Phil 33401, A02—Lecture 6 notes
Putnam and Burge continued.
Distribute paper topics. Word on what’s expected.
Today we’ll wrap up the Burgean and Putnamian arguments for externalism. We’ll focus more on Burge, because we’ve discussed him less.
I. Burge’s argument revisited
Letting “A” be short for “the actual situation”, and “C” short for “the counterfactual situation”:
has the same internal properties in A and C. (supposition)
Recall that internal properties are properties that supervene on one’s internal physical constitution; better, that are constitutively independent of one’s environment (so that there are no facts about the environment the shuffling about of which would just by itself change how they are). They are a matter of how things stand with oneself independently of one’s relationships to the environment.
2. In A, “arthritis” means arthritis in Bob’s linguistic community. (supposition)
A, when Bob sincerely asserts the sentence, “I have arthritis in my thigh” he
expresses the belief that he has arthritis in his thigh. (from 2)
Burge has much to say about possible objections to this, as I’ve said. We’ll talk about them in a moment.
4. In C, “arthritis” means tharitis in Bob’s linguistic community. (supposition)
5. In C, when Bob sincerely asserts the sentence, “I have arthritis in my thigh” he expresses the belief that he has tharitis in his thigh. (from 4)
the fact that Bob believes that he has arthritis in his thigh is not an
internal property of him. (from all of above).
Even though Bob in A and Bob in C are molecule for molecule identical—even though there’s no internal difference between them—they still have different beliefs. Therefore, Bob’s believing what he does is constitutively dependent of how the word “arthritis” is used in his linguistic community. His believing what he does is not just determined by his internal makeup, but is partly constituted by how things stand in his social environment.
7. Therefore, content externalism is true (from 6).
Now, what makes it possible to imagine that Bob could have all the same internal properties in A and C even though what “arthritis” means in his language is different is that he incompletely understands its meaning, and so we can imagine such a change without seeing it as impinging on him at all. Since he’s not abreast of the feature of the meaning of “arthritis” that changes from A to C, that change can be envisioned without forcing us to think that he differs internally at all.
But this feature also raises a question about the viability of the argument. With respect to the inferences from 2 to 3 and 4 to 5 there are two questions:
a. In uttering “I have arthritis in my thigh,” did Bob say that he has arthritis in his thigh?
b. Assuming a positive answer to (a), in saying that he has arthritis in his thigh, did Bob express the belief that he has arthritis in his thigh?
What raises these questions, what can make them seem tricky, is, precisely that Bob has an incomplete understanding of the meaning of the word “arthritis”. He doesn’t fully know what it means. Given that, can he be said to say that he has arthritis? And even if so, can he be said to believe that?
With respect to (a), a word on saying.
The idea of what one says in making an utterance seems simple enough, but it’s actually rather complex. For one thing, we have to distinguish between what one literally says and what one implies in a given utterance.
Consider a famous example of Paul Grice. Suppose someone asks me what I think of a particular student. I say, “She has very good handwriting.” Now there’s clearly a notion of meaning such that I may be said to mean that the student is not very strong. This kind of meaning is opposed to the kind of meaning identical with what one says in making an utterance. I didn’t literally say that she’s not so strong. But given the circumstances, it may be reasonably thought that that’s what I meant. This kind of meaning Grice labeled speaker’s meaning. It’s closely related, if not identical, to the ordinary idea of what one implies in saying something.
Grice developed various principles that govern speaker’s meaning. Generally, there are maxims in conversations to say things that are relevant, informative, etc. If someone says something that is on its face irrelevant to what is being asked, as is clearly the case with my remark, the assumption is that in saying it the speaker intends to convey something else, something that is relevant. In this case, I intend you to reason: if that’s the best thing he can say about the student then she’s no good.
Now, it’s crucial to this kind of reasoning, and so by Grice’s lights to the possibility of speaker’s meaning, that there be something that is literally said in an utterance. The pertinent question is, what did he mean in saying that….? If this question is to make sense, there are two kinds of meaning at stake here, what is meant and what is said.
Now, what determines what one literally says? One natural thought is that it’s what the words one utters literally mean in the language (and not what one takes them to mean, if that is different).
I think this is a fair assumption (though one that not all philosophers of accept).
I think it’s the case that even in cases of total absence of understanding of the meaning of a word, we will still say that the person says what the words in the language actually mean.
Suppose I imitate someone in the use of some big word of which I have absolutely no understanding. I utter the sentence, “Bob’s didacticism is tiresome.” Now, if I have absolutely no idea at all what didacticism means, I can’t be expressing the belief that Bob’s didacticism is tiresome. But that’s what I said. If something hinged on this, you might rightly accuse me, say in a court of law, of having said that Bob’s didacticism is tiresome. (E.g., if Bob’s tiresome didacticism would violate a contract he had not to be tiresomely didactic.)
Note that it matters for this example that I’m a competent speaker of English. If a non-English speaker says that sentence, or even more so if a parrot makes that utterance, we will say not only that they don’t express the belief that Bob’s didacticism is tiresome, but moreover that the person didn’t even say that.
The reason is that you can only say things in a language you speak.
But given that you do speak a language, which I do, you can say things that you yourself don’t understand. To speak a language can’t mean to understand every word in it. Then none of us would speak English.
Why is it that people are crediting with saying what their words mean in the language even if they don’t understand it at all?
Perhaps part of the reason is that it’s crucial for communication that people, as it were, go on the record as saying this or that. (An idea of Wiggins.) A public language like English, with its repository of words that already have meanings, enables one to do this. But for this to be so, it must be the case that what speaker of a language can be credited as saying can be determined by what the words he utters really mean, not by what he thinks the words mean, if that is different.
Obviously, much more could be said on this issue. Raises very interesting questions about the nature of natural languages and their role in communication. But I will move onto question (b) because it raises issues closer to Burge’s discussion. Burge’s extended discussion of objections to his inference is best understood as directed toward doubts about a positive answer to question (b). He basically assumes that the answer to (a) should be yes.
III. Meaning-content identity thesis
Recall a principle that I introduced in discussing the general idea of mental content:
Meaning-content identity: The meaning of an assertion is identical to the content of the belief the assertion expresses.
One point here is that notion of meaning at work in the meaning-content identity thesis is linked to the idea of saying. The meaning of an assertion in this sense is what one says in making the assertion. So the claim is that one believes what one says in making a sincere assertion
If this principle is true, then it licenses a positive answer to (b).
How plausible is the meaning-content identity thesis with respect to literal meaning?
Well, on the face of it, it seems quite plausible. If I sincerely assert a sentence with the literal meaning that I’m going to Hong Kong next month, surely I should be taken as expressing the belief or knowledge that I’m going to Hong Kong next month. Surely such equations must generally obtain if speaking is to be, as it clearly is, the most straightforward and direct way of conveying what one means or knows.
Or consider what is involved in interpreting a foreign language, a context discussed much by Quine and Davidson. Davidson and Quine showed that you cannot make sense of a language new to you without at the same time ascribing beliefs to its speakers. Suppose someone utters “gavagai” in the presence of rabbits and not otherwise. It’s natural to hypothesize that “gavagai” means something like ‘hey, some rabbits’. But in making this hypothesis we’re implicitly assuming that the speaker will share our belief that rabbits are present. If we don’t take the speaker to believe a rabbit is present when one is, we surely have no reason to take the speaker to be saying that a rabbit is present. And clearly that conditional presupposes the idea that in saying that a rabbit is present, the speaker is expressing the belief that a rabbit is present. Break that connection and we have nothing to go on in interpretation.
On the other hand, clearly the meaning-identity thesis will need to be hedged in at least some respects. Consider lapses of concentration. I may utter the sentence, “I’ll be there next week,” when I really mean that I’ll be there the week after next. I don’t in uttering that sentence express the belief that I’ll be there next week. I have no such belief. Nonetheless, it seems undeniable that that’s what I said. If something important turns on when I’ll be there, you could rightly accuse me of having said that I’ll be there next week. I made a mistake in saying that, but nonetheless that’s what I’ve said. I’ve gone on record, as it were, in saying that I’ll be there next week.
(Note that in cases of slips of the tongue, say a Spoonerism like, “There goes our queer old Dean,” Burge things that we shouldn’t even say that Spooner said that there goes our queer old Dean, let alone that he believes that (p. 547). But lapses like the one given above are different. Shows how complex the phenomena are.)
Or consider the case of a total lack of understanding. Didacticism example. We’ve already said that I said that Bob’s didacticism is tiresome. But it seems radically implausible to take me in that case, to express the belief that that is so. How could I express that belief with my utterance, given that I don’t know that “didacticism” means didacticism?
So meaning-content identity can’t be understood to hold across the board. Taken as a completely general claim, it’s false. The relevant question must be: is it plausible that it holds in this case (whatever kind of case is under consideration)?
Burge grants the point about total lack of understanding for sake of argument. But he thinks the case of incomplete understanding is quite different. He thinks the fact that we don’t ascribe a belief whose content corresponds to the literal meaning of a saying in the case of total lack of understanding of a component expression doesn’t entail that we don’t or shouldn’t ascribe a belief whose content corresponds to the literal meaning of a saying in the case of partial misunderstanding or lack of understanding.
Indeed, he thinks it’s obvious that we do make such ascriptions, and moreover that we’re right to do so.
There are two sides to his defense, a positive and a negative side.
On the positive side, one simple point he makes is that the case of Bob is far from anomalous. Is it really true that the rest of us are smarter than Bob and knew before reading the article that arthritis is specifically a joint ailment? All of us, with respect to many common words, have an incomplete or faulty grasp of its meaning, of how it is to be applied. Do you know what cut of meat precisely is a brisket? Can’t you be wrong about that? Or what a contract is, or a recession, or what processes count as fermentation, or which trees are elms and which beeches, or what the baroque period is or the Mesozoic era, or a bassoon or a Murphy bed, or what have you.
Are we to say in all such cases where there’s a misunderstanding that people don’t actually believe what is expressed by the relevant sentences when they’re used correctly? Such a move, Burge argues, does tremendous violence to our ordinary practice in belief-ascription.
The doctor didn’t hesitate to ascribe the false belief to Bob. He didn’t instead say, “Well, yes what you believe is correct. You do have arthritis in your sense in the thigh.” What a pointless and misleading thing that would be to say.
It would seem that many considerations that speak in favor of our ascribing canonical literal sayings to people even when they don’t fully understand their words supports as well ascribing corresponding beliefs. For if we don’t ascribe the corresponding beliefs in cases of incomplete understanding, then given the ubiquity of such cases, we’re committed to thinking that we’re constantly talking past each other. I thought I was arguing with you about the causes of our current recession; in fact, it turns out you were expressing beliefs about a ‘theression’, because you thought it only took two quarters of decline to count as one, rather than three.
Or when someone on T.V. says, “In spite of a common misconception, clavichords must always have brass wedges hitting the strings, rather than quills plucking the strings,” that person is saying something false. Those subject to this misconception simply don’t have any beliefs about clavichords at all. So it’s not a misconception. It’s a correct belief about some other kind of thing, a ‘schmavichord.’
And so on. (Comment on ubiquity of such cases.)
On the negative side, which consumes more of the paper, Burge proceeds as follows. He says, “Okay. Suppose we don’t ascribe the belief to Bob that he has arthritis in his thigh. Or that more generally we don’t ascribe the apparently expressed beliefs when there is incomplete understanding. What beliefs should we ascribe in their place?” He then argues that any of the answers we might give suffer from devastating problems, or at the very least are less palatable, and so lend no support to overthrowing, our ordinary practice of ascribing the beliefs that correspond to the literal meanings of the words.
I won’t discuss the answer that has to do with the de re/de dicto distinction, because that’s a complicated issue that I think Burge doesn’t quite get right (there are two de re/de dicto distinctions).
1. One such answer, one such ‘reinterpretation’ of the
speaker’s belief, Burge ascribes to Descartes (p. 549).
The thought is that in cases of incomplete understanding the speaker’s belief is indefinite. It for that reason can’t be precisely specified with an English sentence.
Burge’s response is that this makes ash of the evident fact that, as he puts it, we all know what would confirm or disconfirm beliefs on the part of that speaker that would be expressed with the problem word. So Bob knows what would confirm or disconfirm his belief that, say, his friend Margie has arthritis. There appears to be a perfectly definite belief there, one that he and someone else could debate. Bob would know for the most part just what evidence to appeal to (She walks slowly, grimaces while climbing stairs, etc.)
2. Another answer is to ascribe a belief involving a
concept whose boundaries are delimited precisely by Bob’s misconception or
partial conception. So let us use ‘tharitis’ as a word for any rheumatoid
ailment. Then we can say that Bob believes that he has tharitis in his thigh,
even in A. And then of course his belief is true.
Burge has many things to say about this view (549-550). For one, it can often be hard or impossible to find a concept that precisely corresponds to a person’s misconception. Perhaps the subject doesn’t have a set of clear and precise beliefs about what arthritis is that determines in all cases what is to count as tharitis (i.e., the concept corresponding to his misconception of arthritis).
Moreover, people do not generally, when informed of an error such as Bob’s say, “But I wasn’t wrong. I was right. For I meant….” (We would find this reply, Burge says, quite disingenuous.) Bob will probably instead be relieved to be wrong. For he didn’t want arthritis, knowing what he does about it. He was thinking about arthritis, and he’s glad he doesn’t have it.
3. Burge also discusses at length the thought that Bob
and others like him should be taken to have a belief not about arthritis or
whatever, but about English, e.g., “I have something in my thigh that the word
‘arthritis’ is true of.” Then Bob will have the same belief in A and C, and
it’s being wrong in one case and right in the other is just a matter of the
relevant facts being different, not of the content of the belief being
Again, Burge argues squares horribly with things we’re inclined to say. Won’t elaborate here.
Burge is happy to keep appealing to ordinary practice. For he thinks we need some reason to overturn it, and he doesn’t think any of these alternatives provide a good case for doing so.
IV. Putnam again
Just a couple points on Putnam.
Recall that Putnam asks us to imagine a planet very much like ours existing somewhere else in the universe. So similar there are even doppelgangers of us—people that are molecule for molecule identical to us. There’s only one real difference between our world and Twin Earth: wherever there’s water here, there’s some other liquid—call it schwater—there. Schwater is like water in almost every discernible way—it’s colorless, odorless, quenches thirst, etc. Nonetheless, it is not water. It’s in fact a totally different substance—whereas water is H2O, schwater has some complicated chemical composition which Putnam abbreviates as XYZ.
(Note that it is an irrelevant feature of Putnam’s specific choice of example that it’s hard to imagine doppelgangers of us in a world without water—because our own bodies are themselves made out of water. So you’d have to imagine that the environment contains schwater wherever ours contains water but nonetheless that bodies are made out of water. This complication has no bearing on the argument, could be eliminated by choosing some other substance, we’ll ignore it.)
Now consider some particular English speaker on our world, Jane, and her doppelganger, Twin Jane. Jane and Twin Jane, by supposition, are the same with respect to internal properties. They have all the same microphysical properties, all the same chemical properties, all the same neurological properties and so on. That is, for every internal fact of the form “Jane is F” there is a corresponding fact of the form “Twin Jane is F”. There is simply nothing to distinguish Jane and Twin Jane internally, nothing to distinguish them, that is, apart from the fact that they inhabit environments with different features. So now we may argue (this is just one way to set up the argument):
[Run parallel argument for Putnam without writing it down.]
Two quick points here.
First of all, although Putnam makes much of what he calls the linguistic division of labor, his argument for externalism does not depend on the social environment like Burge’s. It’s true that we nowadays have scientists who can fix the extension of “water” or “gold” precisely even if the rest of us can’t. But we don’t need that fact to run the thought experiment, as Putnam himself acknowledges in asking us to imagine that it happens in 1750, when no one knows the chemical composition of water.
What the case turns on is rather his idea, which we’ve discussed, that the meaning of the word “water” is determined in part by the actual nature of the natural kind with which speakers of English are in causal contact.
He thinks it’s perfectly possible to imagine a case in which the nature of the natural kind to which we apply the word “water” is different from what it actually is, but not in a way that any of us could notice. Such a case would be Twin Earth (circa 1750).
If so, then, the word “water” would mean something different but not in any respect that we notice and so that impinges on our internal properties.
Putnam does want to think, as we’ve discussed, of the ‘essences’ of natural kinds as scientifically discoverable. But for these essences to set the boundaries of the extension of our natural kind terms they don’t have to actually be scientifically discovered. Precisely not. They do their work whether or not we know what they involve.
So we don’t need a linguistic community at all to run the thought experiment (unlike Burge). Imagine that I encounter a new natural kind while traveling in some faraway place, say, a kind of plant. I decide to call that kind of plant, “boojum”. No one else can have any thoughts about boojum, because no one but me has ever had any contact with it. So now I think, “Boojums are probably pretty tasty.”
Now imagine a twin me exploring a remote place on Twin Earth. The only difference is that everywhere boojum is here there’s twin boojum, a different but very similar looking plant. Now….
The second point is that Putnam himself might not accept the externalist conclusion, at least at the time of these essays. For he seems to commit himself, after some waffling, to the view that genuine mental states are narrow, that is to say, that possession of genuine mental states simply cannot be constitutively dependent on the environment. That would rule out externalism by fiat.
Why anyone would think that will be the subject of our next unit, so I’ll hold off on it for now.