October 2006 Archives

Assume you're on trial for murder (and are innocent) and are on the witness stand. How would you answer the following question:

Isn't it true that you killed your spouse Monday, October 3rd 2006?

I would answer it 'yes', as in, "Yes, it is not true that..." But I take it that this would not be advisable from the Defense's point of view. Why? Doesn't the opposite answer demonstrate a lack of appreciation of the embedded negation? Clearly BOTH answers are unacceptable.

Or are they? If bivalence doesn't hold, then answering 'no' might not commit oneself to an admission of guilt. But, because I'm a classicist, I'm going to keep bivalence and stick to my 'yes'.

Note: My wife is just fine.

Taking as my starting point Lynch's "A Functionalist Theory of Truth" (in The Nature of Truth, ed. Lynch, MIT: 2001, pp. 723-750). I want to bring up something I take to be a positive consequence for this view when it is conjoined to the view that to know the meaning of a declarative sentence is to know its truth conditions. First, let's take a gander at functionalism regarding truth.

To be a functionalist regarding truth is to hold that truth is whatever property plays a certain functional role in the discourse in question. The functional role that truth plays will be demarcated by a set of (a priori?) platitudes concerning truth, such as "The proposition that p is true if and only if p," and "every proposition has a negation," and so on. However, functionalism regarding truth is neutral with respect to the metaphysics of what particular property realizes truth's functional role in a given discourse. Some discourses, say those about medium sized objects, might require a robustly referential realization of truth, while others, say discourse about the law, might require some other kind of property the realization of which is the property of truth in that discourse. Lastly, the functionalist, for Lynch (in agreement, he says, with Wright), sees the platitudes as constituting our grasp of the concept of truth.

My reflections on this begin with the Davidsonian assumption that understanding a language consists in understanding its truth contitions as embodied in a finite theory of truth for the language in question. Now, it seems that these two positions, functionalism about truth and Davidson's thesis, are quite compatible. A restriction of a theory of truth for a language L is that it entail all and only the true T-sentences of the form: S is true if and only if P (where s is a description of an object language sentence and P correctly translates S). But T-sentences are a member of the set of platitudes of which Lynch speaks that specify the functional role of our truth concept. Adding the remaining platitudes (see p. 730 of Lynch) can then turn a theory of truth for L into a functionally respectible theory of truth for L.

A modification to Lynch's alethic functionalism is called for at this point. Truth is defined in terms of satisfaction on a Tarskian construal. Hence, if the functionalists are to be right, then we must also include in our platitudes about truth certain platitudes about satisfaction (or reference). This will give us a functionalist account of reference/satisfaction (in itself not a bad thing, let's say). Presumably this would be done by specifying a set of platitudes about reference. But is there a consistent set of platitudes about reference that hold in all domains, the realization of which might be different from domain to domain. Here I will give some possible platitudes concerning reference.

Reference Platitude 1 (R1) : Reference is a relation between words and the world.
R2: When in a context C, I utter a referring expression x as part of a sentence S of a language L, then usually I have referred to whatever it is that is the referent of x.
R3: When uttering a declarative sentence S in a context C, if an expression I use fails to refer then I have said something false. [Taking Russell's side for the moment.]

Of course, 'the world' in R1 can be construed broadly enough so that in its extension we can include words, some ideas, and all sorts of things. So maybe this will work. Then we can regard a property P as realizing the concept of reference if it plays the functional role spelled out by the platitude(s). Might this work? Here are the positives: (1) it doesn't require reworking Davidson's program; and (2) it will require rejecting his primitivism about truth. Furthermore, (3) we can keep his indeterminacy about reference. And lastly, (4) because the meaning of 'true' is fixed by the platitudes, and not by the particular property realizing it in a given domain, the truth conditions for sentences in one domain and those for another domain can be given in the same metalanguage without fear of equivocation. Thoughts?

The Southern Society for Philosophy and Psychology has announced a CFP for its 2007 Conference, to be held April 5-7, 2007 in Atlanta, GA. This year's President's Invited Speaker is John Searle (UC, Berkeley). Also invited are Colin McGinn (Rutgers) and David Rosenthal (CUNY). Also of interest are two symposia, one on Normative Naturalized Epistemology and the other on Realization. For more information visit their site.

The University of Otago Philosophy Department (Dunedin, New Zealand) is hosting a conference titled "Truth and Reality" (January 8-12, 2007). For a list of speakers and other conference information, go here.

The International Conference on Linguistics and Epistemology (Saturday/Sunday, May 12/13, 2007, Scotland, UK) has issued a CFP. Details here. Also, graduate students might notice the following: "Support to help graduate students defray conference expenses might be available."

A deflationist has two jobs to do. First, they must explain why it is we need a notion of truth. This they usually attribute to our need of certain expressive resources. Second, they must show that the deflationary notion is all we need to make sense of most uses of 'true'. Many examples we are familiar with already. We don't need a notion of truth to say "Snow is white." Saying '"snow is white" is true' is redundant. And what is going on in cases like "Everything John said was true" can be said to be something like an abbreviated form of a really large conjunction. So, how might a deflatoinist handle the following:

(S) Sue hid the truth from Bob.

Maybe (S) can be symbolized without 'true' as follows:

(S1) There is an x such that (Hsxb & x)

The existential quantification here must be read substitutionally or S1 is incoherent. Will this work as it stands? Just food for thought.

Ok, remember your SAT's? I think fascism is to truth as ethical egoism is to duty. Fascists can adopt some reductive account of truth that reduces truth to some power relation, but they need to present themselves as believing in a more robust notion (to give their position gravitas). This is just like a good egoist. They can be egoists, but they should tell everyone and try to convince everyone that there are moral absolutes and duties. This way the egoists can better take advantage of them. (See also my previous entry on fascism)

For those interested, Paul Boghossian (NYU) will be delivering a lecture, Friday, November 3 in UK's Student Center 228 at 4pm (on metaphysics and epistemology, specific topic TBA). I'm looking forward to it as the philosophy reading group has just finished reading his Fear of Knowledge (OUP, 2006). Comments on the talk to follow.

Hello, I'm Michael Horton, a Ph.D. candidate in the University of Kentucky's Department of Philosophy, working on my dissertation under the direction of Brandon Look. Currently I'm located at the University of South Alabama, Mobile, where I'm teaching PHL 251: Philosophy and Cognitive Science as well as some introductory level courses.  I have a personal philosophy website with student materials and professional materials here. 

My dissertation is on truth and meaning, and, specifically, whether one can employ a deflationary notion of truth while doing (some version of) truth-theoretic semantics. My conclusion is negative: I think one needs some inflationary notion to do semantics. Feel free to write with suggestions and other ideas for improvement of the site. Best, MH P.S. A special thanks to M. Mullins for helping get this site started. Make sure to check out his blog The Prosblogion. P.P.S. If anyone is interested, a CV can be sent upon request. 

[Updated 13 April 2009]

In note 23 to Chapter 8 of Popper's Open Society, Popper says that Russell "has shown the connection between a relativist theory of truth and the creed of fascism" (Princeton U.P., 1966). So, I haven't read the Russell, but here is a Popperian-inspired argument for this conclusion. Tarski made truth talk respectable. Truth talk within science is o.k. provided we employ Tarski's predicates (or something similar). The "truth" of the relativist is not Tarski's "truth". According to Popper at least, Tarski's truth is absolutist and correspondence (see Open Society, p. 273, note 23.). Hence, relativism's "truth" is unscientific. Fascism is unscientific insofar as it exhibits historicist tendencies--viewing history as exhibiting lawlike behavior such that if one knew the laws governing history one could predict history's culmination. Relativism's being incompatible with Tarski but compatible with fascism justifies our rejecting it.
But, many (Horwich, Field) argue that we can get by with a deflationary conception of truth in science. We just need a certain logical device to enable us to make blind ascriptions, convey our assent to a large (potentially infinite) set of sentences, and so on. So, if this is the bare minimum we need, then it looks like we cannot rule out fascism using only our theory of truth and our theory of science. But this is what we'd want. Our theory of truth may inform other areas, such as meaning, but it seems improbable that our theory of truth would rule out a political view such as fascism. Fascism stands or falls on its own. But, and here is where I'll end it, does fascism require a certain notion of truth? Can one be both a fascist and an absolutist or a realist or a correspondence theorist? I see no reason why not.

[B.T.W., most of my posts will not be political, but I've had this on my brain lately as I'm teaching the Popper to my students.]

I draw my inspiration for this from a paper I heard delivered by Claire Horisk at the 2005 CSPA. In that paper, a quite good paper, she argued that if an interpretation of 'is true' turns the T-sentences into necessary truths, then we can reject that interpretation. Brandom's 'is true', the anaphoric theory of 'is true', does just that. Q.E.D. While I want a robust notion of truth to figure in theories of truth/meaning, I fear that Brandom can circumvent this objection. So here is Horisk's argument and a response on behalf of Brandom. My question will be whether there is a way around the response.

Paraphrase of her argument. On Brandom's anaphoric theory of 'is true' (I don't call it a theory of truth because it is really not), 'is true' is a prosentence-forming operator. As such, it inherits its content from its anaphoric antecedent. Hence, '"Snow is white" is true' has the same content as 'Snow is white'. Therefore, a T-sentence like

(S) 'Snow is white' is true if and only if Snow is white

is a necessary truth. In fact, the left and right hand sides are synonymous. But according to most versions of Davidsonian semantics, T-sentences are contingent (though perhaps lawlike). Hence, the anaphoric theory of 'is true' cannot be used by a Davidsonian semanticist. [The foregoing argument paraphrased from "Kolbel and Williams on Skim Semantics," Claire Horisk, presented (in absentia) at the 2005 CSPA in Lexington, KY).

My response is as follows: rather than view 'is true' as purely an anaphoric sentence-forming operator, let's also view the result of applying it to a sentence as a sentence in a different language--the metalanguage. Recall also that T-sentences on a Tarskian construal are sentences in the metalanguage employing a structural-descriptive name of the object language sentence of which truth is being predicated. Similarly for the anaphoric theory we can regard T-sentences as metalinguistic constructions using a structural-descriptive name of a sentence to which the anaphoric operator has been applied. I think this will solve our problem (so long as we make sure we're not talking about propositions). Consider the following T-sentence:

(R) 'Il neige' is true if and only if It is snowing.

Clearly this is not a necessary truth because nothing in it guarantees that 'Il neige' has the same truth-conditions as 'it is snowing'. Simply undoing the pro-sentence, so to speak, won't help as it (seems to) will in the homophonic case below:

(R*) 'It is snowing' is true if and only if it is snowing.

Because from the theoretical perspective (R) is like (R*), we shouldn't be misled in the heterophonic case either. Neither (R) nor (R*) are necessary truths. Going metalinguistic is what helped us out of the bind. Is this enough? Are propositions a problem if they're the truth bearers?