Like everyone else, metaphysicians want their beliefs to be true. But true in what sense? Necessarily true? Monopolistically or exclusively true? True in the sense of corresponding to how things are in the world? In the sense rather of corresponding to the way they are? In the sense of being verified or warranted by the evidence to date? Of cohering with the whole body of all our beliefs? Of being what our friends will let us get away with saying? None of the above?

Each of these possibilities represents a particular theory or account of truth. This includes accounts according to which truth is not the sort of thing about which we should expect to have some deep theory; `true' is just the compliment we pay to the sentences we happen to accept. Which account should we believe?

Note first that metaphysics is not committed to any particular account of truth. Some metaphysicians are committed to one, some to another, some to none at all. Consequently, objections to any given account of truth spell no trouble for metaphysics as such, but only for those varieties (if any) that happen to adopt that account. At the same time, metaphysicians do want to say something about extra-linguistic matters, even if there is no agreed sense in which what they say of them is supposed to be true. Earlier, we saw how Quine, the structuralists, and the deconstructionists all question the very idea of a determinate tie of aboutness or reference between words and the world; they are by no means alone in doing so. Rejecting determinate reference would undermine metaphysics to the extent that metaphysical assertions -- indeed, any assertions -- are to be determinately about matters beyond language or beyond the text. Rejecting determinate reference would also undermine certain accounts of truth, realist accounts in particular, according to which a true sentence is made true by something in the world, not by our evidence, and not by consensus, coherence, interpretation, linguistic projection, or conceptual scheme. Realists about truth believe that truth is a word-world affair, not word-word, and in such a way that the conditions of truth are in the world, not within the mind.

Like realism as regards the world, this realism as regards truth is distinguished by a kind of "declaration of independence": truth is independent of such perspectival or mental matters as consensus, interpretation, possession of evidence, coherence with our other beliefs, conceptual scheme, and so on. But even though realism as regards truth and realism as regards the world share this declaration of independence, neither entails the other. It is possible to be a realist about the world and also an anti-realist about truth. Many philosophers are; they believe that there are mind-independent items in the world, but not that true sentences are made true by them. It is possible to be the reverse: a realist about truth yet an idealist about the world; a true sentence is made true by items in the world, yet the items are mind-dependent. Or one can be a realist in both senses or in neither.

In the previous chapter we noted how shaky is the inference from certain theories of meaning to anti-realism about the world. By contrast, the inference from them to anti-realism about truth is in much better shape. The question is not whether anti-realism about truth follows from them -- mostly it does -- but whether they are adequate as accounts of meaning. If not, they provide no reason to reject realism as regards either truth or the world, hence none to reject those varieties of metaphysics committed to either sort of realism.

Theories of meaning that would eliminate reference usually begin by supposing or arguing that the alleged reference of a bit of language, like all meaning, must be determined by a certain kind of fact. They then explain how the reference is not determined by such fact, and conclude there can be no determinate relation of reference between words and the world. For example, recall how Quine supposes that meaning, hence any alleged reference, would have to be determined ultimately by the behavioral or at least by the neuro-physiological states or dispositions of the speaker. He then explains, via the "Gavagai" example, how such matters do not after all fix the reference. They could be the same yet the reference vary, between rabbits, their stages, undetached parts, and so on. The conclusion is that there can be no determinate relation of reference. Reference is inscrutable, and inscrutable for the simple reason that there is nothing there -- no determinate reference -- to scrutinize. Structuralists and deconstructionists afford another example of the same general strategy. They suppose that all meaning is determined by certain word-word relations internal to a language. Since reference is not so determined, the conclusion to be drawn is either agnosticism about or outright rejection of any such word-world relation.

What these strategies suggest is that reference is not determined without remainder by anything "in the head": not by brain states, not by dispositions to behave in certain ways, not by our associations of words with certain words, not by any mental imagery we may have that seems to be "of" or to "present" the alleged referent. Auxiliary arguments suggest further that reference is also not determined by our intentions. An additional argument, advanced by Putnam, seems to many to seal the verdict that, as he puts it, " `meanings' just ain't in the head." Imagine that far, far away there is a planet called Twin Earth, particle-for-particle just like Earth, as are the people on it. Each of us has a Twin-Earth double whose behavior, brain states, awareness, word-word associations, and conscious intentions are the same as ours. So too do the Twin Earthlings include among themselves speakers of a language phonologically and syntactically the same as English. In circumstances in which we would utter `There goes a rabbit!' our Twins would too. But it happens that Twin rabbits, though otherwise like Earth rabbits, are made not of carbon-based organic molecules but silicon-based. Hence Twin rabbits are not rabbits. Whereas our utterance refers to a rabbit, the same utterance by our Twins refers to a Twin rabbit; the reference is just not the same.

Putnam's ingenious bit of science fiction is widely taken to show that reference does not depend solely on -- is not determined without remainder by -- the speaker's word-word associations, behavior, imagery, or mental states, intentional or otherwise. For these are all the same on Twin Earth as here. Yet the reference differs: our Twins are talking about Twin rabbits, we are talking about rabbits. Meaning, in any sense that could determine aboutness and reference, "just ain't in the head." If one adds to this the assumption that it cannot be anywhere else, as at least implicitly do Putnam, Quine, the structuralists and deconstructionists, it follows that there can be no determinate reference relation between words and the world.

Philosophers there are who would reverse this verdict. There is such a relation, all right, but it is not to be found in the head. Those who assume it cannot be anywhere else seem to do so under the influence of meaning-rationalism. They think that speakers would have to have a priori knowledge of what they mean, and the only way to have such knowledge is for the meaning to be in the head or mind, where the speaker can have a priori access to it. Contrary to this meaning-rationalism, the moral of Twin Earth, we are told, is not skepticism about reference but skepticism about meaning-rationalism and about theories of meaning that reduce meaning to what is in the head -- to the speaker's associations of words with words, to brain states, imagery, dispositions to behave in certain ways, to intentions, or whatever. The problem, of course, is to say just what reference does consist in, if not some such intra-cranial affair. Solving this problem is necessary if metaphysicians wish to pierce the veil of language -- not in order to look on reality bare, whatever that would mean, or to have the things-in-themselves presented to consciousness with dazzling immediacy, but simply to say something determinately or objectively true of what there is.