Must Singular Thoughts be Object-Dependent?

What follows are some ideas from London Ed about a book he is writing.  He solicits comments.  Mine are in blue.

The logical form thing was entertaining but rather off-topic re the fictional names thing. On which, Peter requested some more. 

Let’s step right back. I want to kick off the book with an observation about how illusion impedes the progress of science. It looks as though the sun is going round the earth, so early theories of the universe had the earth standing still. It seems as though objects are continuously solid, and so science rejected the atomists’ theory and adopted Aristotle’s theory for more than a millenium.

A final example from the psychology of perception: in 1638, Descartes takes a eye of a bull and shows how images are projected onto the retina. He finally disproves the ‘emissive theory of sight’. The emissive theory is the naturally occurring idea that eyesight is emitted from your eye and travels to and hits the distant object you are looking at. If you ask a young child why you can’t see when your eyes are shut, he replies (‘because the eyesight can’t get out’). 

Scientific progress is [often] about rejecting theories based on what our cognitive and perceptual framework suggests to us, and adopting theories based on diligent observation and logic.

We reject ‘eyebeams’. We reject the natural idea that the mental or sensitive faculty can act at a distance.  When we look at the moon, science rejects the idea that a little ethereal piece of us is travelling a quarter of a million miles into space. Yet – turning to the main subject of the book – some philosophers think that objects themselves somehow enter our thoughts. Russell writes to Frege, saying “I believe that in spite of all its snowfields Mont Blanc itself is a component part of what is actually asserted in the proposition ‘Mont Blanc is more than 4,000 metres high”. Kaplan mentions, with apparent approval, the idea that the proposition ‘John is tall’ has two components: the property expressed by the predicate ‘is tall’, and the individual John. “That’s right, John himself, right there, trapped in a proposition”.  The dominant theory in modern philosophical logic is ‘direct reference’, or object-dependent theories of semantics: a proper name has no meaning except its bearer, and so the meaning of ‘John is tall’ has precisely the components that Kaplan describes.

BV:  I too find the notion that there are Russellian (as opposed to Fregean) propositions very hard to swallow.  If belief is a propositional attitude, and I believe that Peter is now doing his grades, it is surely not Peter himself, intestinal contents and all, who is a constituent of the proposition that is the accusative of my act of belief.  The subject constituent of the proposition cannot be that infinitely propertied gnarly chunk of external reality, but must be a thinner sort of object, one manageable by a finite mind, something along the lines of a Fregean sense.

The purpose of the proposed book is to advance science by showing how such object-dependent theories are deeply mistaken, and also to explain why they are so compelling, because of their basis on a cognitive illusion as powerful as the illusions that underlie the geocentric theory, or the emissive theory of sight.

What is the illusion? The argument for object dependence is roughly as follows

(1) We can have so-called ‘singular thoughts’, such as when we think that John is tall, i.e. when we have thoughts expressable [expressible] by propositions [sentences, not propositions] whose subject term is a proper name or some other non-descriptive singular term.

(2) A singular term tells us which individual the proposition is about, without telling us anything about it. I.e. singular terms, proper names, demonstratives, etc. are non-descriptive. They are ‘bare individuators’.

BV:  This is not quite right.  Consider the the first-person singular pronoun, 'I.'  This is an indexical expression.  If BV says, 'I am hungry,' he refers to BV; if PL says 'I am hungry,' he refers to PL.  Either way, something is conveyed about the nature of the referent, namely, that it is a person or a self. So what Ed said is false as it stands.  A use of 'I' does tell us something about the individual the sentence containing 'I'  is about.

Examples are easily multiplied.  Apart from the innovations of the Pee Cee, 'she' tells us that the individual referred to is female.  'Here' tells us that the item denoted is a place, typically.  'Now' picks out times.  And there are other examples.

There are no bare items.  Hence there cannot be reference to bare items.  All reference conveys some property of the thing referred to.  But variables may be a counterexample.  Consider 'For any x, x = x.' One could perhaps uses variables in such a way that there is no restriction on what they range over.  But it might be best to stay away from this labyrinth.

One criticism, then, is that there are no bare individuators.  A second is that it is not a singular term, but a use of a singular term that individuates.  Thus 'I' individuates nothing.  It is PL's use of 'I' that picks out PL.  

(3) If a singular term is non-descriptive, its meaning is the individual it individuates. A singular term cannot tell us which individual the proposition is about, unless there exists such an individual.

The course of the book is then to show why we don’t have to be forced into assumption (3).  There doesn’t have to be (or to exist) an individual that is individuated. The theory of non-descriptive singular terms is then developed in the way I suggested in my earlier posts. Consider the inference

Frodo is a hobbit
Frodo has large feet
-------
Some hobbit has large feet

I want to argue that the semantics of ‘Frodo’ is purely inferential. I.e. to understand the meaning of ‘Frodo’ in that argument, it is enough to understand the inference that it generates. That is all.

BV:  'Frodo' doesn't generate anything.  What you want to say is that the meaning of 'Frodo' is exhausted by the inferential role this term plays in the (valid) argument depicted.  Sorry to be such a linguistic prick.

What you are saying is that 'Frodo,' though empty, has a meaning, but this meaning is wholly reducible to the purely syntactical role it plays in the above argument.  (So you are not an eliminativist about the meanings of empty names.)  But if the role is purely syntactical, then the role of 'Frodo' is the same as the role of the arbitrary individual constant 'f' in the following valid schema:

Hf
Lf
-------
(Ex)(Hx & Lx).

But then what distinguishes the meaning of 'Frodo' from that of 'Gandalf'?

Meinongian nonentities are out.  Fregean senses are out.  There are no referents in the cases of empty names.  And yet they have meaning.  So the meaning is purely syntactical.  Well, I don't see how you can squeeze meaning out of bare syntax.  Again, what distinguishes the meaning of the empty names just cited?  The obviously differ in meaning, despite lacking both Sinn and Bedeutung.

We don’t need an object-dependent semantics to explain such inferences, and hence we don’t need object-dependence to explain the semantics of proper names. If an inferential semantics is sufficient, then the Razor tells us it is necesssary: Frustra fit per plura, quod potest fieri per pauciora.

BV:  You should state explicitly that you intend your inferential semantics to hold both for empty and nonempty names.

And now we see the illusion. The proposition

John is thinking of Obama (or Frodo, or whomever)

has a relational form: “—is thinking of –”.  But it does not express a relation. The illusion consists in the way that the relation of the language so strongly suggests a relation in reality. It is the illusion that causes us “to multiply the things principally signified by terms in accordance with the multiplication of the terms”.

That’s the main idea. Obviously a lot of middle terms have been left out.  Have at it.

BV:  So I suppose what you are saying is that belief in the intentionality of thought is as illusory as the belief in the emissive theory of sight.  Just as the the eye does not emit an ethereal something that travels to the moon, e.g., the mind or the 'I' of the mind -- all puns intended! -- does not shoot out a ray of intentionality that gloms onto some Meinongian object, or some Thomistic merely intentional object, or some really existent object. 

You face two main hurdles.  The first I already mentioned.  You have to explain how to squeeze semantics from mere syntax.  The second is that there are theoretical alternatives to the view that intentionality is a relation other than yours.  To mention just one: there are adverbial theories of intentionality that avoid an act-object analysis of mental reference. 

Related articles

The Aporetics of Existence and Self-Identity

Are Propositions Counterexamples to Brentano's Thesis?

Comments

[london ed]

Thanks for these comments Bill. I think you have your finger on the main challenge (and hence the main interest) of the theory.

>>Consider the the first-person singular pronoun, 'I.'

Can we set these aside for now? Indexicals are easily dealt with. (And yes, I should have specified this, my mistake).

>>What you are saying is that 'Frodo,' though empty, has a meaning, but this meaning is wholly reducible to the purely syntactical role it plays in the above argument.

Yes, but the idea of a 'syntactical role' is complex. Consider:

(A) There is a character in LOTR called 'Frodo'. Frodo is a hobbit.
(B) I have a pet called 'Frodo'. Frodo is a dog.

Clearly we can't, and we don't, infer 'some hobbit is a dog'. This is because the first Frodo 'refers back' to 'character in LOTR', whereas the second refers back to 'a pet'. What is this back reference? It can't be to an object existing in reality, because the first example is about something which doesn't exist in reality. Ergo the back reference must reduce to syntax. There must be purely syntactical rules that allow us to generate the inference 'some character in LOTR is a hobbit' from the two sentences in (A), and the inference 'some pet of mine is a dog' from the two sentences in (B).

'Syntax' is slippery of course. My understanding is as follows. If there is a set of rules or an algorithm that would allow a computer, armed with an understanding of English, to crack through the text of LOTR and sort out all the referring expressions into sets corresponding to the different characters, then the semantics of referring expressions reduces to 'syntax'.

Note this is not 'eliminativist' wrt semantics. Clearly names and referring expressions have a meaning. But we can reduce their meaning to syntax, i.e. linguistic rules, perhaps very complex. Demonstratives and other indexicals are an apparent exception which I would like to set to one side for now.

>>But then what distinguishes the meaning of 'Frodo' from that of 'Gandalf'?

Consider

(*) There is a wizard called 'Gandalf', and there is a hobbit called 'Frodo'. Frodo lives in the Shire.

From this we can infer that some hobbit lives in the Shire. If we changed the second 'Frodo' to 'Gandalf', we could not make this inference, although we could make the inference that some wizard lives in the Shire. Changing the name changes the meaning of the whole, ergo the names 'Frodo' and 'Gandalf', in that context, have different meanings. Or rather, their meaning is completely reducible to the inference they generate.

[BV]

Although there is a prima facie distinction between reduction and elimination, you may find it difficult to keep your reduction from collapsing into an elimination. (I have written many posts on this problem.) You don't want to be eliminativist about meaning, but you may end up being eliminativist willy-nilly, i.e., nolens volens.

Further, you may not even be able to get as far as a reduction. If you show that for every mental state there is corresponding brain state, that doesn't show that mental states reduce to brain states. Correlation is not identity. If for every student there is a chair, it doesn't follow that students reduce to chairs. In general, a function that maps each x onto a unique y does not reduce the xs to the ys.

Suppose you show that every name plays a unique syntactic role, that would not suffice to show that each name just is its syntactic role.

And if you did show this, then, by the first point, you would have eliminated meaning.

Meaning is irreducible because meaning comes from mind and mind is irreducible.

[Peter Lupu]

Bill,

I assume that in the phrase "...that would not suffice to show that each name just is its syntactic role." you meant to say "...that would not suffice to show that [the meaning of] each name just is its syntactic role."

[london ed]

>>Meaning is irreducible because meaning comes from mind and mind is irreducible.

That is a labyrinth I would like to avoid. Whether or not meaning is mind-dependent, the question is whether some meanings are 'object dependent'. Some have argued either that the object is trapped in the meaning (as Kaplan jokingly suggests), or that there is some kind of irreducible relation between the meaning and the object, such that, with the object destroyed, the meaning is destroyed too.

Bill, do you agree (1) that in this mini-story

(*) There is a wizard called 'Gandalf', and there is a hobbit called 'Frodo'. Frodo lives in the Shire.

the meaning of the story is changed if we change the name 'Frodo' in the second sentence to 'Gandalf'? And do you agree (2) that, if so, this must be because the names 'Gandalf', and 'Frodo', as used in the story, have different meanings? And (3) do you agree that at least part of the change in meaning is explained by difference inference we make from the story? E.g. with the name 'Frodo', we infer that some hobbit lives in the Shire. With the name 'Gandalf', we infer that some wizard lives in the Shire.

The question is also whether there is any more to the meanings of these empty names than the inferences they generate. Do we also need 'objects' to explain the meaning of 'Frodo' and 'Gandalf' as those names are used in the story above?

[BV]

Peter,
You are exactly right. That is what I meant to say.

[Ed]

>>Suppose you show that every name plays a unique syntactic role, that would not suffice to show that [the meaning of] each name just is its syntactic role.

Now I understand. Let's take that as a topic for another post.

[BV]

Ed,

You think Kaplan was joking? I don't think so. Don't be misled by the word 'trapped.'

Yes I agree with your three points. But you are making a far more radical claim. You are suggesting that there is nothing more to the meaning of 'Frodo' than the inferential role that that name plays in the context of TLR. No one is going to deny that there is some connection between the meaning of this name and the inferential role it plays. The latter is or is close to a Moorean fact. You, however, are making a wild theoretical claim. It is arguably as wild or bold as what Meinongians claim. They say that some things have no being at all! You say that there are meaningful non-logical expressions whose meanings reduce to syntactic roles!

I should like to avoid both of these wild theories.

Fearing the Jungle, you embrace the Razor.

Note that you cannot move directly from the denial of objects to inferential semantics. Why can't a name have sense without reference?

I can understand an inferential semantics of logical connectives/constants such as the ampersand and the wedge. But names are non-logical expressions. What you have said so far gives me no reason to think that the meaning of a name can be reduced to its inferential role.

At the end of the day you may be an eliminativist about meaning, nolens volens.

[Ed]

>>You say that there are meaningful non-logical expressions whose meanings reduce to syntactic roles!

Not quite, but depends on what we mean by 'syntactic'. We agreed, did we not, that to resolve the problem of ambiguity we had to include meaning within the ambit of logical form. We distinguish 'Frodo' (as in LOTR) from 'Frodo' (guy in Ohio).

>>Note that you cannot move directly from the denial of objects to inferential semantics. Why can't a name have sense without reference?

Well surely it can: an object-independent sense. However, I deny that there is any more to that sense than is needed to generate that inference.

Let me work on another post to discuss by email, if that's OK with you.

[london ed]

PS we can’t approach this discussion without at some point looking at Searle’s Chinese Room argument. Searle argues that we cannot reduce semantics to syntax, IMO without being entirely precise about what ‘syntax’ or ‘semantics’ really means. I do a lot of translation work and very often have to refer to a dictionary. So I follow the rule: look in the Latin dictionary for the word I want to translate, and find the English word that corresponds to it. Then I have translated from English to Latin. Few translators are fluent enough to do without a dictionary, yet can’t we say we understand Latin?

[BV]

I think I'll do a post on the Chinese Room. But for now:

It is hard to see what your translation adventures have to do with the Chinese room argument. For one thing, Searle stuck in the room knows no Chinese at all, indeed he cannot even identify the characters as Chinese rather than Japanese, but you know Latin fairly well. You know Latin and you know English, hence your translating, even when you must have recourse to a dictionary, is not a mechanical manipulation of symbols that are (to you) meaningless. Let us say you come across the Latin *cicuta.* From the context you suspect it refers to a poison or something harmful to humans, but you don't know that it means hemlock. So you consult the dictionary. But you did know that the Latin word is feminine, a noun, a Latin word, a word for something harmful, the rough pronunciation of the word , etc. Searle in his room knows none of these sorts of things. It is not essential to his thought experiment that he know even that the symbols he manipulates are characters of Chinese, or even of a natural language as opposed to say Esperanto.

[Peter Lupu]

Ed,

First, you compare refuted scientific theories to philosophical theories. In particular, you compare direct-reference semantics to eyebeam-theory of perception as well as a few other refuted scientific theories. I think this comparison is not apt. Unlike refuted scientific theories which are generally "dead for good", philosophical theories seem to exhibit the exceptional aptitude of *resurrection*: they are down (for a period), but never out. Look at the history of philosophy.

Second, I do not see how you can avoid across the board a semantics that is object-oriented. At some point or other language must link to reality. O/w I do not see how meaning can emerge. Consider the problem of empty-names or the subclass of fictional-names (your primary concern). What is the problem with them? Well, the problem is that unlike names that have a referent in a language as well in most uses, empty/fictional names lack a referent. So the very problem of empty/fictional names arises by comparison to non-empty names which do have a referent. So even in stating the very problem of fictional-names you cannot avoid but comparing them to names that do refer. Hence, word-object relations are unavoidable.

Third, Bill has been harping on the idea that you cannot tease out meaning from syntax or so called "inferential-semantics" beyond what can be done regarding the meanings of logical constants. Consider this problem. Imagine a narrative with only two syntactic expressions: 'Fa'; 'Ga'. You may go on and derive as many inferences from these two expressions until you are blue in the face (there are of course infinitely many, strictly speaking) and you won't know what 'F' and 'G' mean; whether they have the same meaning or different meanings; whether one entails the other; and so on. (This is just the Chinese Room argument all over again [I think]). So this is Bill's challenge to you: How can you get any meaning of the non-logical terms from "inferential semantics" (forget the word 'syntax' in this context) provided inferential-semantics means the traditional inferences sanctioned in Prop. and First-Order Quantification calculi? And you cannot assume here the values true/false as given to you in advance, for these depend upon meaning of the parts of the bearers of true/false.

Lets start with these.