Philosophy 257 Behavior and Other Minds

Asst. Prof James Pryor Dept. of Philosophy

Truth, Metaphysics, and Understanding

Truth and Metaphysics

Principle 1 Don't automatically assume that debates about realism have much to do with debates about the analysis of truth. These debates might turn out to be wholly independent.

To see this, it helps to consider a deflationist theory of propositional truth.

Different Versions of Deflationism about Truth

Redundancy and Performative Theories

These theories say that "is true" doesn't express a property, and that "X is true" doesn't assert anything about X.

The problem both theories face is to explain why inferences of the following form are valid:

1. Oscar's claim is true.
2. Oscar's claim = the proposition that snow is white.
3. So, the proposition that snow is white is true.

Disquotational Theories (Tarski, Quine, Field)

Any deflationary theory of sentential truth makes "'Snow is white' is true (in English)" analytically equivalent to "Snow is white." Hence "'Snow is white' is true (in English) iff snow is white" comes out a simple tautology, and can carry no information about the semantic properties that the sentence 'Snow is white' has in English. In addition, the sentence (i) "If English speakers had used the word 'snow' to refer to grass, then 'Snow is white' would not have been true in English" comes out equivalent to the false claim (ii) "If English speakers had used the word 'snow' to refer to grass, then it would not have been the case that snow is white."

Insofar as one thinks that "'Snow is white' is true (in English) iff snow is white" expresses a contingent, a posteriori fact about the semantic properties of the expression 'Snow is white', and insofar as one thinks that (i) expresses a truth, one should reject deflationary accounts of sentential truth. This is quite compatible with continuing to accept some kind of deflationary account of propositional truth. One just says that it's a substantial, empirical matter what proposition a given sentence expresses.

Horwich's Minimalism about Propositional Truth

Horwich says that the collection of all instances of the schema:

(iii) the proposition that p is true iff p

together constitute a complete analysis of propositional truth. This seems along the right lines. Certainly an adequate analysis of propositional truth should tell us as much as Horwich's minimal theory does. And, as we'll discuss below, an adequate analysis of propositional truth ought to be just as neutral about substantial metaphysical issues like the question of realism as Horwich's theory is. However, Horwich's theory does seem to tell us too little about truth.

1. A complete account of truth should explain pathological and paradoxical uses of "true," as in the sentences "This sentence is true" and "This sentence is not true."
2. Horwich's theory is just a collection of instances. Knowledge of all those facts wouldn't yet provide knowledge of the generalizations of which they are instances. Yet we think a complete analysis of truth ought to provide knowledge of some such general facts.

Some future and better deflationist theory of propositional truth?

We'll have to wait and see.

For a very clear discussion of deflationist theories of truth, see Soames' "The Truth about Deflationism."

Theories of truth can be and should be neutral about issues like:

1. The existence of external reality: for example, a theory of truth shouldn't say that there are mind-independent facts of the matter that correspond to, or in virtue of which, propositions are true. A theory of truth should leave it an open question whether there are any mind-independent facts at all.
2. Whether certain facts are epistemically constrained: a theory of truth shouldn't by itself tell us whether it's possible for a given proposition to obtain undetectably, or for the possible evidence for or against the proposition to fall short of determining whether it's true.

It's best to see coherence theories of truth, epistemic theories of truth, theories which say that truth consists in correspondence to some external reality, and the like, as presenting substantive metaphysical claims as if they were part of an analysis of truth. (They're not.)

Theories of Meaning and Theories of Understanding

1. Formal semantics

   This project aims to tell us what forms of inference are truth-preserving.

2. Semantics for natural language

   This project aims at giving an account of what the sentences of the language strictly and literally say. Among other things, that will tell us what the sentences' truth-conditions are. It may also involve explaining how what the sentences say depends on the conditions in which they're uttered (as with sentences containing indexicals). Call what the sentence says its meaning or content. (With indexical sentences, we'd want to distinguish between the sentence's standing meaning, and what it says on particular occasions of use. But for present purposes, just understand "meaning" to refer to the second.)

3. Explaining speakers' understanding of a language, or their competence with the language

It's easy to get misled here, because the sentence "To understand a sentence is to know what it means" is ambiguous. There's a sense in which "knowing what sentence S means" just means "understanding S." In this sense, it's a harmless tautology to claim that to understand a sentence is to know what it means. But taken literally, "to know what sentence S means" describes having a certain sort of propositional knowledge: knowing of sentence S that it has such-and-such a meaning or content. The claim that understanding consists in any such propositional knowledge is controversial, and we should not accept it without argument.

So, we've distinguished some semantic issues from a psychological issue:

Semantic issues

What is the sentence's meaning? What truth-conditions does it have?

Psychological issue

What psychological states constitute the speaker's ability to use and understand the sentence?

Principle 2 A theory of meaning for a language won't necessarily be a theory of anything knowledge of which constitutes or could explain speakers' ability to speak and understand the language.

Different views about the nature of linguistic competence

0. In general, when you have an ability to X, we can speak of your "knowing how" to X, where this is just another way of saying that you have the ability. It's not meant to offer any psychological explanation of the ability.
1. In some cases, your ability to X might be explicitly guided by a certain body of propositional knowledge that you possess.

3. Closer to (ii) is the case where your exercise of the ability is guided by a certain body of propositional knowledge that you have, but this knowledge is not explicit.
4. Closer to (i) is the case where you don't actually have or employ any propositional knowledge in the exercise of the ability, but we can model your ability in terms of a certain body of knowledge, the possession of which would enable one to perform the ability.

Other views, like the following ones, maintain that linguistic competence consists in, or can be explained by, a sort of propositional knowledge. This would correspond to position (iii). (Variants of these views which corresponded to position (iv) could also be offered.) These views differ over what sort of propositional knowledge is involved:

The pragmatic knowledge account

You know what makes it pragmatically appropriate or inappropriate to assert the sentence. You know what the sentence can be used to "get across." This knowledge is what makes you competent.

The epistemic knowledge account

You know what counts as justification for and against asserting or accepting the sentence. This knowledge is what makes you competent. (Compare the criteria theorist's claims about what's required for competence.)

The semantic knowledge account

You "grasp" certain semantic properties of the sentence (e.g, what proposition it expresses, or in which conditions it's true), and you know that the sentence has these semantic properties, and this knowledge is what makes you competent.

Arguments against the semantic knowledge account of competence: You can be a competent user of "cat," and understand "cat," without having mastered any semantic concepts (like truth or meaning or reference) and so without having any beliefs, even tacit beliefs, about the semantic properties of "cat." You can be a competent user of "cat," and understand "cat," without yet knowing what a language is, or what a linguistic expression is. Hence you can be a competent user of "cat" without yet being able to have any beliefs about the linguistic expression "cat," semantic or otherwise. Knowledge of S's truth-conditions does not by itself put one in a position to understand S. For instance, one might know that S is true iff snow is white, but mistakenly take S to express the proposition that snow is white and arithmetic is incomplete (see Foster, "Meaning and truth theory"). Recall that we need a substantive notion of sentential truth for claims about S's truth-conditions to communicate any information about S's semantic properties. The present point is that, even when we are making claims about S's substantive truth-conditions, information of that sort does not by itself fully determine what S means. It's even unclear whether knowledge of S that it expresses a certain proposition suffices for understanding S (see Soames, "Semantics and Semantics Competence," esp. pp. 584-7). Even when competence with S coincides with knowledge of S's semantic properties, the competence needn't consist in the semantic knowledge. In some cases, you're not in a position to grasp the proposition a sentence expresses except by virtue of your ability to use that sentence to express it. In such cases, it's your competence with the sentence that explains your ability to grasp the relevant proposition, rather than the other way around (see Soames, "Semantics and Semantic Competence," esp. pp. 587-91).

Now, consider the following claims:

U1. Anyone who fully understands and is competent in the use of sentence S will be capable of entertaining the proposition that p (where p is the proposition that S expresses).	This is somewhat plausible, though it may not be true for every sentence (see Donnellan, "The Contingent A Priori and Rigid Designators").
U2. Anyone who fully understands and is competent in the use of S will be in a position to know of S that it expresses the proposition that p, just by appeal to his understanding of S and perhaps some a priori philosophical reasoning.	This is also plausible.
U3. Anyone who fully understands S and is competent in the use of S is required to know of S that it expresses the proposition that p.	This is implausible. As we said above, understanding a sentence doesn't require having semantic concepts or beliefs about language.
U4. Knowledge of S that it expresses the proposition that p is what constitutes or explains competence in the use of S.	This is what the semantic knowledge account of understanding claims. This claim is also implausible.

Principle 3 An argument against the semantic knowledge account of our understanding of certain sentences doesn't by itself constitute an argument against a realist account of what those sentences mean.