This page is mostly for brief responses to published or forthcoming articles that discuss my work. I've starred [*] the more technical entries so nonphilosophers can skip them. (The first entry is probably the best place for nonphilosophers to start.) These responses are "unofficial", but if you'd like to use them in an article, feel free to ask. See also online discussions of my work for some further papers and other discussion (without responses).
Table of contents: A link on the author's name goes to an entry further down on this page.
Journal of Consciousness Studies symposium on the "hard problem". (Articles by Baars, Bilodeau, Churchland, Clark, Clarke, Crick & Koch, Dennett, Hameroff & Penrose, Hardcastle, Hodgson, Hut & Shepard, Libet, Lowe, MacLennan, McGinn, Mills, O'Hara & Scutt, Price, Robinson, Rosenberg, Seager, Shear, Stapp, Varela, Velmans, Warner.)
*Philosophy and Phenomenological Research symposium on The Conscious Mind. (Articles by Hill & McLaughlin, Loar, Shoemaker, Yablo.)
Reviews of The Conscious Mind.
Katalin Balog, Conceivability, possibility, and the mind-body problem.Philosophical Review108:497-528, 1999.
Balog appeals to zombies in order show that Jackson's and my anti-materialist arguments are self-defeating. She argues that a zombie could make the same arguments, with true premises but a false conclusion (materialism is true in the zombie world), and concludes that the arguments are invalid. She locates the problem in the inference from conceivability to possibility, or in the thesis that materialism requires a priori entailment.
This is an intriguing argument, but I think the problem with it is clear. Balog's parallel argument requires that a zombie's claim "I am conscious" is true; otherwise the argument doesn't get off the ground. Balog supports this by suggesting that the zombie's "consciousness" concept will pick out a physical/functional property to which it is causally related. But I think it is much more plausible that the zombie's claim is false. The easiest way to see this is to consider an argument in the zombie world, perhaps between Zombie Chalmers and Zombie Dennett. Zombie Chalmers says "Qualia exist", Zombie Dennett says "Qualia do not exist". Balog's analysis implies that in the zombie world, Zombie Chalmers is right. But this seems wrong. Surely in the zombie world, at least, Zombie Dennett is right.
See "The Content and Epistemology of Phenomenal Belief" for more on this sort of thing.
Tim Bayne, Chalmers on the justification of phenomenal judgment. Philosophy and Phenomenological Research 62:407-19.
Bayne focuses on the epistemological discussion in Chapter 5 of my book. There I argue that a reliabilist account of phenomenal knowledge can't work, since a reliabilist account can't deliver certain knowledge. When knowledge is grounded only in a reliable connection, we can't be certain that the reliable connection holds, so we can't rule out alternative skeptical scenarios; but we are certain that we are conscious, and can rule our alternative skeptical scenarios. Bayne argues that the same applies to my "acquaintance" account. He notes that I accept in the book that acquaintance alone doesn't suffice for full justification (acquaintance alone is compatible with fallibility, e.g. in cases of inattention, etc). So various further "background conditions" (e.g. concerning attention, etc) are required for full justification. But we can't know for certain that those background conditions hold; so we can't know for certain that we are justified. So by parity of reasoning, the acquaintance model can't deliver certain phenomenal knowledge, either.
I think this argument rests on a natural misreading of my discussion of reliabilism. The argument is not: reliabilism can't deliver certainty that we are fully justified, certainty that we are fully justified is required for certainty, so reliabilism can't deliver certainty. That argument would rest on a problematic "CJ" thesis (analogous to the "KJ" thesis that knowledge requires knowledge that one is justified, but with certainty instead of knowledge). Rather, the argument is: when knowledge is grounded only in a reliable connection, we can't be certain that the reliable connection holds, so we can't rule out alternative skeptical scenarios about what's at the other end of the connection, so we can't have certain knowledge.
The first argument (resting on the CJ thesis) just might deliver an analogous argument against the acquaintance model. But I'll respond by denying the CJ thesis (even accepting CJ, one could also argue that "positive" certainty of justification doesn't require "negative" infallibility). And an argument analogous to the second argument won't work. Even if an acquaintance theorist accepts that we can't be certain that we are justified, that merely shows that one can't rule out skeptical scenarios in which one's belief in the experience is not fully justified; it doesn't show that one can't rule out skeptical scenarios in which the experiences are not present. Presumably an acquaintance theorist can hold that we're certain about qualia but not about such cognitive matters as full justification, so that we can rule out skeptical scenarios without qualia, but not those without full justification. (To resist this would require the dubious CJ claim.) The analogous move is not open to a reliabilist: by its nature, reliabilism can never deliver "knowledge beyond skepticism". Again, see "The Content and Epistemology of Phenomenal Belief" for more.
*George Bealer, Modal epistemology and the rationalist renaissance. In (T. Gendler and J. Hawthorne, eds) Conceivability and Possibility. Oxford University Press, 2002.
Bealer also says that the claim that "water" picks out XYZ relative to some worlds considered as actual is a violation of English, because the English expression picks out H2O in all worlds. But the latter theoretical claim is wholly grounded in intuitive claims about the referent of "water" in subjunctively considered counterfactual situations, and these claims about subjunctive evaluation are perfectly compatible with the claims that "water" behaves as I say it does under epistemic evaluation. He also worries about the claim that "water" picks out something other than water in some centered worlds considered as actual. This and most of the semantic issues that Bealer discusses are addressed in "On Sense and Intension", especially the second half of section 7.
(N.B. None of these semantic points bear on the modal epistemology. Bealer gives no argument against the claim that expressions can be associated with intensions in the way I describe (whether or not this association is "semantic"), and he gives no argument against the thesis e.g. that a posteriori sentences have 1-intension that is false in some world. This is all that is needed for the modal framework.)
Bealer also addresses the use of the 2-D framework to argue against physicalism, arguing that it fails because physical terms have different primary and secondary intension. In The Conscious Mind I suggested that these terms have the same intensions; in more recent work I deny this. But either way (as pointed out in the book and in recent work), this loophole leads only to the view I call "type-F monism" or "panprotopsychism". (Bealer doesn't like the latter term, but it's the view rather than the term that matters.) I've never claimed to have an argument against this view, and in fact am sympathetic to it, so there is no objection here. The formalization of the argument given in the appendix of "Does Conceivability Entail Possibility" and section 6 of Consciousness and Its Place in Nature" makes all this pretty clear.
Bealer puts forward his own positive view, which appeals to "semantically stable" expressions to close the gap between apriority and necessity. I think this doesn't quite work: for reasons discussed in section 5(v) of DCEP and in "The Foundations of 2-D Semantics", some semantically stable sentences are necessary a posteriori. Also, the best anti-materialist conclusion one can get from this strategy is a rejection of psychophysical property identities, but that rejection is compatible with the truth of many other robust forms of materialism. Still, with a few adjustments, Bealer's appeal to semantic stability is in fact quite compatible with a two-dimensional approach.
Mark Bishop, Dancing with pixies: Strong artificial intelligence and panpsychism. In (J. Preston and J.M. Bishop, eds) Views into the Chinese Room. Oxford University Press, 2002.
Mark Bishop, Counterfactuals cannot count: A rejoinder to David Chalmers. Consciousness and Cognition, 11:642-52, 2002.
In response: it's not at all obvious just how the gradual transition will work in a combinatorial state-automaton, but let's grant that something like it is possible. Then this process will gradually transform a counterfactually-sensitive system into a "wind-up" system that implements just one run. This plausibly will affect the system's cognitive states (such as beliefs), gradually destroying them, so the fading qualia argument (which relies on preserving cognitive states) doesn't apply. Bishop addresses a version of this point by saying that "input sensitivity" can't be crucial, since a non-input-sensitive system (e.g. a blind system, or one with constant input signals) could be conscious. But what really matters is counterfactually-sensitive cognitive processes, which the blind system still has (it isn't a wind-up system). Compare an ordinary human, a blind human, and a humanoid system preprogrammed to go through a single specific series of brain states. Here it seems most plausible to say that the first two systems are conscious but that the third is not.
Maybe it is initially hard to see just how mere counterfactual sensitivity can affect an intrinsic property such as consciousness. But it's hard to see how any physical property can affect consciousness. Given that the connection between physical properties and consciousness is contingent, the hypothesis that some of the relevant physical properties are dispositional rather than categorical doesn't seem all that hard to accept. And it seems to fit well with our intuitive ascriptions of consciousness.
*Ned Block and Robert Stalnaker, Conceptual analysis, dualism, and the explanatory gap. Philosophical Review 108:1-46, 1999.
Frank Jackson and I have written a response to this article: "Conceptual Analysis and Reductive Explanation" (Philosophical Review 110: 315-61, 2001).
*Andrew Botterell, Conceiving what is not there. Journal of Consciousness Studies 8:21-42, 2001.
*David Braddon-Mitchell. Qualia and analytic conditionals. Journal of Philosophy 100:111-35, 2003.
My reply to this is basically the same as my reply to the very similar arguments by Hawthorne and Stalnaker below. At best, this semantic hypothesis explains why it is conceivable that a zombie world is metaphysically possible. But the key conceivability claim was not this but rather that it is conceivable that a zombie world is actual (i.e. P&~Q is not ruled out a priori, where P is the microphysical truth about our world and Q is the claim that someone has qualia). If this conceivability claim is correct, Braddon-Mitchell's semantic hypothesis is false (as this hypothesis entails that P&~Q is ruled out a priori). Braddon-Mitchell responds very briefly by simply denying the conceivability claim, but he gives no argument for this denial of what seems to be an intuitive datum. The view has equal trouble accommodating the apparent epistemic datum in the knowledge argument.
Braddon-Mitchell suggests at one point that his hypothesis is reasonable because if we discovered that our world were entirely physical (if an oracle told us this, for example), we would not then deny that qualia exist, but would say that qualia are physical. I agree that we would do this, but I don't think this provides any support for Braddon-Mitchell's hypothesis. Our reaction in this case would clearly be an inference from our knowledge that qualia exist, and this knowledge is a posteriori. If this a posteriori knowledge were somehow suspended, there would be no good reason to make the inference. If so, there is no a priori entailment from the claim that our world is purely physical to the claim that qualia exist.
(Update: see also the discussion in section 5 of "The Two-Dimensional Argument Against Materialism".)
*Anthony Brueckner. Chalmers' conceivability argument for dualism. Analysis 61:187-93, 2001.
*Alex Byrne, Cosmic hermeneutics. Philosophical Perspectives 13:347-83, 1999.
Byrne argues against the a priori entailment of most macro facts by micro facts. Along the way it addresses two of my arguments: the argument from conceivability, in which I argue that a scenario in which the micro facts are the same but the macro facts are different is inconceivable; and one of the arguments from epistemology, in which I argue that a good enough reasoner given only the micro facts would be in a position to ascertain the macro facts. Byrne's main objection to the conceivability argument is that given that we don't yet know just what the micro facts are, we can't know that the scenario in question is inconceivable. His main objection to the epistemological argument is that the strategy I outline relies on the reasoner reaching certain intermediate high-level conclusions, and that the entailment of these conclusions is just as questionable as the entailment of the original macro facts.
I think Byrne misses a key aspect of the arguments in the book (which are admittedly brief). The argument in general is to argue that the micro facts specify micro structure and dynamics, that micro structure and dynamics a priori entail macro structure and dynamics, and that macro structure and dynamics entails most macro facts (except those that depend on mentality or indexicals). The middle level of macro structure and dynamics can be thought of as a "geometric" characterization of the structure and evolution of macro objects -- their overall shapes, masses, positions, causal relations, etc. As long as microphysics has information about mass, position, etc, of micro objects, the information about macro objects will be easily derivable from the micro information (e.g. by considerations about macroscopic mass densities in various locations, and so on). So this point is robust over microphysical theories. The second entailment is made plausible not least by the fact that this sort of macro structure and dynamics (plus facts about appearance, which I can also help myself to) is all the information we have to go on in ordinary perception.
Byrne's objection to the conceivability argument can be rebutted by noting that however microphysics turns out, it will be inconceivable that there be the microstructure without the macrostructure, and it will be inconceivable that there be the macrostructure without the macro facts. His objection to the epistemological argument can be rebutted by noting that the reasoner will easily be able to ascertain facts about macrostructure, and from facts about macrostructure (plus facts about appearance) they will be able to ascertain the macro facts in question (just as we can through perceptual information). Byrne does not address my second epistemological argument, which is that the absence of a priori entailment of macro facts will lead to a sort of skeptical problem concerning those macro facts (two epistemic possiblities with the same micro facts and different macro facts would "look" just the same) that we are not in fact faced with in most cases, except in the case of consciousness. All this is now discussed at greater length in "Conceptual Analysis and Reductive Explanation".
*Alex Byrne, Chalmers on epistemic content. SOFIA conference on Metaphysics of Mind, December 2001.
In response: I wouldn't use "think about an object" this way. On normal usage, thinking about an object requires that the object be the extension of a concept involved in a thought (and maybe requires a bit more, such as acquaintance), but there's no rigidity requirement. Nevertheless I think we can come halfway to satisfying Byrne's requirement. The 2-intensions of our concepts of external objects can be rigid, although the 1-intensions usually cannot: one might say that such concepts can be subjunctively rigid but not epistemically rigid. I don't think this is counterintuitive, or has bad consequences. Byrne doesn't really suggest any bad consequences, apart from saying that it seems wrong to say we can't think about (e.g.) Bismarck. I agree that seems wrong, but I think that's a problem with a reading of "think about" that requires epistemic rigidity, not a problem with my account.
Byrne suggests that the account divides epistemic space in the wrong way. Say that W1 is a world where the primary intension of "Chalmers" picks out Russell, who is a philosopher, and in which Chalmers himself is not a philosopher. Then on my account, the epistemic content of "Chalmers is a philosopher" is true at the scenario W1, but Byrne thinks that it ought to be false there, or at least that it would be false if the thought were truly about Chalmers. I agree that it would be false there if the thought were "about Chalmers" in the sense requiring epistemic rigidity, but I don't think this applies to the ordinary sense. Intuitively, "Hesperus" and "Phosphorus" thoughts are both about Venus, but it's epistemically possible that Hesperus is not Phosphorus, and intuitively there are points in epistemic space (although perhaps not in subjunctive space) at which the thought is false. So Byrne's suggested connection between "thinking about an object" (in the ordinary sense) and divisions of epistemic space seems too strong.
*Alex Byrne, Intentionalism defended. Philosophical Review 110:199-240, 2002.
Alex Byrne and Ned Hall, Chalmers on consciousness and quantum mechanics. Philosophy of Science 66:370-90, 1999.
I have to eat some crow on this one: there is a mistake in the reasoning in the book. The mistake (first pointed out to me by Jacques Mallah) is a technical one in the discussion on p. 350 arguing that a superposition implements the original computation. I establish a mapping between states of the superposed system and the original implementing states of the original system by projection onto the hyperplane of the original system. I then say that since the Schrodinger equation is linear, the state-transitions among the superposed states as defined will mirror the transitions among the original states. But this last step does not follow. In effect, it presupposes that projection commutes with Schrodinger evolution, which is false in general.
How to fix this up? One needs to find a special class of superpositions, with associated projections (or other relevant operators) onto subspaces, such that within this class, projection commutes (or approximately commutes) with Schrodinger evolution. This is what's needed to capture the Everett-style idea that evolution in distinct "components" of the superposition will sometimes be more or less independent. I think that ideas drawn from decoherence theory may be useful here -- so the result won't hold for superpositions in general, but will hold for "decoherent superpositions". This makes for some interesting links with other ideas, but I haven't yet worked out all the details. If you have any ideas about this, please let me know!
Byrne and Hall have some other objections to my treatment, but I think that this is easily the most serious. I think they mislocate the source of the absurd consequence in a principle about causal relevance (ultimately stemming from preferred bases), rather than in the mistake above, which they don't discuss. If the above problem can be fixed (e.g. via the decoherence idea), I think all the causal relevance one needs will follow, on a natural treatment of causal relevance. And the independence of evolution in the subspaces will deliver all the "preferredness" of bases that one needs.
*Alex Byrne and Jim Pryor, Bad intensions. Barcelona conference on Two-Dimensionalism, June 2001.
Rather, my account requires that people have a conditional ability to identify the referent of a term given nontrivially sufficient information about the world (e.g. to identify the referent of 'water' given a full enough non-'water'-involving description of the world), and idealized rational reflection. This involves the (idealized) a priori knowability of certain conditionals ("if such-and-such, then N is such-and-such"), but it doesn't require any particular knowledge of the form "N is the D". And Byrne and Pryor don't give any reason to doubt that we have this conditional ability. At one point they suggest that even if we can evaluate these conditionals, this may reflect a posteriori rather than a priori reasoning. But they don't give any argument for this claim. See "Conceptual Analysis and Reductive Explanation", especially section 3 and section 5(6), for reasons to believe in the apriority of the conditionals.
Byrne and Pryor try to apply Kripke's epistemic argument against descriptivism to the 2-D framework, but this application mostly turns on the false assumption about associated properties above. They also make two objections to the metalinguistic understanding of the primary intensions of names (e.g. the idea that 1-intension of "Godel" picks out something like the person called "Godel" by the person from whom the speaker acquired the name). B&P say this requires the speaker (implausibly) to grasp sophisticated semantic concepts. But that's false: the conditional account does not require that they possess those concepts (again, B&P are working with the "identifying knowledge" model). They also say that this account can't explain what goes on in Frege cases: e.g. when someone discovers that Bob Dylan is Robert Zimmerman and is excited, what excites her is the knowledge that her old prom date is a famous singer, not that the person called 'BD' is the person called 'RZ'. But it's often the case that when someone is excited to learn that P, what excites them is not the content of P but the content of some associated Q (someone tells me that the guy in the corner is Joe's father, I get excited because I just talked to a famous mathematician). So this point tells us little about the content of "Bob Dylan is Robert Zimmerman".
Patricia Churchland, The hornswoggle problem. Journal of Consciousness Studies 3: 402-8, 1996.
Paul Churchland. The rediscovery of light. Journal of Philosophy 93:211-28, 1996.
This parodies various antireductionist arguments, including my "hard problem"/"easy problem" arguments, by giving analogous arguments that someone might have given about "light" a couple of centuries ago. A response is here (contained within my JCS response).
Daniel Dennett, Facing backward on the problem of consciousness. Journal of Consciousness Studies 3: 4-6, 1996.
Daniel Dennett, The fantasy of first-person science. Northwestern University, February 2001.
Dennett's central argument is that in cases where reports about consciousness go wrong, introspection will also go wrong, so heterophenomenology misses nothing that phenomenology covers. I think this mislocates the debate: the reasons for holding that explaining the reports doesn't explain everything had nothing to do with the fallibility of reports. Even if reports were infallible, the same issue would arise: to explain reports is not to explain the experiences that the reports are reports of.
Dennett says that this view requires a non-neutral attitude toward a subject's reports: taking them uncritically as a guide to their subject matter. And he says scientists have to take a neutral attitude (taking reports themselves as data, but making no claims about their truth), because reports can go wrong. But this misses the natural intermediate option that Max Velmans has called critical phenomenology: accept verbal reports as a prima facie guide to a subject's conscious experience, except where there are specific reasons to doubt their reliability. This seems to be most scientists' attitudes toward verbal reports and consciousness: it's not "uncritical acceptance", but it's also far from the "neutrality" of heterophenomenology. On this view, we're not interested so much in reports as data to be explained in their own rights (though we might be in part interested in that). Rather, we are interested in them as a (fallible) guide to the first-person data about consciousness that we really want to be explained. This way of thinking about things accommodates a major role for reports, while preserving the distinction between explaining the reports (and third-person data in general), and explaining the experiences (first-person data) that they are reports of.
Dennett makes something of the fact that his method appeals to a subject's beliefs rather than just to reports, but the same issues arise. On Dennett's account of beliefs, to believe that P is roughly to be disposed to report that P and to behave in appropriate ways; and explaining such beliefs is in large part to explain the associated dispositions. If one takes this view of beliefs, everything above applies to beliefs as well as reports. If one does not take this view, it is no longer obvious that third-person methods can explain our specific beliefs about consciousness, as these beliefs may themselves have a phenomenal element: see my paper on phenomenal belief for discussion.
Dennett "challenges" me to name an experiment that "transcends" the heterophenomenological method. But of course both views can accommodate experiments equally: every time I say we're using a verbal report or introspective judgment as a guide to first-person data, he can say we're using it as third-person data, and vice versa. So the difference between the views doesn't lie in the range of experiments "compatible" with them. Rather, it lies in the way that experimental results are interpreted. And I think the interpretation I'm giving (on which reports are given prima facie credence as a guide to conscious experience) is by far the most common attitude among scientists in the field. Witness the debate about unconscious perception among cognitive psychologists about precisely which third-person measures (direct report, discrimination, etc) are the best guide to the presence of conscious perception. Here, third-person data are being used as a (fallible) guide to first-person data about consciousness, which are of primary interest. On the heterophenomenological view, this debate is without much content: some states subserve report, some subserve discrimination, etc, and that's about all there is to say. I think something like this is Dennett's attitude to those debates, but it's not the attitude of most of the scientists working in the field.
In any case, the fundamental reasons for rejecting the heterophenomenological view lie prior to these questions about experiments. Fundamental question: Does explaining behavior and other third-person data suffice on its own to explain conscious experience? I think there are overwhelming grounds to say no (see e.g. "Moving Forward on the Problem of Consciousness"). The question is then: given that this is so, can there be a science of consciousness anyway? I think so, as long as we accept that science can deal with first-person data, and as long as we allow that verbal reports and the like can be used as an indirect guide to the first-person data about consciousness.
William Greenberg, On Chalmers' principle of organizational invariance and his `dancing qualia' and `fading qualia' thought experiments. Journal of Consciousness Studies 5:53-58, 1998.
Greenberg makes the same sort of point against my "dancing qualia" argument as that made by van Heuveln et al (below): i.e., because there are two "instantiations" of the system, there is no "inner eye" before which the experiences dance. My response is similar: I think the existence of a single subject is enough to make the argument go through. Greenberg responds to my discussion of this in the book by granting that the systems might have an experienced sense of personal identity, and will report continued personal identity, but noting that that's not enough. I agree that's not enough, but this misses the point of making the change small. Presumably the "sense" and the report will also be preserved in a 75% replacement, but actual personal identity won't be; whereas replacing a small subsystem seems enough to preserve not just sense and report but identity.
Greenberg also objects to my talk of "tepid pink" as the consequence of "fading" red qualia, but obviously this is quite inessential to the argument. And he says the argument assumes that functionally equivalent silicon replacement of neurons is possible, but that isn't so; see my JCS response (#3.5), for more on that.
*John Hawthorne, Advice to physicalists. Philosophical Studies 109:17-52, 2002.
The second part of the paper argues that the epistemic gap could be explained by the suggestion that phenomenal concepts are a certain sort of demonstrative concept. I don't think this works: as I've argued elsewhere, (pure) phenomenal concepts just don't behave like demonstrative concepts (see e.g. my papers on phenomenal belief and on Perry). The third part addresses the "panprotopsychist" option and argues that it's good enough for the physicalist and doesn't leave the intrinsic properties "hidden". I think that whether the view is "physicalist" doesn't matter much, but I do think there's an epistemic limitation here, brought out by the fact that multiple hypotheses about the nature and distribution of intrinsic properties are epistemically compatible with all of our physical knowledge. Hawthorne doesn't address this line of reasoning.
*John Hawthorne, Direct reference and dancing qualia. In (T. Alter and S. Walter) Phenomenal Concepts and Phenomenal Knowledge (OUP, 2006).
I think that whether or not such a thought is a priori justified, it is a priori justifiable in the sense that appeals to an idealization of cognitive functioning. Obviously, the "dancing qualia" scenario involves highly unusual cognitive functioning. An analogy: someone arranges Fred's mind so that sometimes he thinks 5+7=12 (the usual way), sometimes he thinks 5+7=14, without being able to tell the difference. Here, perhaps neither thought will be a priori justified, but the first remains a priori justifiable. As usual, failures of justification due to non-ideal cognitive functioning don't defeat idealized apriority.
Hawthorne very briefly addresses this line by saying that R1=R2 is intuitively more like Hesperus is Phosphorus than Hesperus is Hesperus. But of course there are too many respects of similarity and difference (e.g. a priori justification, knowledge) for this to establish anything about a priori justifiability. One might instead ask whether R1=R2 here is more like 7+5=12 in the case above or Hesperus is Phosphorus. I think the answer is plausibly the former.
A distraction is that Hawthorne expresses R1=R2 as "thus = thus". This is misleading, as it suggests demonstrative phenomenal concepts, and it's clear that the corresponding identity involving demonstrative phenomenal concepts (E1=E2, in the language of CEPB) is a posteriori, requiring introspection for its justification. R1 and R2 by contrast are pure phenomenal concepts, and introspection plays no justificatory role here.
*Christopher Hill. Chalmers on the a priority of modal knowledge. Analysis 58:20-26, 1998.
Hill notes that the arguments in my book depend on the claim that we have a priori access to logical possibility of statements according to their primary intensions, and suggests that my arguments for this claim in Chapter 2 beg the question, because they assume a priori access to the space of logically possible worlds (which Hill disputes).
I think Hill misunderstands the passages he discusses. At this point (p. 68) I have argued that the two-dimensional framework handles all the standard Kripkean a posteriori necessities, with just one space of worlds (the conceivable worlds) and two intensions. The standard examples thus give no reason to deny that conceivability of a world (as opposed to conceivable truth of a statement) implies possibility. I then stipulate that a "logically possible" world is an (ideally) conceivable world, noting that any reasons for believing in a narrower class of "metaphysically possible" worlds must be independent of the standard reasons. Given this stipulation, the claim about access to logical possibility that Hill finds question-begging is relatively trivial.
The real locus of disagreement is whether the class of metaphysically possible worlds is narrower than the class of logically possible worlds (so construed). This issue is discussed in the section on "strong metaphysical necessity" in Chapter 4; the passages Hill discusses in Chapter 2 are irrelevant. For more clarification of the structure of my argument here, see "Materialism and the Metaphysics of Modality", section 3.1.
Christopher Hill. Imaginability, conceivability, possibility, and the mind-body problem. Philosophical Studies 87:61-85, 1999.
This interesting paper argues that there is a cognitive explanation for our conceivability intuitions about consciousness (in terms of different modes of access to physical and phenomenal states), so that these conceivability intuitions don't imply anything about metaphysical possibility or about ontology. I address this matter in section 3.4 of "Materialism and the Metaphysics of Modality", in which I respond to a related paper by Hill and McLaughlin.
*Jenann Ismael. Science and the phenomenal. Philosophy of Science 66:351-69, 1999.
Ismael argues that phenomenal properties are ostensively identified properties of physical space, used to "interpret the map" associated with physical theories. The epistemic gap between physical and phenomenal in effect stems from the gap between knowing the objective "co-ordinates" of the property and knowing that the property is the ostended one (analogous to the gap between knowing the co-ordinates of a location and know that that location is here).
The philosophy of science here is very interesting, but all the argumentive work against me is done by the analogy with indexicals such as "I" and "here". I discuss this analogy in the book, but it's a popular strategy (e.g. it plays a role in ? [not found] John Perry's recent work), so it may be useful to say more here. I think the analogy can't work, for a number of reasons. First, indexical information "disappears" from the third-person point of view: an objectively omniscient third party has no significant epistemic ignorance regarding any indexical facts about another. But phenomenal information does not: a physically omniscient third party may be epistemically ignorant of phenomenal facts about another. So phenomenal information is "objective" in a way that indexicals are not. This can also be brought out by the fact that the identity between indexically identified states and phenomenal states is cognitively significant, as discussed in "The Content and Epistemology of Phenomenal Belief".
Indexical ignorance always stems from ignorance of "locating information" concerning the identification of certain objective parts of the map with "I" and "here" and "this". (We can think of this information as coming in the forms of "arrows" pointing to certain qualitative parts of the map, giving information about the "center" of a centered world.) Once an agent has all the qualitative information and all the locating information, they're in a position to know everything. But we can give Mary, or any third party, all the locating information we like about their own location in the world -- even information about the phenomenal states they ostend, if you like (though there are tricky issues here) -- and they'll still be ignorant about phenomenal states of others (what it's like to see red, what it's like to be a bat). So locating information cannot cross the relevant epistemic gap.
Ismael handles the zombie scenario by noting that imagining C-fiber firings (etc) that no-one is conscious of is just like imagining bits of space that are not "here" for anybody. But I think this point backfires: the space scenario (unlike the zombie scenario) involves a change in objective physical matters. Given all the physical information, one will know whether or not the bit of space is "here" for anyone (modulo worries about consciousness); but one will not analogously know whether the physical states are "conscious" for anyone. As before, there's no epistemic gap for facts about the objectified "x is at P" relation, but there is for the objectified "x is conscious of Q" relation.
*Mark Johnston. Manifest kinds. Journal of Philosophy 94:564-83, 1997.
This article mostly argues that the water-H2O relation (and such) should be seen as constitution rather than identity. An appendix applies the discussion to issues about the explanatory gap raised by Frank Jackson and me. He argues that when things are put in terms of constitution rather than identity, the problems fall away. For example, the identity of pain with C-fibers doesn't follow a priori from a story about C-fibers playing a relevant causal role, but the constitution of pain by C-fibers does.
In response: first, identity plays no special role in my arguments. All that matters is that the macro facts are entailed by the micro facts. In fact I'm happier with talk of constitution than with talk about identity here. Johnston thinks the arguments don't go through when constitution is invoked, as he claims that "if C-fibers explain the causal powers of pain, then C-fibers constitute pain" is a priori. Response: whether or not this is a priori, it is not the relevant conditional for reductive explanation. What matters is entailment of the pain facts by the physical facts. Johnston's conditional can't ground such an entailment, as it builds claims about "pain" into the antecedent. For it to do the job, we would have to replace "the causal powers of pain" by a list of specific, neutrally characterized causal powers (e.g., being caused by tissue damage, causing avoidance behavior, etc.). But then the modified conditional clearly will not be a priori, as it's epistemically possible that any such list be satisfied without pain. In the cases of "water" and such, by contrast, it's much more plausible that there is such a list that makes the analogous conditional a priori. Johnston says that derivability may fail even in other cases (slipperiness, redness, wetness), as we can't derive facts about appearance a priori. I agree with this, but think that this failure is a consequence of the underivability of facts about consciousness. As I say in the book, those facts are still derivable "modulo consciousness".
Robert Kirk. Why there couldn't be zombies. Aristotelian Society, Supplementary Volume 73:1-16, 1999.
Kirk is a onetime zombiephile turned zombiephobe. He argues that (1) on the zombiephile view, we are compounds of a physical zombie and nonphysical qualia, (2) a zombie can't detect nonphysical qualia, and qualia can't detect nonphysical qualia, (3) their compound (i.e. us) can't detect nonphysical qualia, either, so (4) the zombiephile view has the absurd consequence that we can't detect qualia.
I could take issue with (1), but I think the deepest problems are with the second conjuct of (2) and with (3). I think there's a case to be made that our phenomenology alone implies knowledge of qualia; e.g. a disembodied being with exactly my phenomenology would know it had qualia. If so, the second conjunct of (2) is in effect false. The problem is that Kirk assumes a "thin" notion of the phenomenal on which it implies nothing cognitive; I am increasingly inclined to think this is wrong. Further, the step from (2) to (3) is invalid: it's simply false that if neither A nor B can know that P, then a compound of A and B cannot know that P (e.g., on a physicalist view, split the brain in two jagged halves). So even if a zombiephile accepts (2), they might reject (3), presumably by holding that phenomenal knowledge is supported by some combination of a physical and a phenomenal state (e.g. by a physical judgment in combination with acquaintance with a quale, as I suggest in the book). So the argument can't do the work it needs to do.
Noa Latham. Chalmers on the addition of consciousness to the physical world. Philosophical Studies 98:71-97, 2000. (Abstract)
Latham gives a number of arguments against my view that consciousness supervenes naturally but not logically on physics. His first main argument is that the view requires an extraordinarily large number of new psychophysical laws, at least one for every basic sensory quality, and possibly more. I think this is thinking about the laws the wrong way. Rather than a swarm of separate laws, one could (for example) imagine a relatively simple transformation operator from physical space to phenomenal space. These spaces may be high-dimensional, and the collection of the projections onto individual dimensions (which seems to be what Latham is looking at) may be quite complex, but this is consistent with the overall mapping being very simple; witness simple mappings between infinite-dimensional spaces in mathematics. It may also be that the dimensionality of phenomenal space can be reduced considerably by constructing it from a few protophenomenal properties. (Here Latham responds that we don't know what these properties are; but of course this is not to say they don't exist.)
Latham also considers cases such as pleasure/pain inversions (keeping physical facts constant), and related cases where experience become "inappropriately" matched with functional role. Latham argues that if these case were logically possible (as they are on my view), then we could not rule out the epistemic possibility that they are actual and that we are ourselves inverted in this way. After all, what could our certainty that we are not so inverted be grounded in? It could not be grounded in our acquaintance with the experiences, as this acquaintance is also present in the inversion cases, and it couldn't be in our beliefs, for the same reason.
In response: it's not obvious to me that the notion of "appropriateness" is entirely clear and absolute (as opposed to a product of familiar association), and it's also not obvious that we are certain we are in an "appropriate matching" state. (I'm certain when I am in pleasure, but I am not certain about functional roles in my brain and behavior.) But even conceding those points, I don't see that the logical possibility of inversion rules out my knowledge that I am not inverted. It is plausible, for example, that the justification of our beliefs about experiences lies not just in the experiences, and not just in the beliefs, but in the relations between the beliefs and the experiences. Those relations are not present in the inversion cases, and so the knowledge may be present in our cases even though it is not present in those. It's an interesting issue to investigate further, though.
Latham also adapts my own "dancing qualia" arguments to suggest that these are not just naturally but logically impossible. He thinks we can be certain our experience hasn't changed radically in the very recent past (whereas we couldn't be certain if dancing qualia were logically possible); I think we can't be certain of this any more than we can be certain that the world wasn't created in the very recent past. Latham suggests that although he can't rule out zombie worlds by this sort of strategy, the strong logical constraints (of functioning on phenomenology) he has established give indirect evidence for logical determination. I think this is a nice strategy, but that all the arguments for logical constraints go wrong.
*Joseph Levine, Review of The Conscious Mind. Mind 107:877-881, 1998.
Although a review, I mention this here because it is substantial and because it makes a point related to Melnyk's point below. Levine also challenges my arguments on the grounds that primary intensions need not be a priori accessible. He has two arguments for this: (1) infants and animals may have concepts without having any sort of access to their primary intensions; (2) certain contemporary theories of content (e.g. nomic covariation theories) suggest that concept's application-conditions may be not be cognitively accessible in this way, because they have "non-ascriptive" modes of presentation.
In response: (1) the infant/animal case isn't a counterexample to the a priori accessibility claim, any more than the fact that animals can't do mathematics shows that mathematics isn't a priori. Cases turning on cognitive limitations are never counterexamples to a claim of a priori knowability. (2) I think this is a better argument against those theories of content than it is against the accessibility claim. Levine suggests that the application-conditions of "water" and "chair" may be inaccessible in this way, but he does not address the obvious point (made in the book) that given qualitative knowledge of the external world, we are in a position to know what "water" and "chair" refer to. (Note: Levine addresses this point in the book chapter discussed below.) I take it that this is a datum that any theory must either take into account or else provide substantial arguments against; I have seen no such arguments in the literature on these theories. So it seems to me that the accessibility claim is not threatened. At best, Levine needs to take the "consciousness" concept as a special case.
I also think that even if Levine is right about inaccessibility, this is not enough to threaten the anti-materialist argument. This sort of inaccessibility would not deliver "strong necessities" (which are what the materialist needs), but at best "inscrutable truths". See "Does Conceivability Entail Possibility" for the distinction.
*Joseph Levine, "Lately things don't seem the same": The conceivability argument. Chapter 2 of Purple Haze. Oxford University Press, 2001.
To the key line of argument that says that "water" claims seem to be derivable from non-"water"-involving claims (e.g. PQTI) without further empirical information, Levine suggests that the relevant claims are "armchair" derivable, but not a priori derivable. I think this is an interesting line but one that ultimately fails: see "Conceptual Analysis and Reductive Explanation" (especially sections 3, 4, and 5(6)) for reasons. Levine briefly makes the interesting suggestion that the theory of reference and confirmation theory may play a role in these judgments, and that these may be a posteriori, so that the judgments are a posteriori. I'm not sure whether either is plausibly a posteriori, and I'm not sure that the theory of reference is needed for these first-order judgments, but set that aside. Insofar as either is a posteriori, they'll be grounded in certain empirical evidence, and the presence of this evidence will itself presumably be armchair derivable from PQTI. One can then use the style of argument in the section 5(6) on "A priori knowledge vs. armchair knowledge" in CA&RE to argue that there must be a priori conditionals nearby.
It should also be noted that even is Levine is right about the a priori coherence of water-zombies, this establishes only their negative conceivability, not their positive conceivability. On the face of it, ordinary zombies are (at least prima facie) positively conceivable, whereas water zombies are not (even prima facie) positively conceivable. So while this line of reasoning might address an argument based on the negative conceivability of zombies, arguments based on the positive conceivability of zombies remain standing.
Harry Lewis, Consciousness: Inexplicable - and useless too? Journal of Consciousness Studies 5:59-66, 1998.
Lewis makes a few criticisms of the view set out in my "hard problem" paper, arguing that the problem is still harder than I say, mostly because I don't take adequate account of the ineffability of consciousness. He doesn't like my listing "awareness" and "reportability" among the easy problems, for reasons bound up with this; I clarify my usage here in my JCS response (#3.1). There are a few places where he seems to misplace my view, but I've clarified those things elsewhere. Lewis's central point is that the ineffability of experience precludes a theory bridging processing to experience, but this is far from clear to me, and he gives little explicit argument for this conclusion. It may be that we can develop some sort of phenomenological formalism (geometric, informational, mathematical, linguistic?) within which experiences can be described and taxonomized. We already know we can do this for some aspects of experience (the more obvious structural aspects), and I think it is an open question how far we can take it. Certainly this is an area that needs to be investigated carefully in coming years.
*Penelope Mackie, Deep contingency and necessary a posteriori truth. Analysis 62:225-36, 2002.
I don't hold the hidden proposition view, although maybe Jackson does. I think Jackson should probably reply by saying that in these cases, knowing what the 2-proposition is requires more than just knowing what the referent is, but also knowing the referent's essential properties, so that one is able to evaluate the sentence's truth at arbitrary (neutrally described) counterfactual worlds. Speaking for myself, I think that claims about "knowing" the identity of intensions or propositions are irrelevant here. Rather, I'd say simply that when a sentence has a contingent 1-intension but necessary 2-intension, it follows automatically that the sentence's truth is not knowable a priori (since there will be epistemic counterpossibilities), and nor is the sentence's necessity.
*Diego Marconi, Two-dimensional semantics and the articulation problem.
Marconi worries that "water is necessarily H2O" is true when read subjunctively but not when read epistemically. At most, however, this involves an ambiguity in "necessarily" (between subjunctive and epistemic necessity). Once this is disambiguated, we obtain the desired result, thus accommodating both Fregean and Kripkean intuitions. There is no further worry about the semantic value of "water is H2O", which functions as above.
Marconi also suggests that "water could have been XYZ" (and the like) comes out trivially false on the 2-D view, when it should not. I think that if one has the contra-Kripkean intuition that the sentence (read subjunctively) is true, then one should simply hold that the 2-intension of "water" picks out XYZ at some worlds. If one thinks that it is false but not trivially so, then this is predicted by the standard view of the 2D-intension of "water": there are epistemically possible scenarios in which the 1-intension of "water could have been XYZ" is true, so the sentence in question is false a posteriori, and so not trivially false.
In an appendix, Marconi suggests that two definitions I give of two-dimensional intensions come apart in the case of nonrigid expressions. But what Marconi calls my "definition 2" is not a definition at all, but rather an illustration in the case of a rigid expression ("Hesperus"). So there is no conflict here.
*Andrew Melnyk, Physicalism unfalsified: Chalmers' inconclusive conceivability argument. In (C. Gillett & B. Loewer, eds) Physicalism and Its Discontents. Cambridge University Press, 2001.
I imagine that when things are done this way, Melnyk will want to deny that 1-conceivability entails 1-possibility. I'm not sure just how this would go, though. Perhaps Melnyk will deny the "scrutability" of truth and reference, holding that for many expressions E, there is no a priori conditional from a description of the world in some fixed limited vocabulary (e.g. PQTI) to truths involving E. If this principle fails widely, then the conceivability-possibility claim may fail too. (In fact this would defeat only an entailment from negative 1-conceivability to possibility, and not an entailment from positive 1-conceivability: see the section on "Inscrutabilities" in "Does Conceivability Entail Possibility".) Reasons for accepting this sort of scrutability are given in "Conceptual Analysis and Reductive Explanation", especially sections 3 and 4.
David Papineau, Phenomenal and perceptual concepts. In (T. Alter & S. Walter, eds) Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism. Oxford University Press, 2006.
John Perry. Knowledge, Possibility, and Consciousness. MIT Press, 2001.
John Perry. Response to commentators. Philosophy and Phenomenological Research, January 2004.
*Mark Rowlands, Consciousness and supervenience. Chapter 2 of The Nature of Consciousness. Cambridge University Press, 2002.
*Stephen Schiffer, Mental content and epistemic two-dimensional semantics. Pacific APA, March 2002.
Paul Skokowski, "I, zombie". Consciousness and Cognition 11:1-9, 2002.
Tamler Sommers, Of zombies, color scientists, and floating iron bars. PSYCHE 8(22), 2002.
Robert Stalnaker. What is it like to be a zombie? In (T. Gendler & J. Hawthorne, eds) Conceivability and Possibility. Oxford University Press, 2002.
*Robert Stalnaker. On considering a possible world as actual. Proceedings of the Aristotelian Society, Supplementary Volume, 2001.
This discusses two-dimensional semantics as a way to analyze Kripkean a posteriori necessity. Stalnaker defends his own "metasemantic" account and argues against the sort of "semantic" account given by me and others. I think the metasemantic account is unsatisfactory, as it can't capture the distinction between a priori and a posteriori necessities. His objections to the semantic account are brief and inconclusive, but he appears to commit his opponent to a version on which primary intensions are evaluated by finding the extension of a token with the same meaning as the original token, placed in the new context. I think that's the wrong way to do things: see the discussion of "epistemic" and "contextual" interpretations of 2-D semantics in "The Foundations of Two-Dimensional Semantics". See also the discussion of Stalnaker's approach in section 5.1 of that paper.
*Scott Sturgeon, Zombies and ghosts. Chapter 5 of Matters of Mind. Routledge, 2000.
Sturgeon's central argument is that metaphysical modality is "mind-independent", and that we shouldn't expect a priori methods to provide guidance to mind-independent reality. First response: mathematics is plausibly mind-independent, and a priori methods plausibly provide guidance to mathematics. Second response: "mind-independent" is ambiguous. If it means independent of what we happen to think (the most common usage), modality is plausibly mind-independent. But if it means independent of everything in the domain of the mental, including e.g. norms of rational inference and the like (Sturgeon's usage suggests this; he says "even in the rational ideal"), then it's not obvious that metaphysical modality is mind-independent. I've argued that there are constitutive ties between modality and rationality, in which case modality is not mind-independent in this sense. I think it's still perfectly objective and real, however. Compare: principles of logic are not mind-independent in this sense, but they are perfectly objective and real.
Sturgeon also argues against the conceivability-possibility inference as follows: it's conceivable that zombies exist, which seems to support the conclusion that zombies are possible. But it's equally conceivable that zombies are impossible, which equally supports the conclusion that it's possible that zombies are impossible, which plausibly entails that zombies are impossible. These conclusions are contradictory, so the conceivability-possibility inferences are defeated. My response is to deny that it's (ideally) conceivable that zombies are impossible. The issues here are directly analogous to those in Yablo's case of the conceivability of a necessary god in "Concepts and Consciousness", which I reply to in section 3.3 of "Materialism and the Metaphysics of Modality".
Nigel Thomas. Zombie killer. In (S. Hameroff, A. Kaszniak, & A. Scott, eds) Toward a Science of Consciousness II. MIT Press, 1998.
Thomas argues that the concept of a zombie is incoherent, by appealing to the fact that zombies will claim to be "conscious". He considers the possibility that we might interpret this claim as true, false, or meaningless, and argues that trouble threatens each way.
I discuss this sort of thing at length in Chapter 5 of my book, where I argue that the zombie's claim is best taken to be false (though I have occasional sympathy with the "meaningless" option), e.g. because the zombie's "consciousness" concept is empty, a bit like "phlogiston". So the zombie is mistaken. In response, Thomas notes that the zombie's claim is formed by the same "cognitive process" as an ordinary conscious being's claim, so if the zombie is mistaken, this cognitive process is unreliable, throwing strong doubt on our own claims that we are conscious. This line of reasoning was anticipated in the book, where I suggest that the justification for my belief that I am conscious lies not just in my cognitive mechanisms but also in my direct evidence; the zombie lacks that evidence, so his mistake does not threaten the grounds for our beliefs. One can also note that the zombie doesn't have the same beliefs as us, because of the role that experience plays in constituting the contents of those beliefs. (There's more on this in "The Content and Epistemology of Phenomenal Belief", section 4.3.) Thomas's discussion, although it raises some intriguing issues, does not address this natural line of defense.
Robert van Gulick, Conceiving beyond our means: The limits of thought-experiments. In Toward a Science of Consciousness III. MIT Press, 2000.
Third (and related), van Gulick suggests that the conceivability of zombies may be grounded in the "inadequacy" of our phenomenal and/or physical concepts. The notion of a concept being "inadequate" is obscure to me, and I think the conceivability-possibility principles I use apply to all concepts, whether "adequate" or "inadequate". The only counterexample offered by van Gulick is the vitalist example: where L = "life exists", P&~L is conceivable (with a vitalist's concept of L) but not possible. But I think it's clear that with the usual concept even around that time, P&~L was not (ideally) conceivable. At best P'&~L was conceivable, where P' expresses their false beliefs about the physical world, but that's something different. Perhaps one could argue that some vitalists used "life" so that vital spirit was a conceptual requirement on life, so that P&~L as used was conceivable. But then P&~L was also possible, and L as used was false. So there are no counterexamples here.
I think that rather than talking of conceptual inadequacy in these cases, one should talk about insufficient reasoning, or about false beliefs. Then it becomes clear that the room opened up by these cases corresponds at best to type-C materialism (we've reasoned insufficiently about zombies, or we don't know enough about physics), or to eliminativism (we have a false belief that consciousness exists). I think the former view is unstable on examination, and that the latter view is extremely implausible. See "Consciousness and its Place in Nature" (especially section 7) for discussion.
Bram van Heuveln, Eric Dietrich, & Michiharu Oshima, Let's dance! The equivocation in Chalmers' dancing qualia argument. Minds and Machines 8:237-49, 1998.
This paper tries to escape the dancing qualia argument by on the grounds that the argument assumes that one system is present throughout, but it may be the case that on switching subsystems two different systems come into and out of existence. So really two distinct 'phenomenal worlds" are present, and it's not surprising that no change is noticed.
What really matters here is whether two distinct individuals (i.e. subjects) are involved. If two distinct individuals are involved, the argument won't go through. I tried to handle this point in the book by making the changed area small enough (less than 10 percent of the brain) that it was plausible that the same individual was present throughout. It seems unlikely `that replacing some small part of the visual cortex (say) will be enough to change my identity. If one thinks 10 percent is too much, we can even make the replacement 5 percent or less, without too much cost. The authors don't really address this point. I think they hold that the mere presence of two physically distunct subsystems is enough to make their point, but I don't see this. If we grant that all this involves changes in my consciousness, as seems plausible, I think the argument goes through.
The purported "equivocation" is one between 'change in experience" and "experience a change". They argue that although the experiences changes, no single individual experiences a change. I do require that a single individual experiences a change, at least in the sense that the experiences that change are those of an individual, although not in the sense that the individual notices the change, which obviously can't happen in this set-up. The point of the small replacement is precisely to make it plausible that this happens.
Tillmann Vierkant, Zombie Mary and the blue banana. PSYCHE 8(19), 2002.
*Sara Worley, Conceivability, possibility, and physicalism. Analysis 63:15-23, 2003.
*Stephen Yablo, Textbook Kripkeanism and the open texture of concepts. Pacific Philosophical Quarterly 81:98-122.
*Stephen Yablo, Coulda, woulda, shoulda. In (T. Gendler & J. Hawthorne, eds) Conceivability and Possibility. Oxford University Press, 2002.