Friday, August 10, 2007

Some Thoughts on Theories of Truth

In studying philosophy, one of the more perplexing things one finds (if you’re anything like me) is that there are a multiplicity of “theories of truth”. How can there be any room for disagreement here? In examining different “theories of truth”, one might think that such theories are beset with a problem similar to that of the “paradox of analysis”. The paradox of analysis, to recall, is that a purported “analysis” of an expression e is either synonymous with e or not. If it is the analysis is unenlightening, and if it is not the analysis cannot be correct. Similarly, one might think, if there really are rival “theories of truth” we can evaluate a given theory of truth according to its own account of truth or according to a rival one: If its own, it will trivially be vindicated (assuming it is consistent), if another, it will be trivially refuted. Now, if a “theory of truth” is merely a proposed definition of the word ‘truth’ or the predicate ‘is true’ there is no problem, for people are free to define their terms as they please. But this is not an adequate account of what goes on in debates over such theories (at least, it is not an adequate account of what the various disputants think is going on.) Firstly, if the disputants thought of themselves as offering (stipulative) definitions, their arguments would center mainly on which definition was the most pragmatically useful (and for which purposes). In general, however, they do not. Instead, most of the parties to these debates seem to view their opponents as offering genuinely incompatible accounts. Secondly, the accounts which are offered are called theories, and this implies that the accounts all have a common subject matter which they variously attempt to characterize in a way that is accurate, or, in other words, true. These are theories of a special sort: Whereas most theories attempt to characterize the truth concerning some mundane subject, theories of truth aim to characterize the truth concerning nature of truth itself.

This enterprise has an air of paradox about it: After all, one might think, to ask about “the nature of truth” is to ask “Which theory of truth is true?”, and unless we know the answer we can make no sense of the question. That would be too quick, however. Knowledge is not an all-or-nothing affair; it may very well be that we know enough about something in order to identify it without knowing everything there is to know about it. The proponents of the various “theories of truth” can maintain that all the rival theories can be compared, and thus legitimately be called theories of truth (as opposed to theories of correspondence, or theories of coherence, or theories of warranted assertability…) because there is a neutral core conception of truth which is common to all. The T-schema (“True (p) <--> p”) would seem well suited to play such a role. One could then fix the reference of “truth” as being whatever property it is that satisfies the T-schema.

If a correspondence theorist, a coherence theorist, and a warranted assertability theorist can all agree that truth satisfies the T-schema, what exactly is it that the theories would have in common? Should we hold, for example, that while all the theorists accept the T-schema, they each interpret it differently? If so, the question arises of what it is that is interpreted differently. Clearly it is not a meaning that is interpreted differently, for though one can grasp or fail to grasp a meaning, interpretation is a matter of assigning a meaning to something, such as a sound wave or a series of ink blotches. One cannot assign a meaning to another meaning. On this view, since the T-schema is ‘interpreted differently’ by the different theorists, it must be viewed syntactically, as a sequence of characters and nothing more. If the theorists simply assign different meanings to this schema, that no more makes their theories have something substantive in common than the fact that “burro” means donkey in Spanish and butter in Italian makes butter and donkeys have something substantive in common. In order for the different theories of truth to have something in common, then, it is not enough to accept the T-schema regarded syntactically. They must also assign it the same meaning. Consequently, their disagreement can only regard theses which are superadded to the schema to produce richer notions of truth.

Once this is granted, we can ask whether these various “rich” notions of truth are mutually consistent. For all that has been said, there may be many properties which satisfy the T-schema, each accurately characterized by its own “theory of truth”. The truth theories would then have something in common without being in competition. We can call this position “alethic pluralism”. On the other hand, those who wish to “keep the debate alive” have three basic options that I can see: The first option is to argue that there is only one property which satisfies the T-schema. The second option is to say that while there are multiple, mutually inconsistent theories of truth, they have something other than the T-schema in common. The third option is to “go deflationary” and insist that the T-schema is all there is to truth. We may as well call this the “null theory” of truth, for while on this view there are true sentences or statements, there is really no such thing as “truth itself”. Whatever the case may be, I hope that the different stances one can take are a little clearer than they were before.