20

The Status of Content INTRODUCTION An irrealist conception of a given region of discourse is the view that no real properties answer to the central predicates of the region in question. Any such conception emerges, invariably, as the result of the interaction of two forces. An account of the meaning of the central predicates, along with a conception of the sorts of property the world may contain, conspire to show that, if the predicates of the region are taken to express properties, their extensions would have to be deemed uniformly empty. The question then becomes whether the predicates are best understood as expressing properties, and hence as founded on error, or whether they ought to be understood along non-factualist lines.I Historically, irrealist models were developed primarily in connection with evaluative discourse, although as physicalism has flourished and as reductionist programs have failed, their application has been extended to many other domains. Indeed, it is one of the more influential suggestions in contemporary philosophy of mind that they apply even to ordinary belief/desire psychology. A correct understanding of the semantics and metaphysics of content-based psychology leaves us, so the proponents of the influential suggestion claim, with no option but to embrace an irrealist conception of that region of discourse.2 The influential suggestion has not gone unchallenged. Many philosophers have dissented from it-some by disputing the assumed metaphysics, others by rejecting the assumed account of psychological concepts-with inconclusive results.3 In this paper I wish to argue that, at least as things now stand, one or another of these dissidents must be right, for the irrealist conclusion itself is demonstrably unacceptable: at least as traditionally formulated, an irrealism about content is not merely implausible, it is incoherent. The present paper is intended as a challenge, to those who wish to propound such an irrealism, to formulate their view in a way that is not subject to the difficulties it raises. The basic arguments are fairly straightforward. They require mostly some clear thinking about what irrealist conceptions involve in general; what they involve as applied to content discourse in particular; and what sorts of consideration fuel content skepticism in the first place. Their combined destructive impact, however, is far-reaching and has not been adequately appreciated. The paper proceeds as follows. Part I explores the two different ways in which one might seek to make sense of the claim that no property answers to a given predicate. Part II outlines and clarifies what such irrealist conceptions look like when applied to contentful psychological idiom. Part III argues that these standard irrealist conceptions are unstable when applied to content discourse. Part IV attempts a redefinition that evades the outlined difficulties. Part V then argues (i) that even if this redefined position were stable, it could not accommodate the standard motivations for content irrealism; and (ii) that there is every good reason to doubt its own stability. 1. IRREALIST CONCEPTIONS Consider a fragment of discourse F, possessing a set of characteristic predicates and a set of declarative sentences involving those predicates. And suppose you come to have a worry of the following form: nothing possesses (or, perhaps, could possess) the sorts of property denoted by the characteristic predicates of F (if the predicates of F denote any sort of property at all). You become convinced, in other words, that if the predicates of F did express properties, their extensions would be uniformly empty: nothing in the world possesses the sorts of property that are the only candidates for being named by such predicates. A conviction of this sort has traditionally given rise to one of two possible conceptions of F, answering, respectively, to the assumption that the predicates of F express properties and to the assumption that they don’t. The first option leads to an error conception of F. An error theorist about a given fragment of discourse takes that fragment’s semantical appearances at face value: predicates denote properties and (hence) declarative sentences express genuine predicative judgments, equipped with truth conditions. However, the error theorist continues, because nothing actually exemplifies the properties so denoted, all the fragment’s (atomic) declarative sentences are systematically false. Most of us are error theorists about witch talk, for example, in something like this sense: we recognize that “is a witch” denotes a property that nothing really has. John Mackie has defended such a view of moral discourse.4 An error thesis about some fragment may give rise to one of two recommendations. It can lead to the “eliminativist” suggestion that the systematic falsity of F’s sentences constitutes sufficient grounds for its (eventual) elimination and replacement. Or, alternatively, it could result in an “instrumentalism” about F: in the view that, the falsity of its sentences notwithstanding, the continued use of F serves an instrumental purpose that will not easily be discharged in some other way. An error conception is the milder of the two possible reactions to our original ontological worry. A more radical reaction would be non factualism. According to this view, although F’s declarative sentences appear to express genuine predicative judgments, that appearance is wholly illusory. In actual fact, a nonfactualist alleges, F’s predicates do not denote properties; nor, as a result, do its declarative sentences express genuine predicative judgments, equipped with truth conditions: seeing as such sentences would be making no claim about the world, so nothing about the world could render them true or false. A non-factualist conception of F may itself come in one of two versions, depending on whether the attributed reference failure is thought of as “intended.” For, on the one hand, a predicate may fail to refer to a property even though it aspires so to refer; and, on the other, it may fail to refer to a property because it is no part of its semantic function so to refer. Naturally, the first verdict would lead to the recommendation that F be eliminated; whereas the second would merely prompt an alternative account of what the declarative sentences of F are designed to accomplish. How, if not as providing a vehicle for a statement of fact, should the semantic function of F’s sentences be understood? Perhaps as providing a vehicle for the expression of certain sorts of attitude. Such an “expressivist” view, applied to moral discourse, was presented with characteristic brio by A. J. Ayer: The presence of an ethical symbol in a proposition adds nothing to its factual content. Thus if I say to someone, You acted wrongly in stealing that money', I am not stating anything more than if I had said,You stole that money’. In adding that this action is wrong I am not making any further statement about it. I am simply evincing my moral disapproval of it. It is as if I had said You stole that money', in a peculiar tone of horror, or written it with the addition of some special exclamation marks.5 Conceptions similar in spirit have been proposed for a wide variety of regions of assertoric discourse, including the aesthetic, the causal, and the counterfactual. What all non-factualist conceptions have in common-what in effect is constitutive of such a conception of a declarative sentence of the form "x is P"-is (1) The claim that the predicate "P" does not denote a property 6 and (hence) (2) The claim that the overall (atomic) declarative sentence in which it appears does not express a truth condition. Now, the first point I wish to underscore is that there is a perspective within the foundations of semantics-specifically, a conception of truth and of truth conditions-from which any such nonfactualist view is bound to appear unintelligible. Curiously, it is again in Ayer that we find an incisive statement of the perspective in question: [T]here is no problem of truth as it is ordinarily conceived. The traditional conception of truth as areal quality’ or a real relation' is due, like most philosophical mistakes, to a failure to analyze sentences correctly. There are sentences ... in which the wordtruth’ seems to stand for something real … [but] our analysis has shown that the word `truth’ does not stand for anything. 7 The conception of truth Ayer is giving voice to here has come to be known as a “deflationary” or “disappearance” view of truth. It is characterized by the claim that there is no such thing as the property of truth, a property that sentences or thoughts may enjoy, and that would be named by the words “true” or “truth.” By way of probing such a view, let us ask how it understands the significance of “truth attributions”-of sentences of the form “x is true.” There are, in effect, two options. On the one hand, a deflationist may attempt a general account of truth attributions that explains in what their semantic function consists, if it is not to consist in that of attributing a property; or, on the other, he may reject the very idea of such a general account, recommending in its place the much more modest task of specifying truth-in-a-particular-language-“truth-in- L.” In a word, the options are “performative” theories, on the one hand, and “disquotational” theories, on the other. Performative theories, originally proposed by William James,8 and later by Peter Strawson,9 are in effect analogues of expressivist conceptions of ethical statements. Acts of calling something true (compare acts of calling something good) are assimilated to acts of expressing praise for something, rather than to acts of describing that thing as possessing some specific property. As Rorty has succinctly put it, the proposal is that “true” is simply a compliment we pay sentences we are prepared to assert.10 Disquotational theories, by contrast, forego any attempt at a general account of truth attributions. According to them, the basic deflationary thought that truth is not a real property is correctly elaborated not by an alternative account of what truth in general consists in, but by the rejection of any such account. To the extent that we can talk about a theory of truth at all, this is to consist in no more than a recursive definition of truth-in-L. What sort of notion is that of truth-ina-language? It is, in Quine’s apt phrase, the idea of a device for semantic ascent-a device for talking about snow or whiteness, by talking about sentences that are about snow or whiteness. To be sure, such a device has a use: it provides a handy way of affirming or rejecting an infinite lot of facts; but it is, for all that, in principle, dispensable. My concern just now is not with a deflationary conception of truth as such;’2 it is, rather, with the tension between such a conception and a non-factualist thesis about a given region of assertoric discourse. 13 To bring the tension into focus, let us ask what conditions a sentence must satisfy if, on a deflationary construal of truth, it is so much as to be a candidate for truth. What, in other words, certifies a sentence as truth-conditional, on a deflationary construal of truth? Two minimal requirements suggest themselves: first, the sentence must be significant, and second, it must be declarative in form. Unpacking somewhat, the requirements are that the sentence possess a role within the language: its use must be appropriately disciplined by norms of correct utterance; and that it possess an appropriate syntax: it must admit of coherent embedding within negation, the conditional, and other connectives, and within contexts of propositional attitude. 14 The status of these two requirements, as individually necessary conditions for candidacy for deflationary truth, is certified by the deflationary conception itself (which is just as it should be). Thus, it is required by the claim that “true” is a device for semantic ascent, that the sentence to which “true” is predicated be both meaningful and declarative. Since, on such a view, the overall effect of asserting that a sentence is true is just to assert the sentence itself, the requirements on truth predication must include whatever requirements attend candidacy for assertion itself; and, clearly, a sentence must be both meaningful and assertoric in form if it is to be a candidate for assertion. Similarly, a sentence must satisfy both these minimal conditions if it is to be a candidate for “performative” truth, for they are both required by the claim that “true” is a compliment we pay sentences we are prepared to assert. The tension between a deflationary understanding of truth and a non-factualist thesis stems from the fact that these requirements would seem also to be jointly sufficient for truth conditionality, on a deflationary understanding of truth. For if they are jointly sufficient, then there is no more to a sentence’s being truthconditional-genuinely apt for (deflationary) truth and falsity-than its being a significant sentence possessing the appropriate syntactic potentialities. But it is constitutive of nonfactualism precisely that it denies, of some targeted significant, declarative sentence that it is truthconditional. On a deflationary conception of what it is to possess truth conditions, there would be, simply, no space for such a possibility. But why must the conditions be thought of as jointly sufficient? Isn’t there room for the suggestion that more is required for truth conditionality, consistent with subscription to a deflationism about truth? Isn’t it imaginable, that is, that someone might require that a given region of discourse meet certain further conditions-reducibility to the vocabulary of basic science, for example-if it is to be genuinely truth-conditional, and yet remain a deflationist about truth? It is hard to make sense of the suggestion. The difficulty lies in seeing how any such further requirement would be motivated. Any proposed requirement on candidacy for truth must be grounded in the preferred account of the nature of truth. On a deflationary account of truth, there is no substantive property -truth-that sentences or thoughts may enjoy; on such a view, on the contrary, all truth talk consists either in the evincing of a certain sort of praise (with the pragmatists) or in the deployment of semantic ascent (with the disquotationalists). Both of these articulations of deflationism require the two conditions outlined. But how could they conceivably require more? Any meaningful, declarative sentence would be (at a minimum) a candidate for assertion; it would be, thereby, a candidate for the compliment we pay sentences we are prepared to assert, or, as the alternative would have it, a candidate for semantic ascent. Any such sentence would count, therefore, as truth-conditional in a deflationary sense.15 It would appear to be a point to which Ayer must have paid inadequate attention. A nonfactualism about any subject matter presupposes a conception of truth richer than the deflationary: it is committed to holding that the predicate “true” stands for some sort of real, languageindependent property, eligibility for which will not be certified solely by the fact that a sentence is declarative and significant. Otherwise, there will be no understanding its claim that a significant sentence, declarative in form, fails to possess truth condi- tions.16 It will be important to the argument later on to observe that we could have approached this very conclusion from a somewhat different direction: by beginning with the non-factualist’s denial that the predicate in “x is P” refers to a property, rather than with his denial that the sentence as a whole possesses factual content. Corresponding to the distinction between deflationary and robust conceptions of truth, there is a distinction between deflationary and robust conceptions of reference. A deflationary understanding of “refers” would be this: a term refers to a property provided it has the syntax of a predicate and possesses a role in the language. What is denied is that the expression “refers to a property” expresses some sort of objective relation that may obtain between predicates and language-independent properties, a relation of the sort that causal theories of predicate reference may be understood to be attempting to elucidate.17 And, again, it is clear that a non-factualist is committed to a conception of predicate reference richer than the deflationary. For what a nonfactualist wishes precisely to say is that some expressions-like “wrong” or “cause” or “beautiful” or “funny”-which have the syntax of predicates and which possess perfectly welldefined roles within the language, nevertheless fail to refer to any real property. This claim is intelligible only against the background of a robust conception of reference. Since, however, it is a platitude that “x is P” is true if and only if the object denoted by “x” has the property expressed by “P,” a non-factualist’s denial that a particular predicate refers to a property is by itself sufficient to anchor his commitment to a robust conception of truth. For if the predicate in “x is P” might fail to refer to a property, then the overall declarative sentence of which it’s a part might fail to possess a truth condition. However, declarative sentences cannot fail to possess truth conditions except against the background of a robust conception of truth. Hence, a commitment to a robust conception of reference would appear to entrain a commitment to a robust conception of truth, just as expected. Our discussion of error theories can afford to be much briefer. An error thesis about the sentence “x is P” is simply the view that, because nothing has the property avowedly denoted by “P,” (3) “x is P” is always false. It is easy to see that an error theory, in contrast with a non-factualist thesis, is not locked into any particular understanding of the central semantic notions. Because an error theory does not rely upon a distinction between apparently referential and genuinely referential devices, or apparently truthconditional and genuinely truth-conditional sentences, it is intelligible on both robust and deflationary understandings of reference and truth. What any error theory is committed to, however, is simply this: that there actually are some sentences that possess truth conditions (on whatever understanding of that notion is favored). The commitment is evident: an error thesis presupposes that the targeted declarative sentences possess truth conditions, otherwise it couldn’t call them “false.” To sum up this general discussion of irrealist conceptions: I have argued that a non-factualist model of a given region of discourse presupposes robust conceptions of truth and reference; and that an error theory of that region presupposes that its sentences are truth-conditional, on whatever construal of truth is favored. What I am going to argue is that an application of these models to content discourse itself runs afoul of their respective presuppositions.18