This excerpt is provided, in screen-viewable form, for personal use only by members of MIT CogNet. Unauthorized use or dissemination of this information is expressly forbidden. If you have any questions about this material, please contact cognetadmin@cognet.mit.edu

Chapter 6

Do We Think in Mentalese ? Remarks on SomeArguments of Peter Carruthers

For better or worse, I am much inclined to the view that the vehicle ofmore or less alI I thought is an innate, nonnatural language. It has becomethe custom to call this putative language in which one thinks Mentalese,and I shall do so in what follows. I'm also much inclined to the view thatthe semantics of Mentalese is informational and atomistic- informationalin the sense that Mentalese symbols mean what they do in virtue of theircausal/nomic connections to things in the world; and atomistic in the sensethat the metaphysically necessary conditions for anyone Mentalese symbol

to mean what it does are independent of the metaphysically necessaryconditions for any other Mentalese symbol to mean what it does.

Actually, Carruthers (1996; all page references in this chapter are to thisbook) agrees with much of this. In particular, he thinks that the vehicle of. thought is linguistic, and that a lot of concepts are probably innate. But hedisagrees on two essential points: He holds that all conscious thinking isdone in (as it might be) English;1. so English is the language of consciousthought. And he denies that English has an atomistic semantics. Ergo, hedenies that the language of conscious thought does.

Refreshingly, Carruthers doesn't claim that he has knock-down arguments for any of this; he holds only that his is the view of preference

ceteris paribus. In particular, he takes it to be introspectively plausible thatone thinks in English;" and, if there are no decisive reasons for supposingthat one doesn't, we ought to accept the introspectively plausible view."All I need to show is that Fodor has no definitely convincing argumentfor the thesis that the language of all thinking is Mentalese, in order forthe introspective thesis to win by default" (134). Accordingly, much ofthe first half of Carruthers's book is given to arguments that are supposedto show that the arguments that are supposed to show that we think inMentalese aren't decisive.3 These are the arguments I propose to discussin what follows.

Not, however, that I disagree with Carruthers's conclusion. It's true thatthe arguments that we think in Mentalese aren't decisive. This is, quitegenerally, not the kind of area in which decisive arguments are available;

rather, it 's the kind of area in which one has had a good day if one hasmanaged not to drown. But I do think the arguments, for the view thatone thinks in Mentalese are pretty plausible; indeed, I think they are themost plausible of the currently available options. And I don

't think thatCarruthers is right about what he says is wrong with them.

Carruthers has three major points to make; one is about the relationbetween the Mentalese story and the Griceian program in semantics; oneis about the relation between the Mentalese story and the notion of functional

definition; and one is about the relation between the Mentalesestory and semantic holism. I propose to consider each of these. First,however, a brief word or two about the introspection argument itself.

To begin with , it's not so introspectively obvious that it

's introspec-tively obvious that one thinks in English. The psychologist Ken Forsteronce told me that when he eavesdrops on his thinking, what he hears ismostly himself saying encouraging things to himself.

"Go on, Ken; youcan do it . Pay attention. Try harder.

" My own thought process es, for whatit 's worth, are no more transparent, though the internal commentary thatgoes with them is generally less sanguine:

"I can't solve this; it 's too hard.I'm not smart enough to solve this. If Kant couldn

't solve this, how canthey possibly expect me t07

" and so forth. It could be, of course, thatpeople

's introspections disagree.Anyhow , as Carruthers is fully aware, it

's surely not introspectivelyavailable that what one does when one talks to oneself is thinking;specifically, it

's not introspectively available that what one finds oneselfsaying to oneself is what plays the causal/functional role of thought in thefixation of belief, the determination of action, and the like. Introspectionitself can't show this, for essentially Humean reasons: Post hoc is onething, propter hoc is another; causal connectedness can

't be perceived, ithas to be argued for. That is as true of the inner sense as it is of the outer.

Finally- though this is a long story, and not one I want to dwell onhere- I sort of doubt that the theory that we think in English can actuallybe squared with the introspective data, even if we

're prepared to take thelatter at face value. The problem is this:

"

thinking in English" can't just be

thinking in (sequences of ) English words since, notoriously, thoughtneeds to be ambiguity-free in ways that mere word sequences are not.

There are, for example, two thoughts that the expression "

everybodyloves somebody" could be used to think, and, so to speak, thought is required

to choose between them; it 's not allowed to be indifferent betweenthe possible arrangements of the quantifier scopes. That

's because, sansdisambiguation,

"

everybody loves somebody" doesn't succeed inspecifying

something that is susceptible to semantic evaluation; and susceptibilityto semantic evaluation is a property that thought has essentially. You can,

64 Chapter 6

Do We Think in Mentalese? 6S

to be sure, say in your head "everybody loves somebody" while remainingcompletely noncommittal as to which quantifier has which scope. Thatjust shows that saying things in your head is one thing, and thinkingthings is

'

quite another.I take the moral to be that "everybody loves somebody" isn't, after all,

a possible vehicle of thought. The closest you could come to thinkingin English would be to think in some ambiguity-free regimentation ofEnglish (perhaps in formulas of what Chomsky calls "LF" (roughly, Logical Form). Maybe, for example, what's in your head when you think thateverybody loves somebody on the interpretation where "everybody" haslong scope, is "every.r somey (x loves y)." That (give or take a bit) isthe right sort of linguistic structure to be the vehicle of a thought. But(dilemma) it 's surely not the sort of linguistic structure that is given toanybody's introspection; if it were, we wouldn't have needed Fregeto teach us about bound variables. Maybe intuition does say that youthink in word sequences; but there's good reason to think that it 's wrongto say so.

It is, to put the same point a little differently, a defining property ofmental representation theories to hold that the causal role of thoughts inmental process es is formally determined by the structure of their vehicles.This is where the representational theory of mind intersects the computa-tional theory of mind, and I take it to be common ground between meand Carruthers. There are, however, two different causal roles thatconverge on the sequence of words "everybody loves somebody." Thatbeing so, this sequence of words corresponds to two different vehiclesof thought even assuming that we think in English. What distinguish esthe vehicles- what the difference in their causal/computational powersdepends on- is their synt.ax.4 And the syntax of English isn't intro-spectively available. Carruthers can have it that we think in English, orhe can have it that we have introspective access to the vehicles of ourthoughts; but he can't have it both ways.

Ambiguity isn't all that's wrong with English sentences as vehicles ofthought, by the way; for sometimes they tell you not too little but toomuch. Was what you thought that it was surprising that John arrived early,or was it that John's arriving early was suprising? If you find that your introspections

are indifferent, then either they are unreliable or you mustn'thave been thinking in English since, of course, English distinguish esbetween these expressions. Carruthers holds that it 's precisely the conscious

thinking that you do in a natural language. So, if your thought thatit was surprising that John arrived early was conscious, there must be an(introspectively available) fact of the matter about which sentence formyou thought it in. Well, but is there?

However, let's put the introspection stuff to one side. I'm anxious to

get to Carruthers's arguments against the arguments that we think in

Mentalese.

First argument: Mentalese and the Grice program Suppose that thoughtsare relations to Mentalese sentences, and that the content of a thoughtis the meaning of the Mentalese sentence to which it is related. Andsuppose not only that Mentalese is a nonnatural language, but that it

'san onto logicaly independent nonnatural language; that is, a Mentalese sen-tence's meaning what it does doesn

't depend on any natural languagesentence meaning what it does. What, on such a view, ought one tosay about the meanings of natural language sentences themselves?

According to Carruthers, the thing for a mend of Mentalese to dowould be to adopt

"a Griceian approach to natural-language semantics,according to which sentence-meaning is to be explained in terms of thecommunication of beliefs. . . . [ This] takes the notions of belief and intention

for granted and uses them in explaining the idea of natural languagemeaning" (74).

However, as Carruthers rightly remarks, "Since Grice wrote, there has

been a rich literature of counterexamples to [the Griceian] account" (77);and, probably, nobody now thinks that the reduction of the meaning ofEnglish sentences to facts about the communicative intentions of Englishspeakers- or, for that matter, to any facts about mental states- is likelyto go through. This is just a special case of the general truth that nothingever reduces to anything, however hard philosophers may try . So, then,here's the way Carruthers sees the geography: The Mentalese story aboutthought is hostage to the Grice story about natural-language meaning;and the Grice story about natural-language meaning is a check that probably

can't be cashed; so the Mentalese story about thought is, to thatextent, in trouble.

However: Carruthers is wrong to think that it matters to Mentalesewhether Grice can cash his checks. More generally, it doesn

't matter toMentalese whether any story about natural-language meaning can bemade to work. More generally still, it doesn

't matter to Mentalesewhether there is such a thing as the meaning of a natural-languageexpression or (in case this is different) whether natural-language expressions

have meanings. What Mentalese cares about- and all that it caresabout that connects it with the Grice program~ is that natural languageis used to express thought.

What's required for natural language to express thought is that knowing a natural language can be identmed with knowing, about each of its

sentences, which thought it is used to express. It might turn out to be possible to parley this condition into a theory of natural-language meaning.

66 Chapter 6

On the face of it, it 's plausible enough that whatever express es thoughtmust ipso facto have a meaning that is determined by the content of thethought that it express es. Cashing that intuition is, in effect, what Grice'stheory of meaning was up to. But- to reiterate- neither the claim thatwe think in Mentalese, nor the claim that knowing a natural language isknowing how to pair its expressions with Mentalese expressions, requiresthat this be so.

Suppose, for example, that there is no answer at all to questions like'What does 'the cat is on the mat' meanf' just as there is no answer at allto questions like

'What do tables and chairs meanf' or 'What does thenumber 3 mean?" English sentences, numbers, and tables and chairs, are,by this supposition, not the sorts of things in which meaning inheres.Perhaps only thoughts or mental representations are. It wouldn't followthat there is no answer to questions like

'What thought is 'the cat is onthe mat' used to expressf

' In fact, it 's used to express the thought thatthe cat is on the mat, whether or not it thereby means that the cat ison the mat; indeed, whether or not it means anything. To repeat: Neitherthoughts having content, nor the use of language to express thoughts,depends on language having content. Not , at least, by any argument thatI can think of; certainly not by any of the arguments that Carruthers hason offer.

But, if that's right, then the Mentalese story about thinking doesn'tdepend on the Griceian reduction going through after all. The most thatcould follow from the failure of Grice's reduction is that he was wrong toassume that whatever is used to express thought must mean something. Itwould, no doubt, be surprising and rather sad if Grice was wrong aboutthis, since the intuition that sentences, words, and the like do mean something

would have to be given up,s and it is a plausible intuition, as previously remarked.6 But, once again, the status of Mentalese wouldn

't be inthe least impugned.

Here's a slightly different way to put this point. Grice had a theory oflinguistic communication and a theory of linguistic meaning, and his idea wassomehow to ground the latter theory in the former. Now, Grice's theoryof linguistic communication is practically untendentious, at least for anybody

who is a realist about intentional states. The theory was that youutter sentences intending, thereby, to indicate or express the content ofthe mental state that you're in; and your hearer understands you if hedetermines from your utterance the content of the mental state that youintended it to express or to indicate. I take this theory of communicationto be practically platitudinously true. So far as I know, none of the myriadcounterexamples to the Grice program that you find in the literature is anobjection to his communication theory.

Do We Think in Mentalese1 67

Notice, however, that this communication theory could be true even ifmental states are the only things in the world in which content inheres. Inparticular, no commitments at all about natural-language meaning enterinto it . This is not an accident; indeed, it 's essential from Grice's point ofview that his account of communication should not presuppose a notionof natural-language meaning. For his goal was to derive the theory ofmeaning from the theory of communication, and he had it in mind thatthe derivation should not be circular.

I suppose Mentalese does need Grice's communication theory; or, anyhow, something like it . An acceptable account of thought ought to say

something about how thoughts are expressed; and if thoughts are whathave content in the first instance, then it is natural to suppose that whatcommunication communicates is the content of thoughts. But thoughit needs his theory of communication, Mentalese doesn't need Grice'stheory of natural-language meaning; or, indeed, any theory of natural language

whatsoever. For the Mentalese story is not just that the content ofthought is prior to natural-language content in order of explanation; theMentalese story is that the content of thought is onto logically prior tonatural-language meaning. That is, you can tell the whole truth aboutwhat the content of a thought is without saying anything whatever aboutnatural-language meaning, including whether there is any. Here, accordingto the Mentalese story, is the whole truth about the content of a thought:it 's the content of the Mentalese sentence whose tokening constitutes thehaving of the thought. (This, by the way, is about as un- Wittgensteiniana view of the conceptual geography as it is possible to hold. If theMentalese story about the content of thought is true, then there couldn't bea private language argument. Good. That explains why there isn

't one.)It may be just as well that Mentalese leaves all the questions about

natural language semantics open, because- so it seems to me - it reallymight turn out that there is no such thing. Maybe all there is to what"cat" means is that it 's the word that English speakers use to say whatthey are thinking about when they are thinking about cats. That, ofcourse, isn't semantics in either of the familiar construals of the term: Itdoesn't specify either a relation between the word and its meaning or arelation between the word and what it 's true of. It can't, however, be heldsimply as a point of dogma that English has a semantics; if you think thatit does, then you will have to argue that it does. Grice failed to say whatnatural-language meaning is; and so too, come to think of it, has everybody

else. Maybe natural-language is the wrong tree for a theory ofmeaning to bark up.

Bottom line: Carruthers thinks Mentalese is hostage to the Grice program. But it 's not. So the failure of the Grice program isn't an argument

against the claim that we think in Mentalese.

68 Chapter 6

I turn, briefly, to Carruthers's last two arguments, both of which concern the doctrine that the semantics of Mentalese is atomistic.

To begin with, not every friend of Mentalese accepts this doctrine.Indeed, the standard view is that the meaning of Mentalese expressionsis, at least partly, determined by their functional (inferential) roles. Car-ruthers himself holds that this is so, both for Mentalese and for English.But I have resisted assimilating the Mentalese story about thought to afunctional role theory of meaning. That's because I believe that functionalrole semantics is intrinsically holistic and that a holistic semantics isincompatible with any serious intentional realism.

The argument for the second step is familiar: If semanti~.s is holistic,then the content of each of your thoughts depends on the content of eachof your others; and, since no two people (indeed no two time slices of thesame person) have all their thoughts in common, semantic holism impliesthat there are no shared intentional states. Hence that there are no soundintentional generalizations. Hence that there are no intentional explanations

. But intentional Realism just is the idea that our intentional mentalstates causally explain our behavior; so holistic semantics is essentiallyirrealistic about intentional mental states. So the story goes.

Carruthers has two points to make against this, neither of which seemsto me to work, but both of which keep cropping up in the recent literature

. Let's have a look at them.First, though Carruthers admits that there are lots of differences

between people's actual beliefs, he thinks that inferential role semantics,rightly formulated, can avoid concluding that there are no, or practicallyno, belief contents that people share. The idea is that "[the contents of]propositional attitudes are individuated in terms of their potential [sic]causal interactions with one another. These .. . may remain the sameacross subjects whose actual propositional attitudes may differ. . . . mhesame conditionals can be true of them" (111). So, for example, you knowthat today is Tuesday, so you infer that tomorrow is Wednesday. I don'tknow what day it is today, so I'm agnostic about what day it will betomorrow. But that's okay because we do have this hypothetical in common

: if we think today is Tuesday, then we think tomorrow is Wednesday. Since, according to the present proposal, the functional role of an

attitude, and hence its content, is to be defined relative to these sharedconditionals, the eccentricity of actual beliefs needn't imply that beliefcontents can't be shared. As I say, this sort of idea is getting aroundd

. BI~ 1993).Well, it may help a little, but it surely doesn't help much. For one thing,

belief hypothetic als shift along with the categoric als when indexicals areinvolved. It's because I believe that today is Tuesday that I believe that ifI ask you, that's what day you'll say it is. If content identity is relativized

Do We Think in Mentalese1 69

to such hypothetic als, it's going to exhibit just the sensitivity to individual

differences that Carruthers is trying to avoid.And even where "eternal" propositions are at issue, it

's a classic problem for functional role semantics that, on the one hand, . what functional

relations hold between your beliefs depends on what you think is evidence for what; and, on the other hand, what you think is evidence for

what depends, pretty globally, on what else you happen to believe - thatis, on what you categorically believe. You think Clinton is a crook becauseyou've read that he is in the Times. I haven't read the Times, so we differin categorical beliefs (viz., about what it says in the paper), so we can'tshare relevant Clinton-belief contents. Thus the old holism. Notice howlittle the move to hypothetic als does to help with this. What, for example,is the status, in the circumstances imagined, of the conditional: "If one hasread the Times, then one thinks that Clinton is a crook." Answer: thoughit 's true of you by assumption, it

's quite likely to be false of me: If I'd read

that about Clinton, dogmatic Democratic that I am, I would have concluded not that he's a crook but that the Times is unreliable. One man's

modus ponens is another man's reductio, as epistemologists are forever

pointing out.Examples like

"if you think it's Tuesday today, then you think it 'sWednesday tomorrow," where belief hypothetic als are plausibly content-constitutive (and hence insensitive to, inter alia, the contents of contingent

categoric als) are exceptional and therefore misleading. Rather,people who differ in their categorical beliefs are often likely to satisfydifferent belief hypothetic als for that very reason.7 Bottom line: There is ashort !:,oute from functionalism about meaning to meaning holism. As faras anybody knows, there's no way to get around that.

What's wrong with functional role semantics (FRS) is that it wants toanalyze the content of a belief in terms of its inferential (causal) relations;whereas, plausibly, the direction of metaphysical dependence actuallygoes the other way ' round. The content of a belief determines its causalrole; at least, it does in a mind that's performing properly. Holism isNature's way of telling FRS that it has the direction of analysis backwards.None of this is discern ably altered if FRS goes hypothetical.

Alright, alright, but doesn't that prove too much? "[ Holism] is a problemfor functional individuation generally, not a problem for functional rolesemantics in particular. So, unless we are prepared to accept that therereally are no functionally individuated entities, we have as

.

yet been givenno reason for rejecting functional individuation of thoughts" (114). (Seealso Devitt , 1996, where this sort of line is pushed very hard.)

Well, but does avoiding the holistic embarrassments of FRS reallyrequire rejecting functional analysis per se1 That, to be sure, would be a

Chapter 6

steep price; more than even people like me, who hate FRS root andbranch, are likely to be willing to pay.

Notice, to begin with, that the claim that the individuation of such andsuches is functional is, in and of itself, a piece of metaphysics pure andsimple; in and of itself, it need have no semantical implications at all. Suppose

, for example, I claim that hearts are functionally individuated; in particular, that hearts just are whatever it is that pumps the blood. Making

this claim presumably commits me to a variety of metaphysical necessities; if it is true, then "hearts are pumps

" and all of its entailments areamong them. But it commits me to nothing whatever that's semantic. Forexample, it leaves wide open all questions about what, if anything, theword "heart" means; or about what the principles of individuation for theconcept HEART are; or about what the possession conditions are forhaving that concept.

The reason that claiming that hearts are pumps leaves all this open isthat you can't assume- not at least without argument- that if it 's metaphysically

necessary that Fs are Gs, then it follows that "F" means something about being G; or that knowing that Fs are Gs is a possession

condition for the concept F (or for the concept G). Thus, for example: It 'smetaphysically necessary that two is the only even prime. It doesn'tfollow that "two" means the only even prime or that having the conceptTWO requires having the concept PRIME, etc. Likewise, it 's metaphysically

necessary that water is H2O. But it doesn't follow that "water"means H2O; or that you can't have the concept WATER unless you havethe concept HYDROGEN, etc.

Functionalism about hearts is, to repeat, a metaphysical thesis with,arguably, noseman tical implications whatever. Whereas, the holistic embarrassments

for FRS arise precisely from its semantical commitments; in particular, from the constraints that it places on concept possession. The intention of

FRS is that if I is an inference that is constitutive of concept C, then having concept C. requires in some sense acknowledging I (e.g., it requires

finding I "primitively compelling," or it requires being disposed to draw Iceteris paribus, or whatever). Holism follows from this construal of concept

possession together with the assumption (independently warranted,in my view) that there is no principled way to distinguish concept-constitutive inferences from the rest.

But, notice, this is semantic functionalism; it 's functionlism about havingthe concept HEART (or about knowing what the word "heart" means).Semantic functionalism does lead to semantic holism, and by familiarroutes. But onto logical functionalism is fully compatible with semanticatomism. That's because, although functionalist theories of hearts bring allsorts of metaphysical necessities in their train, they place no constraints

Do We Think in Mentalese7 71

whatever on concept possession. It's just a fallacy to infer the functional individuation

of concepts from the functional individuation of the properties thatthey express.

So it 's alright to be a functionalist about hearts; at least it's alright for

all the arguments that we've seen so far. But it 's not alright to be a functionalist

about conceptual content; that leads to holism, hence to unsat-isnable constraints on concept possession.

One last point on all of this: It is, for all I know, okay to be afunction -alist about conceptual content if you can, at the same time, contrive toavoid being a semantic holist. And, of course, an FRS functionalist canavoid being a semantic holist if he is prepared to accept an analytic/synthetic distinction; which, in fact, Carruthers does. I don

't want to argueabout whether it's okay to philosophize on the assumption of an analytic/synthetic distinction; everyone has to sleep with his own conscience.Suffice it that you don't need an analytic/synthetic distinction- or anysemantical views whatever- to be a functionalist about hearts and such;you just need to be an essentialist. Being an essentialist is quite a differentthing from endorsing an analytic/synthetic distinction (unless there is reason

to believe that truths about essences are ipso facto analytic- whichthere isn't because they aren't). That's another way to see why functionalism

about concepts doesn't follow from functionalism about things; and

why the former is so much the more tendentious doctrine. So much, then,for Carruthers's third argument.. I want to end by returning to a point I

've already made in passing.I don't think that there are decisive arguments for the theory that allthought is in Mentalese. In fact, I don

't think it 's even true, in any detail,that all thought is in Mentalese. I wouldn

't be in the least surprised, forexample, if it turned out that some arithmetic thinking is carried out byexecuting previously memorized algorithms that are defined over publiclanguage symbols for numbers ("Now carry the '2,' " and so forth). It 's quitelikely that Mentalese co-opts bits of natural language in all sorts of ways;quite likely the story about how it does so will be very complicatedindeed by the time that the psychologists get finished telling it .

But here's a bet that I'm prepared to make: For all our philosophicalpurposes (e.g., for purposes of understanding what thought content is,and what concept possession is, and so forth), nothing essential is lostif you assume that all thought is in Mentalese. Hilary Putnam onceremarked that if you reject the analytic/synthetic distinction, you'll beright about everything except whether there is an analytic/synthetic distinction

. Likewise, I imagine, in the present case: If you suppose that allthought is in Mentalese, you

'll be right about everything except whetherall thought is in Mentalese. More than that is maybe more than it

's reasonable to ask for.

72 Chapter 6

Do We Think in Mentalese1 73

Append i.r: Higher-Order ThoughtsI try never to think about consciousness. Or even to write about it . But Ido think there's something peculiar about Carruthers's treatment. I quote(174): "Any mental state M , of mine, is conscious = Misdisposed tocause an activated [i.e., not merely dispositional] conscious belief thatI have M ." Now, as Carruthers rightly notes, this formulation wouldappear to be circular on the face of it . So he immediately adds "so as toeliminate the occurrence of the word 'conscious' from the right hand sideof the definition" the following :

Any mental state M , of mine, is conscious = M (level 1) is disposed tocause an activated belief that I have M (level 2), which in turn is disposedto cause the belief that I have such a belief (level 3) and so on; and everystate in this series, of level n, is disposed to cause a higher-order belief oflevel n + 1.

But that can't really be what Carruthers wants. It eliminates the occur-rence of the word '1 conscious" from the right hand side alright; but it doesso precisely by leaving it open that each of the higher-order beliefsinvolved might be unconscious. And it is explictly Carruthers's view thathaving the unconscious n-level belief that one is having a certain n - 1level thought is not sufficient to guarantee that the n - 1 level thought isconscious.

As far as I can see, two options are open to Carruthe~s, neither of whichis satisfactory. He could simply reintroduce II conscious" to qualify each ofthe higher level beliefs (e.g., M [level 1] is disposed to cause an activatedconscious belief that I have M [level 2], etc.) - thereby, however, failing toresolve the problem about circularity. Or he could deny that '1 conscious"really occurs in I'conscious-level-n belief" for any level higher than 1;in effect, he could spell 'I conscious-level-n belief" with hyphens- thereby

, however, raising the hard question why a merely conscious-with-ahyphen level n + 1 belief should suffice to make a level-n belief conscious

without a hyphen. Notice, in particular, that conscious-with-a-hyphenbeliefs can't be assumed to be conscious sans phrase; if they were, theanalysis would lapse, once again, into circularity.

1. This is to ignore a lot of subtleties; for example, whether we sometimes think in images,and the like. Such questions may matter in the long run, but not for the project at hand.2. In what follows "English" is generally short for "English or some other natural language"and "sentence" is generally short for "sentence or some other expression." Nothing hangson this except brevity of exposition.

3. Most of the second part of Carruthers's book is about consciousness. I say a little aboutthat in the Appendix.

Notes

4. I gather from emails that Carruthers thinks what makes the differen~ between the twothoughts that everybody loves somebody is not the syntax of the representation that youtoken, but something like the way you intend the token. I'm not sure that I Imderstand that.I don't think that I intend my thoughts at all: I think that I just have then\. But, inany event, the suggested doctrine is pretty clearly not compatible with a computationaltheory of mind, so I don't want any.

S. I'm asswning that if "the cat is on the mar' is used to express the thought that the cat ison the mat, then if it means anything it means that the cat is on the mat.

6. Disquotation would still give a sense to the thought that sentences, words, and the likemean scxnething; to that extent they would be better off than numbers and tables andchairs, for which conventions of (dis)quotation have not been de6ned. Cold comfort, ifyou ask me.

.

7. There are, of coune, all sorts of other ways in whim belief hypothetic als shift with beliefcategorica1s. I am cynical; I think all politicians are crooks. So (given, as usual, reasonablerationality) irs bue of me that if I think Francis is a politician, then I think Francis is acrook. You are naive; you think no politicians are crooks. So, ir s not true of you that ifyou think Francis is a politician then you think Francis is a crook. If the contents of ourbeliefs are relativized to these hypothetic als, we get the usual dreary result: Our beliefsabout politicians aren't contradidories.

74 Chapter 6

This excerpt is provided, in screen-viewable form, for personal use only by members of MIT CogNet. Unauthorized use or dissemination of this information is expressly forbidden. If you have any questions about this material, please contact cognetadmin@cognet.mit.edu

chap6.pdf/var/tmp/PDF65678.pdfpage 2page 3page 4page 5page 6page 7page 8page 9page 10page 11page 12

This except is provided, in screen-viewable form, for personal use only by members of MIT CogNet