback from the class and afterwards when I approached him with this as a potential writing sample….I think I used to have this up on my website at some point when I was in Connecticut but it seems to be lost now (especially after my pre-cloud computer crash back in 2004 or 2005). I  only have a vague notion of what was in that paper and it would be interesting to see it again.

(On another tangent I also recall this book influenced a paper I wrote for my epistemology class I had in the spring of 2002 (also with Kent Bach) where I argued that adopting Perry’s view showed us how we can say that when I play Resident Evil and have thoughts like “there’s a green herb in the basement” they come out true (and how this shows a way to avoid skepticism)).

…But back to Levin’s critique. Here is what she says,

In his (2010), Richard Brown argues that the zombie argument (and its relatives) beg the question; they seem compelling only to those who already assume that qualitative properties are not physical.

One thing that I think has become clear over the years is that I was not clear enough in the original paper about my background assumptions and intentions. I had been used to arguing with people like David Chalmers and Richard Chappell and they were both very strong supporters of a priori reasoning and some version of two-dimensional semantics. So when I said that the zombie argument begged the question I meant that it begged the question against a physicalist who accepted the link between conceivability and possibility. I was trying to show that even if you grant all of the other assumptions that dualists make they still do not have an argument against physicalism. I didn’t really intend this to be a claim that the zombie argument begged the question against those who accepted type-B physicalism, or who otherwise denied the link between conceivability and possibility (as I am in some moods likely to do).

So how was this supposed to play out? As Levin says, I attempt to show this by

…presenting conceivability arguments analogous to the Zombie Argument that aim to support, rather than undermine, physicalism.  In particular, Brown argues for the conceivability of Zoombies; that is (p. 50), ‘creatures non-physically identical to me in every respect and which lack any non-physical phenomenal consciousness’, and also of Shombies; that is (p. 51), ‘creature[s] that [are] microphysically identical to me, ha[ve] conscious experience, and [are] completely physical’, and suggests that arguments with such conceivability premises will seem as compelling to physicalists as the zombie argument seems to dualists.

I can see why she says this but I would like to clear this up a bit.

First, I never intended my paper to offer any support for physicalism. I took the point to be that we did not have any good reason to think it was false, or that the a priori arguments against it did not show it to be false (currently, that is. Whatever their potential to do so in the future amounts to they do not presently constitute a reason to think that physicalism is false). Perhaps this is in fact to offer some kind of indirect support for physicalism but even so the conceivability arguments she is here discussing were aimed at showing that dualism is false. So take the pair {zombie, shombie}. If one accepts a two-dimensional semantics and one buys the general arguments against strong necessitates, then only one of this pair is truly ideally conceivable and the other is necessarily inconceivable. That much is common ground between those who accept two-dimensional semantics )and I do for the purposes of this argument…and sometimes for other purposes as well). But which one of this pair is ideally conceivable? No one has really been able to show that either one leads to a contradiction.

So, as I stressed above, given this set of background assumptions then I think the zombie argument is question begging. It begs the question by assuming that it is zombies that are truly ideally conceivable and not shombies. But if I am working inside 2D semantics then I find shombies to be conceivable and so that means zombies have to be the ones that are ideally inconceivable, even if I cannot yet say why. It is at this point, from within the 2D framework that the standoff manifests most strongly. As the Stanford encyclopedia entry on zombies suggests the best option for the dualist like Dave is to maintain that shombies are inconceivable (Dave has said this as well) but when pressed on why they are all that can be said is that many people have found it very hard to see how physicalism could be true of consciousness. But that is just to say that shombies are incredible for him, just as zombies are for me.

And this is just what Levin herself says, as we’ll see below. She goes one to say,

More precisely, Brown argues that Zoombies and Shombies—along with zombies—are all prima facie conceivable, and contends (like the theorists discussed above) that one’s antecedent theoretical commitments determine one’s convictions about which of these creatures will be ideally conceivable, conceivable ‘in the limit’.  And, though he is officially neutral about what the ultimate outcome will be, he suggests that if we were to learn, and sufficiently attend to, all the (not yet discovered) physical and functional facts about the world, we would be able to recognize that these facts do indeed entail that certain of our internal states are conscious experiences.

I would balk at this way of putting things. It is true that I am neutral about which one is ultimately truly conceivable but I do not think that it must be the case that we end up being able to make these deductions. I only think this is a possibility and that it is not ruled out by the conceivability arguments against physicalism. I do think that for different people different combinations of zombies and shombies will seem to be conceivable/inconceivable and that it is one’s background theoretical commitments that tacitly determine which is which.

I will also say that one thing that has been somewhat disappointing is that the suggestion that we may be able to make deductions a priori from physical states to phenomenal states only when we have the relevant concepts, whatever those turn out to be. So, Mary cannot do it in her room unless she has the relevant phenomenal concept (I can be neutral on what exactly these are and do not mean to endorse any account of them here). I have always thought that this was the best way to respond to the Mary case and I do not see it being discussed very much. I wish it were!

The main problem Levin has with what I say is something she points out as a flaw in Perry’s argument as well. She says,

Brown characterizes his view as a species of (what Chalmers calls) Type-C physicalism, the view that, although zombies may be conceivable now, they would be inconceivable ‘in the limit’.   But of course dualists would deny this, and it’s worth getting clear about what, on Brown’s view, could account for the eventual inconceivability of zombies.  First (p. 49) he likens the (prima facie) conceivability of zombies for us now to the prima facie conceivability of H2O in the absence of water for those living in the early 18th century, and suggests that, just as further empirical discoveries made H2O without water inconceivable, so further empirical discoveries will do the same for zombies. But this argument, like the ones discussed above, relies on a questionable analogy.  At least arguably—as noted before—what enables us (and not our 18th century counterparts) to deduce water facts from H2O facts is not merely that we have increased our empirical information, although this, to be sure, is crucial.  In addition, we, at least implicitly, appeal to certain principles that go beyond one’s empirical or methodological ‘theoretical commitments’; in particular, that the referent of ‘water’ is a  (natural) kind of stuff (e.g. the stuff that fills our lakes and comes out of our faucets or, alternatively—pointing to a lake or a puddle—that kind of stuff), and that natural kinds are to be individuated by their compositional properties (such as being composed of H2O).  But it is not clear that there are analogous principles which, when combined with our knowledge of the structure and function of the brain and central nervous system, would permit the deduction of facts about conscious experience from even a complete statement of the physical and functional facts.  And thus, it seems, it is plausible to think that even those committed to physicalism and open to further information about the structure and function of the brain will continue to find zombies conceivable.

I think I partially agree with what she is saying here. Especially with respect to the ‘extra empirical’ factors relevant to theory adoption in general.

Levin’s comments can be made especially poignant for the ‘water is H2O’ example. To anyone really interested in these issues I would recommend  reading Hasok Cheng’s book Is Water H2O? Evidence, Realism, and Pluralism, which was recommended to me after a talk I gave in Taiwan at the Academia Senica. This book really clarified a lot of the technical philosophical and empirical issues about the theoretical identity of water and H2O and I think it does highlight similar kinds of extra empirical forces at work in the history of science. I highly recommend it to anyone interested in these issues. But I am not sure why this kind of point is supposed to count against  the claim I am making. It seems to me to further support it!

If I am reading Levin right it looks like she is arguing that we may be Type B physicalists at the limit. That is, we may learn all of the physical facts and yet still find that zombies are conceivable even if we become otherwise convinced that physicalism is true. I agree that this is a possibility and this is why I think it is correct to say none of these arguments (employing zombies or shombies) are question begging tout court. They beg the question against a particular way of thinking about physicalism.

I also tend to agree that in the background of my thinking is something like what she points to about natural kinds. Especially in this case where I am interested in being as close to two-dimensional semantics as possible (because part of me thinks it is nice view to have). If one takes that point of view then consciousness is similar enough in that it can be picked out by a primary intension. We can, I think, attend to specific instances of phenomenal consciousness and say ‘that kind of stuff’ and thereby pick something out. So while I agree that my way of thinking is not the only way or forced up on in any way I am more concerned with arguing that this way of thinking is not (currently) threatened by any a priori arguments.

This is because my aim is not to establish physicalism but to show that arguments aimed at undermining it don’t do so without further assumptions. Even if you grant that the ideal conceivability of zombies would show that physicalism is false we need more than that. We need to also know that zombies are in fact ideally conceivable, and that is why homies are problematic. They must be inconceivable but then why can I, and so many others, conceive of them? The reason is, I suggested, that neither side has been able to show what contradiction is (or could be) entailed by the other side. Dualists and type B physicalists correctly point out that no one has made a successful case for why zombies are  inconceivable but, less often pointed out of late, it is also the case that no one has shown what contradiction is entailed by shombies. As far as we can tell at this point either both are ideally conceivable and so 2-dimensional semantics and modal rational fail or only one of the pair is ideally conceivable and we don’t presently know which one it is.

I think that Levin herself arrives at something like this position as well, as we’ll see. She goes on to say,

In a related argument, Brown suggests (p 55) that what could permit ‘the deduction of the phenomenal facts from the physical facts is the (for us) a posteriori discovery of identities between phenomenal and physical properties’.  But given the asymmetry discussed above, one may wonder, on behalf of dualists, what makes it plausible to think that any such a posteriori discovery could occur.  The answer, Brown maintains (pp. 55-56), is that further empirical investigation is likely to show that (contrary to Chalmers, et al and also Type B physicalists) ‘we will have discovered that phenomenal properties can be explained in broadly functional terms’, and goes on to argue that ‘this does not thereby endorse Type A physicalism.  It is just to point out that I cannot really conceive of anything else doing any explanatory work…No one has ever given anything like a proper account of what non-physical properties are or how they explain phenomenal consciousness.’

This, however—as for Perry—seems to be a matter not of zombies’ being inconceivable, but rather incredible, and the plausibility of Type C physicalism depends on the plausibility of the former, stronger, claim.  In short, there is no particular reason to think that any further discoveries of the neural structure of the brain and psychological laws governing mind-body interactions will provide information of a different sort from the information we have now.  And given that there seem to be no a priori links between physical (or physical-functional) and phenomenal concepts, and no a priori principles determining the nature or essential properties of the items denoted by our phenomenal concepts (other than that they must feel a certain way), it’s hard to see how zombies could become inconceivable—even in the limit.

Here I completely agree with her and in fact I think this is part of the point that I was trying to make.  Some people seem to find zombies conceivable, some people seem to find shombies conceivable. Since I am assuming that our reasoning capabilities are good enough for this (another common assumption between Chalmers and I for this argument) that leaves only empirical or conceptual issues left. So I agree that we haven’t yet got the story. I have argued that something like the higher-order theory will help but really the larger point is that we need a theory of consciousness and empirical data to move forward.

Again I would also point out that I think that part of the story here must invoke something about how these identities could be established. Here I think we could have a 2D version but also a version like that of Ned Block (that is a non-two-dimensional account). I am attracted to both kinds of pictures about the way these identities will be discovered.

But all of that to one side the whole point is that we cannot start from the assumption that zombies are in fact ideally conceivable. They may *seem* to be so *to you* given what *we know now* but this is not at all the same thing as their actually really being ideally conceivable, as shombies show. If one or the other were truly ideally conceivable *at this point in time* then we wold be able to show what contradiction is entailed by the zombie or shombie world. Thus, until that can be done, instead of ‘physicalism is false’ the best we get from the zombie argument is ‘it seems to me now that physicalism is false’. I hope it goes with out saying that I think the same is the case for the a priori arguments from the physicalist.

On a final note, Levin says in a footnote that Thomas Nagel in his 1965 paper Physicalism discusses cases like zoombies. Zoombies were supposed to be non-physical duplicates of me which lack consciousness. These creatures have all of my non-physical properties and yet they do not have consciousness. This seems conceivable to me, but is this what Nagel is talking about? When I went and re-read that paper it seemed like a warm up for his Bat paper and at the end he is talking about indexical information (which seems to be part of the story about how I became part of PQTI in Chalmers’ work and may be a precursor to Perry’s work). Maybe this is what she meant? I might have to re-re-read it…anyone else see the similarity?

Cohen & Dennett’s Perfect Experiment

I was re-reading Dennett and Cohen’s recent paper in Trends in Cognitive Science, “Consciousness Cannot be Separated from Function” and I am now puzzled by their view (before I go on, I would like to note that this issue of TICS also has Joe LeDoux’s paper where he mentions the Qualia Fest, and Lau & Rosenthal’s paper…all in all a great issue!).

Cohen and Dennett want to argue against phenomenological overflow, which is a debate I am currently in the middle of myself, by showing that it is essentially an unscientific view. To do this they introduce what they call the ‘perfect experiment’. They imagine that the area of the brain that is responsible for processing color is somehow allowed to function but is isolated in such a way that it cannot be accessed. The subject in this experiment is shown a blue cup, say, and will deny that they see the color even though the isolated brain area is doing what it normally does. They say,

In spite of this frank denial by subjects, theories that posit dissociation between consciousness and function would necessarily assume that participants of the ‘perfect experiment’ are conscious of the apple’s color but simply cannot access that experience. After all, the conditions these theories stipulate for phenomenal consciousness of color are all met, so this experiment does not disprove the existence of isolated consciousness; it merely provides another particularly crisp example of consciousness with- out access.

However, there is a crucial problem with this logic. If this ‘perfect experiment’ could not definitively disprove [overflow theories] theories, then what could? The subject man- ifests all the functional criteria for not being conscious of color so what would ground the claim that the subject nevertheless enjoys a special kind of consciousness: phenomenal consciousness without access consciousness

I think it is interesting to note that this kind of argument against overflow has been around for a long time. Here is a passage from Huxley’s On the Hypothesis that Animals are Automata, and Its History (from 1874),

If the spinal cord is divided in the middle of the back, for example, the skin of the feet may be cut, or pinched, or burned, or wetted with vitrol, without any sensation of touch, or of pain, arising in consciousness. So far as the man is concerned, therefore, the part of the central nervous system which lies beyond the injury is cut off from consciousness. It must be admitted, that, if any one think fit to maintain that the spinal cord below the injury is conscious, but that it is cut off from any means of making its consciousness known to the other consciousness in the brain, there is no means of driving him from his position by logic. But assuredly there is no way of proving it, and in the matter of consciousness, if anything, we may hold the rule, “De non apparentibus et de non existentibus eadem est ratio.”

The latin phrase there means something like “things that can’t be detected don’t exist.” As Huxley himself points out in the matter of consciousness, if anything, this seems correct. What sense can we make of a phenomenally conscious state that one is in no way aware of?

But it seems to me that these kinds of arguments need to deal with the mesh argument that Block defends. If the mesh argument works (I don’t think it does, at least not in favor of overflow), then we have an answer to the perfect experiment. We say that there is unaccessed consciousness in the isolated brain region because that is the (allegedly) the best over-all interpretation of the data coming in from neuroscience and psychology. Just to repeat, I DO NOT think that there is overflow but I do think that the above kinds of arguments need to deal with the mesh argument directly so I don’t find Cohen and Dnnett’s paper to advance the debate.

Zoombie Round-Up

There has been a bit of discussion of zoombies in the blogosphere of late and I want to keep track of them all so that I can respond to them so I am posting links to them.

1.) Intentional Objects‘ David Gawthorne

Richard Brown’s Zoombies and Shombies

2.) Brain Scam‘s Tony Alterman:

Zombie, Scmombie –Richard Brown’s Efforts to Ressurect Materialism

(and his reply to my reply) Return of the Zombie

3.) And then there’s Richard Chappell’s responses.

In our latest exchange he has acknowledged that the primary and secondary intensions of statements in Q may diverge but seems to think that translating those statements into “semantically neutral” language will still let the argument go through. So, just was “the watery stuff isn’t H2O” comes out true at Twin Earth, “the painful stuff isn’t C-Fiber firing” comes out true at the zombie world. But this move won’t work. Here is what Chalmers has to say about this:

Given the discussion above, one might try generating an anti-materialist argument by simply substituting primary possibility for metaphysical possibility in the original argument.

(1) P&~Q is conceivable

(2) If P&~Q is conceivable, P&~Q is 1-possible

(3) If P&~Q is 1-possible, materialism is false.


(4) Materialism is false.

On this reading, (1) and (2) are both plausible theses, but (3) is not obviously plausible. The reason is that materialism requires not the 1-impossibility of P&~Q but the 2-impossibility of P&~Q. That is, materialism requires that it could not have been the case that P were true without Q being true. This is a subjunctive claim about ordinary metaphysical possibility, and so invokes 2-impossibility rather than 1-impossibility.

A materialist might reasonably question (3) by holding that even if there is a world W verifying P&~Q, W might be a world with quite different ingredients from our own. For example, it might be that W does not instantiate true microphysical properties (those instantiated in our world), such as mass and charge, but instead instantiates quite different properties: say, pseudo-mass and pseudo-charge, which stand to mass and charge roughly as XYZ stands to H2O. Likewise, it might be that W does not lack true phenomenal properties, but instead lacks quite different properties: say, pseudophenomenal properties. If so, then the possibility of W has no bearing on whether true microphysical properties necessitate true phenomenal properties. And it is the latter that is relevant for materialism.

Still, it may be that the gap between 1-possibility and 2-possibility could be closed. In particular, when a statement S has the same primary intension and secondary intension, then a world will verify S iff it satisfies S, so S will be 1-possible iff it is 2-possible. If P and Q both have primary intensions that coincide with their secondary intensions, then so will P&~Q, and we could run the following argument:


(1) P&~Q is conceivable

(2) If P&~Q is conceivable, P&~Q is 1-possible

(3) If P&~Q is 1-possible, P&~Q is 2-possible.

(4) If P&~Q is 2-possible, materialism is false.


(5) Materialism is false.

Here, the truth of (3) requires that both P and Q have primary and secondary intensions that coincide. (from The 2-D Argument Against Materialism)

Ut vos es Bellator Victus Mortuus Tempus Fugit

Yes, indeed, time does flee when you are fighting the living dead!

It has been a year since the original zombies wars broke out around here, which led to all of the zoombie and shombie action. I was recently looking over some of these old posts and the debate between Richard Chappell and myself as I get ready to do the re-write for the Journal of Consciousness Studies (where this paper and others from Consciousness Online are slated to appear).

The basic claim that Chappell makes is that the zoombie argument is not really a parody of the zombie argument. The basic reason for this is his claim that there is an important dis-analogy between the zombie argument and the zoombie argument. This dis-analogy arises because the completed micro-physical description of the world contains qualitative facts, if it does at all, only implicitly. This opens up the conceptual gap between the physical and the qualitative in a way that leaves the physicalist susceptible to the conceivability argument.

The dualist on the other had think that the qualitative facts are primitive and must be added to a completed micro-physics explicitly. Since the qualitative is not reducible to any base facts that are not themselves qualitative the dualist is not susceptible to conceivability arguments in the same way that the physicalist is. Another way to make the point, and a way that I think Richard prefers to make the point, is in terms of asking what follows from a completed micro-physics. The original zombie argument tries to show that the qualitative facts do not follow from the complete physical facts. The zoombie argument doesn’t work this way. We do no build up the non-physical facts and then ask whether or not the qualitative facts follow from them. This is because the dualist claims that the qualitative facts have to added explicitly. There is no issue of whether the follow from any other kind of fact since they are taken to be primitive.

Now, clearly, there is this dis-analogy between the two arguments. But does it follow, as Chappell alleged, that this dis-analogy makes it “daft” to mount a conceivability argument against the dualist? Or, another charge from Chappell, does this dis-analogy make the zoombie argument a bad parody of the original zombie argument (actually he said that it was irrational to think so, but let’s not dredge that stuff up) or show that I have misunderstood the original zombie argument (yet a further gem from Chappell)?

To start with the first question, the answer is obviously ‘no’.  A conceivability argument just starts from the claim that something is conceivable and reasons to that thing’s being possible. If something follows from that possibility then fine.If not fine. In this case it is conceivable that NP, being the complete set of non-physical facts about the actual world, obtain without any qualitative properties included in NP. If this is conceivable then it is possible and if it is possible dualism is false. Is this a bad parody of the zombie argument? Only if you think a parody must be identical in every respect to the original. The point is that zoombie argument is like the zombie argument in the right respects; in particular in asserting the conceivability of p and reasoning from there. The fact that what is at issue in the zoombie argument is not “what follows from what” but is rather “what must be included in the complete non-physical description of a world that is non-physically identical to the actual world?” is irrelevant. All that maters is that the possibility of zoombies shows that qualitative properties are not non-physical properties.  Finally, does this show that the proponent of the zoombie argument misunderstands the zombie argument? I don’t think so. The zombie proponent holds that zombies are conceivable because there is a gap between our physical/functional concepts and our qualitative concepts; or in other words that there is no reason as of now to think that qualitative properties are physical properties. The physicalist like me will claim that this gap will be abolished as we reach the ideal limit and so zombies are prima facie conceivable but not ideally conceivable. The zoombie proponent holds that zoombies are conceivable because there is no reason to think that qualitative properties are non-physical and so no reason to think that they must be included in a complete description of a non-physical duplicate of our world. Clearly both of these (zombies and zoombies) can’t be ideally conceivable so a priori arguments won’t help us decide between theories of consciousness.