Use Theory - PowerPoint Presentation

Slide1

Use TheorySlide2

Use theory: history & motivationSlide3

The Causal-Historical Theory

Last time we learned about the causal-historical theory of reference. According to Kripke’s model, names and natural kind terms have their referenced initially fixed by

ostension

(pointing) or description, and then those referents are passed, as links in a chain, along with the words as they move from speaker to speaker.Slide4

Example

A child is born.

It’s parents point to it and say: let this child be named ‘Richard Feynman.’

Other people learn the name ‘Richard Feynman’ from the parents, and in doing so their name inherits the original meaning.

People far away in the “chain” may know nothing, or believe a bunch of false things, and still succeed in referring to Feynman.Slide5

Causal Theories

We didn’t have time to look at other causal theories of reference/ meaning. The general motivation, though, was that causal interaction with the referent was far more determinate than mere description (we looked at mirror universes, eternal recurrence universes, and Twin Earths).Slide6

Causal Isolation

However, it’s widely recognized that causation can’t be essential to all meaning, because some things that are meant can’t be causes or effects. Consider words like ‘and,’ ‘or,’ and ‘not.’ Conjunction (the meaning of ‘and’) can’t cause or be caused by anything. There’s nothing to point to and say “let that be the meaning of ‘and.’” Or, there may be such things, but none of them are in fact the meaning of ‘and.’ Slide7

Use to the Rescue

However, people who have mastered the meaning of ‘and’ are inclined to

use

the word ‘and’ in the following ways:

If they believe ‘A and B’

Then they would be willing to believe ‘A’

And they would be willing to believe ‘B’Slide8

Use to the Rescue

However, people who have mastered the meaning of ‘and’ are inclined to

use

the word ‘and’ in the following ways:

If they believe ‘A’ and they believe ‘B’

Then they would be willing to believe ‘A and B’Slide9

Suggestion

So maybe ‘and’ means what it does because of how people use it in inference. If you didn’t use ‘and’ in those ways, you wouldn’t mean what everyone else means by ‘and,’ and if you use ‘or’ in those ways, then by ‘or’ you mean what everyone else means by ‘and.’ The meaning of these words is determined by and perhaps identical to the use of these words.Slide10

Further Suggestion

And maybe, just maybe, we were wrong to become causal theorists in the first place. Maybe the meaning of ‘Richard Feynman’ and the meaning of ‘water’ is also how we use those words.Slide11

Careful!

But be careful. It’s not enough to say that the meaning of the words is “determined by how they’re used.” That’s in a way accepted by everyone. According to a causal theorist, the meaning of ‘water’ is determined by the fact that your uses of the word ‘water’ are caused by a certain substance (namely, water). A real “use theory” doesn’t say use merely plays a role in meaning– it says that use

is

meaning!Slide12

The Denial of Denotation

One of the big reasons people have had for adopting use theories is that they have come to deny that words (or all words, or many words) have denotations. They don’t think names refer to things, or that common nouns and verbs apply to things, or that sentences can be true or false.Slide13

Denotation Relations

Why do I connect these ideas: refer to, apply to, and truth/ falsity? Because truth/ falsity can be defined in terms of the former:

A sentence “Michael is hungry” is true := “hungry” applies to the referent of “Michael.” Slide14

Denotation Difficulties

Why would anyone want to give up on these relations? Usually, it’s out of an endless parade of historical failures in accounting for denotation.

The idea theory can’t explain why ‘dog’ applies to dogs, because resemblance is indeterminate. Many non-dogs resemble the idea associated with ‘dog.’Slide15

Denotation Difficulties

The verification theory won’t work, for similar reasons. Many non-dogs (e.g. fake dogs) confirm ‘dog’ more than some dogs do (e.g. abnormal dogs).

And the causal theory won’t work, for similar reasons. Dogs often cause me to say ‘dog’ or think DOG. But so do fake dogs, and marsupial “dogs” and pictures of dogs, and so on. Slide16

The Denial of Connotation

The use theory thus denies that denotations even exist. But it does not thus identify meanings with any of the classical connotations. Remember that ideas (mental images) and verification conditions (possible experiences) were posited as meanings (connotations) solely to explain why words had the denotations that they did. If you deny the existence of denotations, why do you think mental images are meanings? What’s special about

them

?Slide17

The Middle Way

Instead, the use theorist maintains that meaning is non-mental (not connotation). It’s out there in the world. But it’s not the stuff out there in the world we think of as denotation either. ‘Michael’ doesn’t, for instance, mean

me

. The meaning of an expression = how it is used. Sure, use is out there in the world. But the (relevant) use of ‘Michael’ need not involve

me

at all.Slide18

Horwich and the use theorySlide19

Meanings are Concepts

Horwich’s first thesis is that meanings are concepts. “Concepts” are what psychologists and philosophers turned to after the whole idea theory didn’t work out. Concepts are mental entities, but they aren’t little pictures in the mind. Horwich, influenced by the Computational Theory of Mind, takes them to be expressions in the “language of thought” a.k.a. “Mentalese.”Slide20

Metasemantics

Remember that a theory of meaning (as we’ve been using that expression) is not a theory that tells you what meanings are (though often it does that as well)– it’s one that tells you why words have the meanings they do, rather than different meanings, or no meanings at all. (This is often called “metasemantics.”)Slide21

Metasemantics

So what’s Horwich’s story of how words get their meanings (why do they mean the concepts they do, rather than other concepts or none at all?)?

To understand this, we’ll have to look at Grice’s distinction between natural and non-natural meaning.Slide22

Natural Meaning

One meaning of the word ‘meaning’ is indication:

Smoke means (indicates the presence of) fire.

These

Koplik’s

spots mean your child has measles.

The fact that there’s 27 rings on this tree stump means that the tree was 27 years old when it was cut down.Slide23

Features of Natural Meaning

We can’t say “these spots mean the child has measles, but the child doesn’t have measles.”

We can’t say “these spots mean ‘the child has measles.’”

It can’t true that someone means the child has measles by these spots.Slide24

Non-Natural Meaning

We

can

say “John’s utterance ‘

l’enfant

a la

rougeole

’ means the child has measles, but the child doesn’t have measles.

We

can

say “This sentence (‘

l’enfant

a la

rougeole

’) means ‘the child has measles.’”

It

can

be true that someone means the child has measles by “

l’enfant

a la

rougeole

.”Slide25

‘Meaning’ is Ambiguous

Grice thus concludes that there are two English verbs ‘to mean.’ One just expresses natural meaning, roughly: “A means B = Whenever A is true, it’s a fact of nature that B is true as well.” The other is non-natural meaning, and it’s what we’re trying to analyze when we do metasemantic theorizing.Slide26

The Univocality

of Meaning

Horwich, however, claims that there’s only one sense of ‘meaning,’ the natural one. The way he understands natural meaning is: ‘smoke means fire = smoke gives us a good reason to believe there’s fire.’ So he says ‘cat’ means the concept CAT = (utterances of) ‘cat’ give us a good reason to believe there’s (in the speaker’s mind) CAT.Slide27

Virtue?

“It is a virtue of this account that it respects the relational appearance of meaning attributions and that it calls for no special, ad hoc assumption about the meaning of ‘means’ in semantic contexts.”

Horwich, in his ‘ad hoc’ remark, seems to forget that there were principled reasons for denying the

univocality

of ‘meaning.’Slide28

Natural Meaning is Transitive

Furthermore, natural meaning is transitive:

Thunder means there’s lightning.

Lightning means there’s unbalanced electric charges in the clouds.

Therefore, thunder means there’s unbalanced electric charges in the clouds.Slide29

Non-Natural Meaning is Not Transitive

If all meaning were natural meaning we’d expect:

‘salt’ means there’s SALT

SALT means there’s PEPPER

Therefore ‘salt’ means PEPPERSlide30

Principle 2

Principle 2: “The overall use of each word stems from its possession of a basic acceptance property.”Slide31

The Robustness of Use

The point of this principle is to overcome a problem (a large problem that exists for lots of theories besides the use theory). Here’s the version for the UT: often, we use words in ways that are not consistent with their meaning. We flub our speech; we make a genuine mistake (and call a cow a ‘horse’); we use words metaphorically; we overstate or understate…Slide32

The Problem of “Error”

If meaning is to be identified with use, then it would seem that these uses, since they are uses, must be part of the meaning. So flubs, mistakes, metaphors, hyperboles, etc. are all literally true. But that’s silly.Slide33

Horwich’s Response

So Principle 2 is Horwich’s response: there is some sort of basic regularity that explains all of the use (including correct use, incorrect use, and poetic use). The regularity that explains all the use is the meaning (we’ll see what that amounts to shortly, and how it’s supposed to be consistent with the fact that concepts are meanings). So erroneous uses, while explained by the basic regularity, are not constitutive of meaning (only the basic regularity is). Slide34

Basic Acceptance Properties

The acceptance property that governs a speakers’ use of “and” is (roughly) his tendency to accept “p and q” if and only if he accepts both “p” and “q.”

The explanatorily fundamental acceptance property underlying our use of “red” is (roughly) the disposition to apply “red” to an observed surface when and only when it is clearly red.Slide35

Basic Acceptance Properties

(c) The acceptance property governing our total use of the word “true” is the inclination to accept instances of the schema ‘the proposition that p is true if and only if p.’Slide36

Principle 3

“Two words express the same concept in virtue of having the same basic acceptance property.”

In Principle 1, Horwich said that meanings are concepts. In Principle 3, he says that concepts are

individuated by

basic acceptance properties of the words that mean them.Slide37

Individuation

Consider the word ‘gift’ in English: it means something like “a present, something of value given without charge.” Now consider the word ‘gift’ in German: it means something like “a poison, venom, or toxin.” Are these one word with two meanings, or two words? The answer to this question is not important for us. What is, is this: if words are individuated by their spelling/ pronunciation, we have one word; if they’re individuated by their meaning, two.Slide38

Horwich on Individuation

For Horwich, concepts are individuated by basic acceptance properties of the words that express them. Let’s call these their “meanings.” Then two concepts have the same meaning = the words expressing them have the same acceptance properties. This is how meaning for Horwich is both concept and use. The way you tell one concept for another is the use of the words that express it.Slide39

Summary of Principles

Words mean concepts, and “meaning” is univocal– it always means just “indication.”

For any word, all of its uses may be explained by a basic acceptance property: a regularity in the use of the word, that explains irregular uses as well.

Concepts are individuated by the basic acceptance properties of the words that express them.Slide40

PRO Argument 2: Explanation

Premise: “What people say is due, in part, to what they mean.”

Premise: “It is relatively unclear how any other sort of property of a word [besides use properties] would constrain its overall use.”

Conclusion: Only the use theory can explain how what people say is due to what they mean.Slide41

Premise 2?

I’m skeptical of premise 2 in this argument. Horwich says that what a word refers to can’t explain its use. Imagine I have a map of Central and on one part of it is written “Wing

Lok

Street.” Why did the mapmaker

use

that name there? Quite sensibly,

because

the street drawn on the map corresponds to Wing

Lok

Street, and

“Wing

Lok

Street” refers to Wing

Lok

Street

. How does a basic acceptance property provide a better explanation than that?Slide42

PRO Argument 3: Attribution

When we judge that two words (in different languages or idiolects) mean the same thing, we check to see if their

uses

are appropriately similar. And what does ‘appropriate’ mean here? Horwich argues that it means differences in use are circumstantial– both words are still governed by the same basic acceptance property. He says we judge they mean differently when differences in use are more than merely circumstantial.Slide43

Theoretical Entities Redux

This is certainly an empirical question. It does run Horwich into some potential trouble though (CON Argument 2: Holism). People with radically different theories (about electrons or whatever) will use words in radically different ways. Horwich can say that they are still talking about the same thing but only up until the point that their uses are governed by the same basic acceptance property. Again, Whether this comports with intuition is an empirical matter.Slide44

PRO Argument 4

Premise 1: We are generally inclined to accept inferences from a sentence S containing word w, S(w), to the sentence S(v), when w and v are synonyms (have the same meaning).Slide45

PRO Argument 4

Premise 2: If the use theory is true, then w and v are synonyms whenever w and v’s uses are governed by the same basic acceptance property. Thus if w’s basic acceptance property leads me to accept S(w), v’s basic acceptance property, which is the same as w’s, will likewise lead me to accept S(v)Slide46

PRO Argument 4

Inference to the best explanation: Since no other theory of meaning explains these facts better than the use theory, the use theory is true.Slide47

Against Application as a ToM

For example, Horwich argues that if the meaning of ‘groundhog’ is what it applies to, then to know the meaning is to know what it applies to. And to know the meaning of ‘woodchuck’ is to know what it applies to. But, he claims, you can know all this without knowing that ‘groundhog’ and ‘woodchuck’ apply to all the same things.Slide48

In Defense of Denotation

Is that really true though? Many philosophers have held that the meaning of a sentence is its truth-conditions (and remember: truth is a notion belonging to the denotation relations). To know what a sentence means is to know the circumstances under which it is true. If S(w) and S(v) are true under the same circumstances, then shouldn’t we know that S(w) if and only if S(v), when we know their meanings?Slide49

In Defense of Horwich

Well… not exactly. There are classic examples where sentences are true under the same circumstances, but not known to be so by people who understand them:

2 + 2 = 4 if and only if Obama is president.

e

i

π

+ 1 = 0 if and only if Obama is president.Slide50

PRO Argument 5: Implicit Definition

An implicit definition is where we define a word or symbol by using the defined symbol in a context. Here’s an example:Slide51

Euclid’s Postulates (Wolfram Alpha)

1. A straight line segment can be drawn joining any two points.

2. Any straight line segment can be extended indefinitely in a straight line.

3. Given any straight line segment, a circle can be drawn having the segment as radius and one endpoint as center.

4. All right angles are congruent.

5. Given

any straight line and a point not on it, there

exists

one and only one straight line which

passes through

that point and never intersects the first line, no matter how far they are extended.Slide52

PRO Argument 5: Implicit Definition

Horwich argues that the use theory is needed to make sense of implicit definition. When people are given a set of axioms or postulates involving new terms, they accept them and use those postulates to decide what other sentences involving those terms to accept. Thus the implicitly defining postulates wind up being the basic acceptance properties governing future use.Slide53

Implicit Definition?

This argument rests quite a bit on the possibility of implicit definition. There’s some reason to think things don’t work this way. For example, in non-Euclidean geometry, lines don’t satisfy Euclid’s postulates (that’s why it’s called

non

-Euclidean geometry). But that doesn’t make sense if Horwich is right: the things in non-Euclidean geometry aren’t

lines

.Slide54

PRO Argument 6: Translation

Why is it that when I say, “I’d like some cheese” in America and “Je

voudrais

du

fromage

” in France, similar things happen in both countries?

Here’s Horwich’s idea. I have this theory:

If I say “I’d like some _____” in America, peons bring me some _____. Slide55

Further Theory

In addition, I have this theory:

If I say, “I’d like ----- _____” in America then peons bring me ----- _____.

For example, If I say “I’d like ALL cheese,” then peons bring me ALL cheese.Slide56

Further Further

Theory

In addition I have this theory:

If I say “

xxxxx

would like ----- _____,” in America then peons bring

xxxxx

----- _____.

For example, if I say “Tony Parker would like no beans,” then peons bring Tony Parker no beans.Slide57

Similar Role for French

But then notice that ‘

voudrais

’ plays a similar role:

If I say “

xxxxx

voudrais

/

voudrait

/ etc. ------ _____” in France, then peons bring

xxxxx

----- _____.” Slide58

Basic Acceptance Property

It’s a simple step here. Horwich claims that the basic acceptance property underlying our uses of ‘would like’ and ‘

voudrais

/t/etc.’ And this is it:

All uses of w (‘would like,’ or ‘

voudrais

’) arise from the fact that we accept that if we say “

xxxx

w ----- _____” then peons bring

xxxxx

----- _____.”Slide59

Why Translation Works

Therefore, identical basic acceptance properties between words in different languages give rise to identical behaviors (or at least, expectations of behaviors) across those languages.

Translation works, Horwich says, because meaning is constituted by basic acceptance properties. Slide60

Other Possibilities?

Horwich doesn’t claim, however, that a denotation-involving theory couldn’t arrive at an explanation of why translation works. For example, for commands, we might think that instead of truth conditions (circumstances under which they are true), they had satisfaction conditions (circumstances under which they are obeyed) as their meanings.Slide61

Alternative Explanation

Then we might say that

In any country, peons satisfy the conditions of your commands (when they speak the language you utter them in).

“I’d like some cheese” and “Je

voudrais

du

fromage

” have the same satisfaction conditions.

Peons will bring me cheese in France when I say “Je

voudrais

du

fromage

.”Slide62

PRO Argument 7: Pragmatic Argument

Horwich’s final argument is that since his theory explains why translation works, it explains why we bother translating things.

I’m not sure this gets to count as an

extra

reason for accepting the theory.Slide63

concluding thoughtsSlide64

Horwich is not the only use theorist, but he’s one of the few that I understand. His views are put forth in admirable clarity.

Here’s a summary of the arguments, color-coded for whether I think

they work

,

don’t work

, or are still

up for grabs

.Slide65

Rainbows

There’s only one sense of ‘meaning.’

UT required for meaning to explain use.

Appropriate similarity in use = same meaning

UT required for synonym equivalence.

UT required for implicit definitions to work.

UT required for efficacy of translation.

UT explains purpose of translation.