When Donald Davidson died unexpectedly on August 30, 2003, the English-speaking world lost one of its most influential philosophers, one who had dominated debates about meaning, mind and language for 40 years.
Yet this reputation was built entirely on articles published in scholarly journals and edited collections of papers.
Even Truth and Predication , Davidson's last book, is a compilation of two lecture series, rather than being written from scratch. The first, the Dewey Lectures at Columbia University, was delivered in 1989 and was first published in the Journal of Philosophy in 1990. The second, the Hermes Lectures, was given at the University of Perugia in 2001 and until now was published only in Italian. It contains important new ideas, confirming that Davidson's thought was still as fertile, subtle and provocative as ever.
The other volume reviewed here is the last of five collections of previously published papers; almost all of its contents date from the 1990s. It contains such seminal pieces as "The folly of trying to define truth" (1996), "A nice derangement of epitaphs" (1986), and "Thinking causes" (1993). No serious philosopher or student of philosophy should be without a copy.
The issue that Davidson explores in his Hermes Lectures is what he calls "the problem of predication" but is more familiarly known as "the problem of the unity of the proposition". He traces the history of the problem from Plato and Aristotle, through Leibniz, Hume and Kant, to more recent manifestations in the work of Bradley, Frege, Russell and Wittgenstein.
Davidson contends that a solution to this problem has existed, unrecognised, for 70 years, in the shape of Alfred Tarski's semantic theory of truth. That Tarski should be the hero of Davidson's tale will come as no surprise to those familiar with Davidson's work in the theory of meaning, which is heavily indebted to Tarski. Even so, it will surprise many that Tarski should be credited with solving the problem of predication, and would probably have surprised Tarski himself.
What is the problem of predication? In a nutshell, it is this. Consider any simple subject-predicate sentence, such as (Davidson's favourite example), "Theaetetus sits". How are we to understand the different roles of the subject and the predicate in this sentence, "Theaetetus" and "sits"
respectively? The role of "Theaetetus" seems straightforward enough: it serves to name, and thereby to refer to or stand for, a certain particular human being. But what about "sits"? Many philosophers have been tempted to say that this also refers to or stands for something, namely, a property or universal that Theaetetus possesses or exemplifies: the property of sitting. This is said to be a universal, rather than a particular, because it can be possessed by many different individuals.
But now we have a problem, for this view of the matter seems to turn the sentence "Theaetetus sits" into a mere list of (two) names, each naming something different, one a particular and one a universal: "Theaetetus, sits." But a list of names is not a sentence because it is not the sort of thing that can be said to be true or false, in the way that "Theaetetus sits" clearly can. The temptation now is to say that reference to something else must be involved in addition to Theaetetus and the property of sitting, namely, the relation of possessing that Theaetetus has to that property. But it should be evident that this way of proceeding will simply generate the same problem, for now we have just turned the original sentence into a list of three names, "Theaetetus, possessing, sits."
Indeed, we are now setting out on a vicious infinite regress, which is commonly known as "Bradley's regress", in recognition of its modern discoverer, the British idealist philosopher F. H. Bradley. Bradley used the regress to argue in favour of absolute idealism, but Davidson is by no means prepared to follow him in that direction.
So what, short of following Bradley, is the solution to this problem? According to Davidson, solving the problem requires a philosophical elucidation of the notion of truth - not a definition of truth, because he does not think that one can be provided, at least where everyday natural language is concerned. But he is emphatic that certain traditional theories of truth, especially the so-called correspondence theory favoured by Russell and (the early) Wittgenstein, are of no use in this regard. This sort of theory says that "Theaetetus sits" is true in virtue of corresponding to a fact. But what fact - and what is a fact supposed to be, according to this view? Answer: a fact is something in the world - a part of extra-linguistic reality - that is of such a kind that its existence suffices to make a certain sentence or proposition true. In this case, the fact in question is simply the fact that Theaetetus sits, or is sitting.
This fact, according to the correspondence theorist, contains Theaetetus and the property of sitting, with the former in the relation of possessing to the latter. But it is easy to see that this "solution" to the problem of predication has just transferred the problem from language to the world.
Where we formerly had the problem of explaining the unity of the sentence or proposition "Theaetetus sits" - that is, the problem of explaining why this is not a mere list of names - we now have the parallel problem of explaining why the fact that Theaetetus sits is not a mere aggregate of objects, Theaetetus and the property of sitting, or these two objects and, in addition, the relation of possessing.
The great German philosopher Frege was aware of the problem and had a solution of his own. He believed that "Theaetetus" and "sits" each refer to something, the first being an object and the second a concept. But he held that concepts differ crucially from objects in being "unsaturated" or, in a certain sense, incomplete. Hence, objects and concepts do not need any kind of metaphysical "glue" to hold them together, because concepts have to apply to objects for their "completion". But Frege did not believe in facts as the worldly correlates of true sentences. According to him, all true sentences refer to, or name, the same thing - the True.
Davidson has great respect for Frege, but considers these doctrines of his to be failures. However, he does agree with Frege that if true sentences were to refer to anything, they would indeed all have to refer to the same thing. He just does not agree that they refer at all. To treat a sentence as referring to anything is to confuse a sentence with a name. Similarly, to treat a predicate, such as "sits", as referring to anything is to confuse it with a name.
So how, according to Davidson, does Tarski solve the problem of predication? In this way: he shows us how we can explain the function of predicates in our language without supposing that they refer to or stand for anything, such as properties or universals. He does this by showing how they contribute systematically to the truth-conditions of all of the sentences of our language in which they occur, starting with the most elementary subject-predicate sentences, such as "Theaetetus sits". The truth-condition of this sentence is simple enough to state: "Theaetetus sits" is true if and only if the predicate "sits" is true of the object referred to by the name "Theaetetus".
It may be protested that this is uninformative, because we are using the notion of a predicate's being true of an object to account for the truth of a sentence about that object. But Davidson sees nothing defective in this, because he has emphasised that he does not think that we can define truth: it is too fundamental a notion for that, but is nonetheless a notion that we all grasp clearly enough for practical purposes. Rather, his point is that Tarski's approach to truth enables us to understand the distinctive role of predicates in our language, and does so without invoking any useless metaphysical baggage, such as properties or universals.
Even so, some may feel dissatisfied with this account. Davidson, as we have seen, rejects the idea that there is anything extra-linguistic, such as a fact, that makes "Theaetetus sits" true. This is not to say he denies the objectivity of truth - far from it. What he denies is that there is some quite specific thing whose existence suffices for the truth of the sentence "Theaetetus sits".
But surely there is such a thing, namely, Theaetetus's act of sitting, which is a particular event. And it may seem odd that Davidson himself does not say this, given that he has long argued that we must recognise the existence of events as well as objects, as he does again in "Method and metaphysics" (1993), in the other volume under review here.
E. J. Lowe is professor of philosophy, Durham University.
Truth and Predication
Author - Donald Davidson
Publisher - Harvard University Press
Pages - 180
Price - £19.95
ISBN - 0 674 01525 8