What the Evans literature is about

    What this literature (Evans, Davies & Humberstone, Jackson, Canberra Plan, Chalmers) is about is, the synthetic identity of properties, and a posteriori necessities.
     In the late 1950's, U.T. Place, Jack Smart and Hilary Putnam liberated us from analytic philosophy, by pointing out that properties can be identical even though the predicates that express them are not synonymous.  Being water and being H2O; clouds and masses of water droplets; lightning and electrical discharge in the sky; genes and segments of DNA molecules.
     This syntheticity of the identities meant that Ordinary-Language and other analytic philosophers could not object, against e.g. the Identity Theory of mind, that "pain" doesn't mean c-fiber firings, that we can imagine or conceive pain without c-fiber firings, etc.  (Of course it doesn't, of course we can, so what?  Get lost.)  It also meant the immediate death of the Open Question Argument in meta-ethics (which demise the "Cornell Realists" discovered after I'd been noting it for twenty years).
     Smart characterized the synthetic property identities as "contingent" identities.  But Kripke then reminded us (following Ruth Marcus) that there's no such thing as a "contingent identity"; what the synthetic identities are is, rather, a posteriori.  They're necessary, like all identities, but known only empirically rather than a priori.
     So matters stood for many years.  A posteriori necessity has been the paradigm for reductive theories in philosophy: mental properties to brain properties, semantic properties to causal etc. properties, moral properties to natural properties, and for that matter middle-sized physical objects to swarms of subatomic particles.
     But since Black's famous objection to Smart (fn 13 of "Sensations and Brain Processes"), there has been a particular undercurrent of unease.  Black and others argued that a posteriori identities don't come for free; certain conditions have to be met.  And a type of argument against the Identity Theory (Black, Kripke, W.D. Hart, Stephen White, David Lewis, George Bealer, Jackson 1993, Chalmers) has taken the form, "Although water/H2O et al. do meet those conditions, mental states / brain states doesn't meet them; so mental states aren't identical with brain states."
     In the Evans literature, a new but closely related form of resistance to the paradigm has emerged.  The general idea is that although of course there are a posteriori identities, there is still underlying each one an a priori identity that is philosophically more important than the original identities themselves, and as philosophers we should be more concerned with those underlying a priori identities than with the originals.  This is where and why Jackson thinks conceptual analysis is so important.
     And that idea leads to a resurgence of the foregoing type of objection:  In the special case of mental states and brain states, the relevant a priori identities fail; therefore mental states are not identical with brain states.
     When I first read this stuff, my appalled reaction was, "Good God, next they'll be bringing back the Open Question Argument!"  I meant that rhetorically.  But, har har: Horgan and Timmons (1992).  See Jackson's Ch. 6 for a discussion of that issue.  Jackson ultimately does not buy the Open Question Argument, but he makes a ghastly concession to it.
     Anyway:  In reading Jackson, bear in mind that it's all about a posteriori necessary identities and our not getting them for free.
    What is FJ going to mean by "concept"? He addresses this, but not very definitely, on pp. 33-34.  Under the heading of "analysis," he means the analysis of concepts as opposed to words, but the concepts seem to be abstracta rather than mental entities, and his triangle example suggests that they at least line up with word meanings.
 

Reply to Jackson on defining/changing the subject

    In his section, "Folk Theory and the Causal-Historical Theory of Reference: A False Opposition" (pp. 37-41), FJ contests my Putnamian view of "belief" and beliefs.  Here are some remarks in reply.

     His first and important substantive disagreement is that "if we give up too many of the properties common sense associates with belief as represented by the folk theory of belief, we do indeed change the subject, and are no longer talking about belief" (p. 38).
     This needs disambiguating as regards "folk theory."  (I do disagree with him on each of the two readings, but for quite different reasons.)  (1) If "folk theory" means what it does to Lewis, viz., the set of commonsensical but mostly (obviously) contingent platitudes about belief, then I disagree because as FJ says, I think we can give up lots and lots of those platitudes and still be talking about belief.  (2) If "folk theory" means what most English speakers would agree were the noncontingent, analytic implications of "belief" as revealed by intuitive semantic judgments about hypothetical cases, then of course if I give up any of the implied properties, I am no longer talking about belief.<1>  My disagreement here is that I don't think "belief" (or any other term) has any analytic implications.
     It sounds as though FJ here intends sense (1), since he has characterized the disagreement in the way he has.

     I don't know what FJ would say about the parallel issue concerning proper names.  I find it tolerably obvious that virtually all our beliefs about person X expressed using a proper name of X could be false, yet X would still be the person those beliefs were about.  (David Kaplan once gave the real-world example of "Robin Hood"; according to some researchers circa 1970, Robin Hood was a real person, that they had identified, but he had none of the properties stereotypically associated with him, including being called "Robin Hood" (well, he was male).)  Would FJ disagree, or would he claim that natural-kind terms work differently in this regard from proper names?

     FJ argues, "[S]urely it is possible to change the subject, and how else could one do it other than by abandoning what is most central to defining one's subject?"  What exactly does FJ mean by "change the subject"?  I guess, to be using the same word but no longer be talking about the same phenomenon one was before; or, a more familiar situation, to be using the same word as another person but using it to refer to a different phenomenon, i.e., to be talking past each other.   Here's how else the latter could happen, according to the Causal-Historical theory: for the respective causal-historical chains to be, unbeknownst to the speakers, grounded in different phenomena.  (The famous story of "Kraut," Kraut, Kraut, and the 1982 Chapel Hill Colloquium.)  This is a little harder to envisage for the intrapersonal case; the speaker's later uses of the word would have not to be primarily grounded in her/his earlier uses.  But it seems perfectly possible, and also, intrapersonal "changing the subject" is itself harder to envisage in the first place.
     (I admit that there is a genuine issue about changing the subject, even within the Causal-Historical camp.  It has been argued that the C-H theory implies that phlogiston really exists:  Phlogiston is oxygen; it's just that the original theorists had lots of false beliefs about it, such as that in combustion it is given off rather than taken on, etc.  Well, yeah.  I confess that I have always wondered why people haven't taken that view; I secretly tend to believe it myself.  But many would regard it as a reductio.  By far the best discussion I know of these issues is in sec. 11.2 of Steve Stich's Deconstructing the Mind.  Stich argues pretty convincingly that in historical cases, the "reduce or eliminate" choice has been largely verbal and governed by contingent sociological and/or political factors.  If true, that counts against FJ's conservative descriptivist view of the matter, though it doesn't directly support my libertine view.)

     In fn 12, FJ says he thinks that prior to the Putnamian revolution it was at best indeterminate whether "water" meant H2O or just watery stuff in general.  Off the record, I agree with him.  I'm not even entirely sure that in ordinary (as opposed to philosophers') English, "water" doesn't mean watery stuff rather than H2O.  Certainly it's hard to get novices to buy Putnam's Twin-Earth argument.  At best, usually, they end up being persuaded that there is a sense in which XYZ isn't water.  But like FJ, I'll publicly and officially go along with the Putnamian orthodoxy.

     FJ says, "Our agreement...reflected our folk theory of water.  Putnam's theory is built precisely on folk intuitions."  This again uses "folk theory" ambiguously, but here FJ seems to mean sense (2) rather than sense (1).
     Notice another important distinction, which FJ has almost ostentatiously ignored:  The theory is built on intuitions, we agree, but what sorts of intuitions?  Modal intuitions--intuitions regarding what is or isn't metaphysically possible.  (E.g., it isn't metaphysically possible for something to be water and not to be H2O; it is metaphysically possible for water to lack all the stereotypical features associated with water.)  FJ talks as though the intuitions in question were semantic intuitions, intuitions about meaning, (semantic) entailment, etc.  (Cf. the potential running together of "entailment" in the neologistic sense of metaphysical determination with real (semantic, conceptual) entailment.)  We must be vigilant here:  FJ is not entitled to the claim, should he go on to make or presuppose it, that the "intuitions" appealed to in philosophical argument or (alleged) conceptual analysis are semantic intuitions.  For example, on p. 28 he argues fallaciously that Chisholm's and Ayer's JTB theory of knowing "counted as a piece of conceptual analysis," "because it was intended to survive the method of possible cases."  Non sequitur.  (Historically, Chisholm and Ayer themselves probably did regard the theory as a piece of conceptual analysis, but they need not have.)

     Perhaps FJ is right that something might be a believer without sharing an informational natural kind with us, depending on what kind of "information" is at issue.  I mean to leave it quite loose what natural kind "believer" is.

     On p. 41, FJ argues that, supposing I was right about the informational natural kind, what would show that is, our having the corresponding [modal!] intuitions.  Right.  But now, in the argument he then makes against me, a major confusion occurs: essentially the "folk theory" equivocation again.

But then what is being revealed by these responses is precisely that the property intuition associates with belief is belonging to the right informational natural kind.  So it cannot be right to say, as Lycan does, that a state might be a belief without having the properties we usually associate with belief.  If [my italics] intuition delivers the answers it needs to for Lycan's claim to be plausible, the property we folk associate with belief is belonging to the right informational natural kind, and that property is precisely the one that Lycan thinks that all believers have.

Modest and immodest roles for conceptual analysis

    Here's why I don't understand what FJ is getting at on pp. 42-44.  Geach is presented as a "modest" analyst, because he is only explicating the folk concept of "change," and not drawing any metaphysical or other factual conclusion from the explication.  But others have used Geach's analysis to defend a metaphysics of temporal becoming, by mounting the argument attacking four-dimensionalism; so their use of the analysis is immodest.   Fair, and clear, enough.  But why is that difference between Geach's and the others' intellectual biographies interesting or important?  The second Geach's explication is offered, it's available for metaphysical use and certainly will be so used by someone, and if the explication is a correct piece of conceptual analysis, rightly so.  FJ says he is, and we should be, "suspicious of conceptual analysis in its immodest role," but why?  If Geach's analysis is correct, four-dimensionalism is indeed false, right there, and what occasion is there for suspicion?<2>
   I wonder slightly whether FJ is hinting at a difference between the folk concept of an F and some deeper or more "real" concept of an F.  (I don't think he is, but some philosophers have suggested such a distinction.)   The anti-four-dimensionalist argument tacitly assumes that the folk sense of "change" is what's in question throughout the argument:  If different things' having different properties is not change in that sense and things do in fact change in that sense, then (since four-dimensionalism holds that the only change there is is, different things' having different properties) four-dimensionalism is false.   But what if the folk sense is defective in some way, and does not get at the "real" concept of change?  I can imagine someone responding, "So much the worse for the folk sense; I don't care whether different things' having different properties is not change in that ignorant plebeian sense.  Different things' having different properties is change in the only sense that is metaphysically sound.  So four-dimensionalism remains unrefuted."
    I myself would reject that response.  If the word "change" really does mean something over and above different things' having different properties, but four-dimensionalism holds that that is the only change there is, and things really do change, then four-dimensionalism is false.  FJ notes that "the four-dimensionalist concept...may not be the same as the folk concept," which suggests some sympathy with the foregoing response, but at best the four-dimensionalist has made the sort of Quinean move described by FJ on pp. 44-45.  The four-dimensionalist would have to put her/his view by denying premise 2 (the Moorean fact) and then issuing a qualification: "Strictly speaking, entirely contrary to common sense, things never change.  But they do do something enough like change that if we talk about it, our talk will still do the theoretical job we folk give the concept of 'change'...."  Of course, it would be easier for the four-dimensionalist just to reject Geach's analysis in the first place.
 

Whence A-intensions?

    Let me say a little more about the grounding of (the alleged) A-intensions in 2-D modal logic, and about their dubiousness.
     Here's why I have never understood FJ's and others' characterization in terms of evaluating a term at a world w "under the supposition that w is the actual world" as opposed to w's being a merely counterfactual world:  First off, if w is not the actual world, then it follows that w is a counterfactual world.  FJ must intend some distinction between "merely" counterfactual worlds and in some way non-"merely" counterfactual worlds, but I have no idea what distinction it might be.  (I think Dorit's Kaplanian idea of "world-indifference" is on the right track--see below--but it's not the same idea as FJ's and is not defined in terms of "actual.")
     Second, think of a world as a maximally consistent set of propositions (intuitively, all and only the propositions that are true at that world).   What does it mean to "suppose" that a world is actual?  Obviously, to suppose that the collectively false set of propositions were true.  OK, I'm supposing that an XYZ-world is actual, so I'm supposing that, among other things, it's true that the lakes are filled with XYZ and XYZ comes out of the taps.  What does "water" refer to at that world?  To H2O, if anything, just as always, and there isn't any H2O there, so "water" is empty at that world, contra FJ's intentions.  So I don't know what he means.
     (Maybe it's tendentious to think of a world as a set of propositions.  David Lewis would insist that a world is a gigantic concrete entity, just like the real world only disconnected from it in spacetime.  But a Lewisian concretist has a special difficulty making sense of counterfactual suppositions of the form "If w were actual," or so I argued on pp. 34-35 of Modality and Meaning.)
     So I don't understand FJ's characterization.
     What's gone wrong in the foregoing reasoning is, as Katya said, that I've been insisting on using the word "water" with its actual meaning, what it means in our language, English, in the actual world.  But the Ur-idea of 2-D modal logic was to evaluate a term at a world w, not according to what it means in our language, but according to what the term would mean at w.  (Hence Gupta's appeal to the old leg/tail joke.  The 2-D logician says, although, yes, a horse still has only four legs even if we call a tail a "leg" (since calling a tail a "leg" doesn't make it a leg), there's also a sense in which if we call a tail a "leg" the horse has five legs.  That's why the joke question was perceived as a riddle or puzzle in the first place.)
     So the idea is that to find "water's" A-extension at a world w, find out what "water" would refer to if it were being used by a denizen of w, meaning what it would mean there in that person's mouth as opposed to what it means in our speech.
     Of course, there is an immediate puzzle, pointed out by Keith in his seminar a year ago:  "What it would mean there"??  Depending on the world, it (the sign-design "water") might mean anything--arsenic, or lions or tigers or bears (oh my), or it might be a conjunction and mean "and," or it might be a proper name, or anything at all, or nothing.  What the 2-D proponent must mean is, what it would mean there if various other stuff in the other world is held fixed relative to our actual world.  But what stuff?
     No one has said.  Dorit's indexical example suggests, and FJ et al. intend, that what's held fixed constitutes a kind of meaning, a kind other than ordinary referential meaning.  A world-indifferent kind, as Dorit said; "water" would have the same "meaning" of that new kind at Twin World as it does for us.  But that does nothing to tell us what is supposed to be held fixed or why; so until we are told, we have to reason to believe in the alleged new kind of meaning.
     I believe the idea is this, though I think it lacks motivation:  Let's stipulate and hold fixed that at Twin World, "water" is a natural-kind term and works semantically much as it does here.  That is, it refers to an underlying chemical substance that is the scientific essence of the familiar liquid in question, and its reference is fixed (in Kripke's sense) by superficial descriptions such as "the stuff that fills the lakes" and "the stuff that comes out of the taps."  Everything else about English is held fixed also.  Under those suppositions, what does "water" mean at Twin World?  Well, everything is the same except that the underlying substance "water" refers to there is XYZ.  So at Twin World, "water" means XYZ, not H2O, in the same way that in our language "water" does mean H2O.
     Now we can generate FJ's A-intension for "water."  At any relevant world, i.e., any world at which the sign-design "water" is used as we use it except for there being a different underlying substance, it will mean, not H2O, but the underlying substance at that world.  It will refer to "whatever plays the watery role" there.  So (this is the bold 2-D inference), even for us in the actual world, "water" has a kind of flaccid intension or meaning along with its normal, rigid referential meaning.  That intension is the function that sucks up a world and spits out whatever plays the watery role at that world.  Thus, "water"'s flaccid meaning for us is, stuff that plays the watery role.
     And that meaning is world-indifferent within the class of worlds circumscribed above.  It is also a candidate for "narrow content" in the sense that's supposed to be common to Earthling and Twin Earthling users of "water," though that case would have to be made with some care.  (N.b., narrowness of attitude content is a different though related issue.)

     Here's why I don't think English words have A-intensions.  To generate an A-intension, one needs a transworld "role," as in "plays the watery role."  But by what are these roles supposed to be constituted?  By reference-fixing descriptions, that are the same across the relevant worlds.  (Notice the retro, antiKripkean move here:  Kripke taught us that we need not believe that referring expressions abbreviate descriptions just because those expressions are "backed by" descriptions; the descriptions merely fix reference, and do not fix sense.  But here FJ is saying, OK, so they don't fix sense in the usual sense of "sense," but there's another sense of "sense" in which they damn well do fix sense.)  All this presupposes that an English word has a distinctive and stable set of reference-fixers.  And that presupposition I deny.
     The presupposition is not silly.  Dictionary entries contain common stereotypical information about water, and philosophers have had no trouble coming up with sets of stereotypical reference-fixers such as those mentioned above.  There is a loose body of information about water that we 21st-century Americans share.  That seems to constitute a "watery role."
     But there being that body of information is a highly contingent fact.  First, no one such description is essential.  One could still have the word "water" explained to one even if there were no lakes and even if there were no taps, by reference to some other mode of acquaintance with water.  In fact, even the collective body of all the reference-fixers that have been mentioned in the Putnam literature is inessential; under unusual circumstances, one could have the word "water" explained to one even if there were no lakes and no taps and no rain and no drinking and no colorless liquid and and and..., so long as there were (real or imaginary) water around doing something or being some way.  And "water" would still mean just what it does now, in real-world English.
     (In saying that, I am not begging the question by insisting that "water" means H2O, period.  I mean only that "water" would mean what it does now, whatever that is, even if its reference were fixed, in some context, by descriptions entirely different from the usual ones.)
     Perhaps surprisingly, it is not a linguistic fact that "water" satisfies the descriptions that you find in dictionaries.  Dictionaries contain lots of nonlinguistic information.  The difference between what information turns up in dictionaries and what information doesn't is only the difference between information that is, though entirely contingent, widely known among a particular dictionary's intended readership, and information that isn't so widely known.
     No distinctive, stable set of reference-fixers, no stereotypical "role."  No stereotypical role, no Jacksonian A-intension.  (And that distinguishes A-intensions from Kaplanian characters.  Everyone agrees that indexicals have Kaplanian characters.)  So we've been given little reason to think that English words have A-intensions at all.
 

A concession

    On p. 53 FJ makes an argument against Quine that I believe is potentially a good argument, though I would explain it a bit differently.  He says that every representation partitions the space of posssibilties.  Now, consider a case in which representations R1 and R2 delimit sets of possibilities, one of which is a subset of the other.  If R1's set is a subset of R2's, then the proposition, "If R1 then R2" is true independently of world, and hence a priori in FJ's sense.
    It's actually worse than that from the Quinean point of view.  One can show that, assuming there is determinate meaning at all, there is a perfectly good sense in which some sentences are true in virtue of meaning.  Suppose that the meaning of a sentence is a set of worlds,  a correct theory of meaning for the containing language.  And suppose again that R1's set is a subset of R2's.  Then as before, "If R1 then R2" is true at every world (hence necessary), and/but, more to the point, this follows from the correct theory of meaning aforementioned.  Is "If R1 then R2" not then true in virtue of its meaning?  And so, isn't it, well, analytic?
    I think we may concede a sense of "analytic" in which some sentences are analytic.  In Modality and Meaning (see section 3 of Ch. 12, Chs. 11 and 12 being posted here) I offered one line of resistance to that concession, and I think my line is still sound so far as it goes, but it's OK with me if people want to use "analytic" in this way; I called the new sense "TD-analyticity," "TD" for "truth definition."  But, n.b., it is not a sense in which sentences are true contentlessly and with no contribution from the nonlinguistic world; nor is it any help either to the metaphysician in search of immutable necessary truths or to the anti-metaphysician bent on debunking such things; nor does it (always) afford a priori knowledge.
    That last disclaimer needs explaining, in light of FJ's contention that since  "If R1 then R2" is true independently of world, it is a priori in FJ's sense.  By "a priori," FJ means "[w]hat we can know independently of knowing what the actual world is like" (p. 51).  Assuming that means, independently of knowing what the a.w. is like except for the word meanings in it, I guess there can be a priori knowledge in that sense.  (But notice the odd result that, in that sense, I can know the word meanings a priori; e.g., I can know "a priori" that "cow" means cows.)  But a priori knowledge even in that sense is not guaranteed by TD-analyticity.  For I may not have anything like a good grasp of my sentence's actual truth-condition.  I may not know a priori, or at all, that if R1 then R2; I may never come to know it.  Thus, TD-analyicity is of no help in settling philosophical disputes; because in a hot and complex philosophical discussion, the semantics of the relevant expressions is likely to be just as much up for grabs as are the actual propositions being disputed, even though (I very reluctantly concede to FJ) those two things are different and separable.
 

The central(?!) argument of Chapter 3

    FJ is concerned to defend, as he did in "Armchair Metaphysics" and other pilot papers, the Strong entry by entailment thesis, to be very carefully distinguished from the original or "Weak" entry by entailment thesis.  (He does the distinguishing himself most clearly on p. 68, where he raises the "question" (italics his) of which of the two he means.)  The Weak thesis says only that "the one and only way of having a place in an account told in some set of preferred terms is by being entailed by that account" (p. 5); but "entailed" there means only metaphysically determined, so the Weak thesis is unexciting--some would call it obvious.   By contrast, the Strong thesis says that the one and only way of having a place in an account [etc.] is by being entailed by that account, where "entailed" means entailed, i.e., semantically/conceptually entailed.  The Strong thesis is insane, or, to use less contentious terms, very startling and highly controversial. On the Kripke-Putnam view, the type-identities that figure in scientific reductions are a posteriori, and so they do not afford semantic entailments going upward from micro- propositions to macro- ones.  That is one reason why we should expect the Strong thesis to fail (there are at least two other important reasons<3>).
    It is the Strong entry by entailment thesis that figures in Jackson's and Chalmers' arguments against materialism in the philosophy of mind (see again my handout, "Against the New A Priorism in Metaphysics"), though FJ does not make that anti-materialist argument in his book.<4>  The Weak thesis would not support such arguments.
    Now, the Strong thesis was the central claim of "Armchair Metaphysics," and I expected it to be a central claim of the book and the central claim of Ch. 3. Yet it occurs very unemphatically, on pp. 59-60, and most of the rest of the chapter is spent on another (though clearly related) matter.   You have to look closely to see the argument.  FJ points out that there is a nonstandard way of interpreting the standard Armstrong-Lewis role-occupant type of argument, viz., according to the 2-D picture.  Instead of taking the first premise as true in virtue of sense-fixing descriptions, we can take it to be true in virtue of merely reference-fixing descriptions and "temperature"'s alleged A-intension; on that interpretation, the premise would be a priori and contingent, rather than a priori because a necessary conceptual truth.  That, FJ argues, will afford him at least potential derivations of macro- propositions from micro- ones, vindicating the claim of semantic entailment.
    As we saw in class, the type of derivation that he offers, based on the 2-D model, is very ingenious.  But, strangely, he does not spell it out at any length in the book, though he does give a quick example on p. 82.  See again the handout aforementioned; as there, I would argue that, ingenious as they are, Jackson-Chalmers derivations are fallacious.
 

Jackson's attack on the "metaphysical"/"conceptual" distinction

    Here again is the standard picture of logical space (drawn and colored for us by Adam, to whom thanks).
 

    OK, what's supposed (by FJ) to be wrong with this picture?  He is concerned to deny that there is a difference between "metaphysical" and "conceptual" possibility/necessity.  (The view he attributes to his opponent in the first half of p. 69 is a version of the standard picture, and one that I and many others firmly accept.)
    Let's get one distraction out of the way:  FJ denies that there are two senses of the terms "necessary" and "possible," and he continues to refer to his target view as "the two senses view."  But no one (I hope) thinks that the alethic terms are ambiguous, as between different senses.  Rather, as I put it earlier, there are types and grades of necessity and possibility, designated by adverbs such as "conceptually" and "legally."  FJ is surely aware of that, and his arguments do not turn on the difference between different senses and different types.  His allusions to "senses" are just sloppiness.  Mainly, he wants to deny that there is a difference between metaphysical and merely conceptual possibility/necessity.
    He offers two reasons for that denial.  The first is "Occamist" (pp. 70-74).  As its name implies, it is an appeal to parsimony.  "The phenomena of the necessary a posteriori, and of essential properties, can be explained in terms of one unitary notion of a set of possible worlds" (p. 70).  The main phenomenon that FJ is concerned to explain is disconcertingly specific:  "...how a sentence can be necessarily true and understood by someone, and yet the fact of its necessity be obscure to that person" (p. 71).  He gives a predictable and (if we waive the issue of the stable set of reference-fixers) not implausible 2-D-style explanation.<5>  He concludes, "And the important point for us is that this story about the necessary a posteriori does not require acknowledging two sorts of necessity.  The story was all in terms of the one set of possible worlds."
    The latter remark is perfectly true, but unresponsive.  First, that the "metaphysical"/"conceptual" distinction is not required to explain the specific phenomenon of how a sentence can be both necessary and a posteriori does not entail that it is not required for any explanatory purpose.  Second and more importantly, the distinction is not (or not primarily) an explanatory posit in the first place.  As before, conceptual possibility outruns metaphysical possibility because, just as a proposition may be (semantically) entailed by the laws of nature without being true in all metaphysically possible worlds, a proposition may be entailed by some fact or law of metaphysics without being a conceptual truth: "All water is H2O," and "Mark Twain is Samuel Clemens," for example.  If we think of possible worlds as sets of propositions, the (merely) conceptually possible worlds contain propositions that no metaphysically possible world contains.  Those are (broadly) logical facts, not hypotheses invoked to explain anything in particular.  All the distinctions between grades of necessity and possibility--nomic vs. legal vs. biological vs. moral etc. etc--remain in place just as before; why should metaphysical vs. conceptual be any different?

    FJ's second argument is brief:  "The key point is that the right way to describe a counterfactual world sometimes depends in part on how the actual world is, and not solely on how the counterfactual world is in itself.  The point is not one about the space of possible worlds in some newly recognized sense of 'possible', but instead one about the role of the actual world in determining the correct way to describe certain counterfactual possible worlds" (pp. 77-78).  I have my usual reservations about that 2-D way of putting it, but let's suspend those.  My problem with this second argument is that I don't see that it is, in fact, an argument.  Suppose FJ's "key point," expressed in the first sentence of the foregoing quotation, is correct.  How is it supposed to follow that there is no difference between metaphysical and conceptual possibility?
    Perhaps this is another appeal to parsimony; we can construe the arguments of Kripke and Putnam in the 2-D way, so there is no need to posit a special realm of conceptual-but-not-metaphysical possibilities.  But if that's the right interpretation, I would (obviously) make the same reply as I did to the first argument.  The realm isn't a special realm, and certainly not "some newly recognized" one; it's an already recognized sector of logical space like any other sector.
    If the argument is not another appeal to parsimony, then I don't know what it is.
    Here as previously, perhaps it is tendentious of me to think of possible worlds as sets of propositions.  Suppose Lewis is right, and other worlds are gigantic concrete entities like our own universe but merely disconnected from it.  Then there might be a debate over exactly which types of concrete world exist.  For example, Lewis himself refuses to countenance impossible worlds; in The Plurality of Worlds he gives an argument against them.  In this concretist context, FJ's appeal to parsimony would make more sense, because gigantic concrete entities are more plausibly understood as posits than are sets of propositions.
    Just three points in response to that.  First, on p. 10-11 FJ disavows ontological concern about worlds; he explicitly refuses to take sides as between concretists and nonconcretists; nor do I know of any evidence that he is a concretist.  Second, concretists are very few, and there's a good reason for that: Concretism is, to say the least, highly problematic.  (See Ch. 4 of my Modality and Meaning.  Of course, there is no unproblematic ontological interpretation of worlds.)  Third, even if concretism is true, there is no sound motivation for selectivity about which kinds of worlds to countenance.  In particular, Lewis' argument against impossible worlds fails (Modality and Meaning, pp. 39-40).  And, to date, I have seen no argument for the "existence" in whatever sense  of possible worlds  that is not equally an argument for that of impossible ones.

    A further argument against the "metaphysical"/"conceptual" distinction is suggested by FJ's remarks on the "methodological objection" (p. 80).  He reminds us that the alethic modalities are not primarily features of sentences.  He says that what we should be talking about is possibilities themselves and how many kinds of them there are.  That doesn't answer the obvious question of what items the modalities are, then, features of.  But presumably they are features of propositions or states of affairs.
    Now, Kripkean-Putnamian believers in a posteriori necessities hold their belief primarily because of a posteriori identities and the Marcus-Kripke point that genuine identities (identities whose terms are rigid designators) are necessary.  But someone might argue that any two true identity sentences whose terms are rigid designators of the same individual express the same proposition.  "Mark Twain = Samuel Clemens" expresses just the same singular proposition as does "Samuel Clemens = Samuel Clemens," viz., the proposition that that person is that person.  "Water = H2O" expresses just the same proposition as do "Water = water" and "H2O = H2O," viz., that that stuff is that stuff.  Moreover and more generally, outside intensional contexts coreferring rigid designators may be substituted in sentences salva propositione, so "Some water is not H2O" expresses the same proposition as "Some water is not water" and "Some H2O is not H2O."
    So if we are individuating worlds according to sets of propositions, there is (at least as yet) no proposition that holds in a conceptually possible world but not in any metaphysically possible world.  In particular, the proposition that some water is not H2O does not hold in a conceptually possible world, because it is one and the same as the proposition that some H2O is not H2O.  Same for the proposition that Twain is taller than Clemens, which is just the proposition that Clemens is taller than Clemens.
    (It may be that there are other types of sentence that do express propositions that are conceptually possible though metaphysically impossible: "There are distinct physical objects that occupy the same region of space at the same time"; "Some abstract entities have causal powers."  But let's waive that and pursue the question just with respect to a posteriori necessary identities.)
    Now:  A first problem with the foregoing argument is that the principle (P) "Any two true identity sentences whose terms are rigid designators of the same individual express the same proposition," presupposes that the terms are also directly referential, i.e., that the terms are Millian names; or at least, the principle is plausible only if the terms are Millian names.  According to the fairly popular Direct Reference theory of proper names, such names are indeed Millian.  But notice carefully that Millianness does not follow from rigidity.  According to some other theories of proper names, names are semantically equivalent to rigidified or otherwise rigid descriptions.  So principle (P) is without support.
    Of course, anyone who does in fact accept DR is subject to the present argument; s/he need not accept the unsupported (P).  And I should admit that I myself hold a slightly weakened version of the DR theory (see Ch. 7 of Modality and Meaning).  But DR faces terrible problems, and not even its most ardent fans should accept it with any great confidence.
    In any case, the argument breaks down when it comes to natural-kind terms.  For no one, not even a raving DR theorist who holds that any two coreferring proper names are synonymous, thinks that "water" and "H2O" are synonymous.  It is an unquestioned given of the Kripke-Putnam literature that "water" and "H2O" differ in meaning.  Since "water" and "H2O" do differ in meaning, "Some water is not H2O" expresses a different proposition from that expressed by "Some water is not water," and ditto for "Some H2O is not H2O."
    Of course, on the Kripke-Putnam view, "water" rigidly designates H2O, and "water" and "H2O" are necessarily equivalent.  Being necessarily equivalent, they apply to exactly the same bits of stuff at every metaphysically possible world.  But necessary equivalence does not suffice for synonymy; if it did, all tautologies would be synonymous.<6>  A proposition is not just a set of worlds, but a set of worlds structured according to the proposition's internal conceptual structure.  (Cf. Carnap on intensional isomorphism; this goes back to the matter of hyperintensionality, taken up in my first Peacocke handout.)  So there is after all a proposition, expressed by  "Some water is not H2O," that holds in a conceptually possible world but in no metaphysically possible one, and the foregoing argument fails.

    Finally, notice that even if one is persuaded by FJ that even if there is no "metaphysical"/"conceptual" distinction as between worlds, the all-important "question" of a priori deducibility of macro- from micro- would still remain.  That is why I'm a bit puzzled that both FJ and Chalmers make such a big deal of trying to collapse the "metaphysical"/"conceptual" distinction as applied to worlds.  I await enlightenment on this point.

    Ad "A Simple Argument to Finish With" (pp. 83-84), I'll content myself with a very quick answer to FJ's rhetorical question, "...where in that progression can the physicalist plausibly claim that failure of a priori deducibility of important facts about these organisms and creatures emerges?":  Why, at the point in the progression where natural selection cuts in and starts dramatically cross-classifying biological structures and functions against chemical kinds.  (Granted, that "point" is pretty vague.)
 

Chapter 4

    At the end of Ch. 3 (p. 86), FJ says that the theory of color to be promoted in Ch. 4 will "draw...on the methodology defended and explained in the first three chapters."  And of course there notoriously is a "location problem" for color.  So we might expect that FJ would endeavor to show that ascriptions of color are (semantically, conceptually) entailed by the truths of microphysics, either by the method of Jackson-Chalmers 2-D derivations or otherwise.  But unless I've missed it, he does not.  I myself love and share the primary quality theory of color.<7>   And FJ's argument for the theory is ingenious (though, I believe, unfortunately flawed<8>).  But so far as I can tell, neither the theory nor the argument does draw on the methodology defended in Chs. 1-3.
    FJ calls his opening instance of the "prime intuition" (p. 89) "a subject-determining platitude."  The reference to subject-determining suggests his thesis that "if we give up too many of the properties common sense associates with...[X] as represented by the folk theory of...[X], we do indeed change the subject, and are no longer talking about...[X]" (p. 38).  I assume "folk theory" in the latter passage is meant in sense (1), the Lewisian sense, but that disambiguation does not automatically apply to p. 89.
    If on p. 89 FJ has sense (1) in mind, what are the other platitudes that make up the folk theory of color?  On Lewis' implicit-definition view, any given platitude may be rejected; it's just that we cannot reject too many of the platitudes without changing the subject.   So, on the sense-(1) interpretation, we're left unsure as to what penalty would be incurred if we balked at the prime intuition.  If on the other hand FJ intends sense (2), that would be tantamount to saying that the prime intuition is itself a conceptual truth, analytic.<9>  But that would hardly convince. The opening instance on p. 89 is very plausible, at least to me, but it will be controversial, hardly true in virtue of meaning.
    No matter!  My main point here is that nothing about FJ's argument requires us to settle either of the foregoing issues.  What's important is only whether the prime intuition is correct.   If Premise 1 is true, no matter what its modal or epistemic status, and the rest of the argument is sound, the argument succeeds; if Premise 1 is not true, the argument fails.
    What about the Strong entry by entailment thesis?  So far as I've seen in Ch. 4, it is neither appealed to nor defended for the case of color.  Possibly FJ might defend it after the fact:  Perhaps there is a "red" role, perhaps given by the formulation at the top of p. 97, and a Jackson-Chalmers derivation might be constructed:

   This object's surface has complex physical property P.

   P is the actual occupant of the "red" role.  [Derivable somehow(?!) from microphysics]

   The actual occupant of the "red" role = redness.  [Allegedly contingent a priori]

 \  The object's surface is red.  [Substituting identicals]  QED

But as before, the soundness of FJ's argument for the primary quality theory does not depend on the soundness of such a derivation, or in any other way that I can see on the Strong entry by entailment thesis, or more generally on the 2-D apparatus.  Perhaps I'm missing something.
 
 

1 Qualification:  FJ's remarks on Quinean paraphrase (pp. 45-46) suggest a sense in which I might be talking about something much the same as a belief, as near as matters for practical purposes to a belief, even though the thing is not strictly a belief.
     This may help to explain FJ's distinction (pp. 42-43) between a modest and an immodest role for conceptual analysis, which distinction I do not yet understand (see below).

2 FJ refers us to his current doubts about the Knowledge Argument.  But I don't see how the failure of that argument turned on any immodest use of conceptual analysis.  It did give a big role to "intuitions about possibilities," but only one small subgroup of the argument's detractors disputed the intuition on which the argument was based, that Mary could know all the scientific and other objective facts without knowing what it is like to see red; much more commonly, the argument was rejected as invalid on its intended reading.

3 One is that the vocabulary of macro- theories bears no semantic relation to that of micro- ones; what possible semantic relation could there be between "lung" and "lepton"?  Another is that the taxonomies of the special sciences--biology, psychology, economics, etc--all cross-classify with each other and with chemistry; there are no "bridge principles" of the sort assumed by the positivists.  (Yet another reason is that nothing semantically or conceptually entails anything, but it would be unilluminating to bang the table about that.)

4 The argument depends on rejecting so-called Analytical Functionalism, and (I have been told) FJ is no longer sure he wants to do so.

5 In the process, he vindicates Dorit's interpretation by explicitly assimilating what-is-understood to Kaplanian character.

6 As we've seen, Stalnaker takes the extreme position that necessary equivalence does suffice for synonymy; but I know of no one who has been persuaded, and I'm fairly sure that FJ himself has not.

7 Consciousness and Experience (Bradford/MIT, 1996), pp. 72-74.

8 Premise 1 (p. 93), an instance of the "prime intuition," is that yellowness is the property of an object putatively presented in visual experience when the object looks yellow.   Notice that "putatively" there is ineliminable, for many theorists deny that redness is really a property of objects; indeed, projectivisms about color are widespread.   But Premise 2 tries to apply a normal-cause psychosemantics to looks such as looking-yellow, as motivated back on pp. 89-90, in a way that seems to ignore the "putatively" even though the word is reproduced there:  How could the property be a normal cause of anything unless it were not merely putative but really a property of the object?  Thus, it seems to me that premise 2 begs the question against eliminativist and projectivist opponents.  But perhaps I've misunderstood  FJ in some way.

9 Incidentally, I assume that the opening instance of the prime intuition is not essentially metalinguistic, even though it has been formulated metalinguistically.  Subsequent passages suggest that FJ would be happy to reformulate it as "Redness is the property of an object...."