Critical Study

Critical Study

Richard Schantz (ed.), What is Truth? Berlin: de Gruyter, 2002. 339 pages. US$ 54.95 (hardback).

What is Truth? is a collection of original philosophical articles by many of the central figures in the field. Most of the contributions are focused on deflationism, for and against, although other approaches have a fair airing, and some novel accounts are presented. The intrinsic worth of many of the papers apart, the interest of the collection arises, I think, from its bringing into relief a number of problematic lacunae within the extant deflationisms, which, I predict, will be the main area of controversy in the years to come.

The currently entertained options about truth may be usefully coordinated in the space opened by Frege’s work. Roughly, Frege distinguished two types of construction in which the adjective true predicatively occurs. In the first type, the predicate takes a full finite complement and has an expletive subject; e.g., It is true that violets are blue (these types may be read as extraposed versions of the type (e.g.) [the thought] that violets are blue is true.) For these constructions, Frege suggested that the contents they express are flat with the contents of their complements, i.e., to entertain the truth of a thought is not to picture the thought or its constituents in relation or correspondence with any external elements, be they facts or objects; rather, the contents of truth predications may be expressed by the contents to which truth is predicated. Hence, we have the origin of the idea behind the familiar schema

(ST) TRUE(P) iff P

(Let the left flank be proxy for any type of truth predication to an explicitly presented full sentential clause. Let ST to be a generalisation over the various schemata that have been forwarded.) The second type of construction which occupied Frege is where the adjective is predicated of a quantifier noun phrase with no sentential complement; e.g., Everything Bob said is true. Here, there is no articulable content with which the predication may be flush. After all, one need not know anything of what Bob said to entertain a general thought about his veracity. In light of this difference and the infamous ‘regress argument’, Frege concluded that our concept of truth is sui generis: it neither unpacks into a uniform relation of correspondence or some such, as the fist type of construction shows, but nor does it unpack merely into those contents to which it is predicated, as the second type of construction shows.[1]

In the main, the contributions to the present volume may be plotted in this space. Roughly, correspondence theories (or, more generally, ‘inflationist’ ones) depart from Frege by, we may say, elevating truth inasmuch as the concept is credited with an underlying relational structure. Thus, the impression given by instances of ST is, in some sense, misleading, and one limb of Frege’s argument for the primitiveness of truth is fractured. The deflationists, on the other hand, think that Frege was right (under a reading which is not Frege’s) about the flatness of the first type of construction, but wrong about what the second type entails. The central thought is that truth is flat with the contents to which it applies tout court. What the apparent recalcitrant elevation of the truth predicate in the second type of construction reveals is merely the raison d’être of our concept of truth given the indirection of the predicate in the first type of construction. In brief, the truth predicate has an ineliminable expressive role in enabling us to mimic the effect of making an indefinite or infinite number of assertions of those contents covered by quantifier noun phrases; yet the predicate is semantically inert, for the generalisations unpack, content-wise, into infinite disjunctions/conjunctions upon which truth is flattened, e.g., the conjunction of the instances of the schema, If Bob said ‘p’, then p.[2]

A third, somewhat motley, type of position cleaves to Frege’s reasoning. On this view, truth is a primitive notion qua neither analysable as a relation nor merely flat with antecedent contents. Thus, on the kind of view held by Davidson[3], Dummett, Wiggins and others, truth is immanent to our understanding of the contents themselves; truth is not an expressive notion subservient to an antecedent understanding of content.

The collection is broken into five sections. The first group of essays deal with the correspondence theory. Contributions from Alston, Schantz and Devitt argue, in their different ways, that a deflationary account of the function or meaning of the truth predicate as an expressive device does not lead us to any particular account of the property of truth, the kind of account a correspondence theory offers. If, then, deflationism is to reveal that truth is insubstantial - show that no correspondence theory inter alia is available - much more needs to be offered than a functional story about the predicate, for that can be common ground. Devitt attributes deflationism’s myopia here to “a use/mention sloppiness”; similarly, Alston identifies a conflation of concept with property.

Independent of the virtues, if any, of the correspondence theory, this challenge is formidable. Deflationists tend to be proprietary about instances of ST, but as we just sketched, in themselves, they are acceptable to all major disputants. So, if the instances are generally acceptable to mutually inconsistent theories, then, by themselves, the instances hardly constitute a distinctive, complete account of truth. Nor does the expressive tale about the truth predicate’s role in generalisations alter this independence, for that account is licensed precisely by the reading of ST under which the truth predicate is revealed as being essentially semantically inert. If such a reading is denied, then it is hardly mandatory to take the role of the predicate in generalisations to be merely expressive. Further, it is prima facie mistaken to think of generalisations as compendious assertions: access to the contents which would comprise the putative conjuncts/disjuncts appears to be independent of the competence called upon to entertain the generalisation. In brief, then, Devitt and Alston may be read as suggesting that the ST equivalencies do not assert their own triviality, as it were; if they are trivial, it seems that this must be grounded in some conception of the property of truth. Somewhat analogous to Boghossian’s ‘reductio’[4], this thought effects a sort of deconstruction - a description which would perhaps be abjured - of deflationism: the doctrine requires a motivating metaphysics that it officially denies.

Contributions from Armstrong and Boyd seek to articulate the kind of metaphysics of facts and ‘truth making’ apparently involved in the correspondence conception. Schantz also nods in this direction by suggesting, a la pre-deflationary Field, that truth may be grounded in a causal theory of reference. While these articles supply some thought on how to fill out a correspondence theory, they leave the deflationary conception unmolested.. One can appreciate the critical force of the ‘deconstruction’ without having any sympathy for correspondence theories in particular. These latter papers do not do much to make this reviewer sympathetic.

The second section comprises contributions from Horwich, Brandom, Grover and Michael Williams. Each offer tweaks and twists to their respective well known deflationary positions. The papers by Brandom, Williams and Grover are to be especially recommended for their emphasis that deflationism ought not to be seen as the mere negative thesis that there is no property of truth. This is not to say that deflationism in fact has a story to tell about such a property, but that, for these writers, at least, deflationism is a genuinely explanatory thesis, although a circumscribed one. Horwich details how his brand of deflationism may capture certain norms as to why we ought to believe truths. Grover essays a very elegant account of how the ‘world’ plays an explanatory role in deflationism, although the world here is not a relatum to which truths correspond, but what-is-true: just as the content of the truth predicate is flat with the contents to which it applies, so being true is flat with what is true. Most central, though, is the claim of Brandom and Williams that a deflationary account of truth should be integrated into an account of meaning or content. Here we see the sort of response a deflationist may make to the ‘deconstruction’: the exhaustive character of the ST equivalencies is grounded in an account of content rather than a putative metaphysics of the property of truth. This is a claim also pressed in Horwich and Field’s recent work.[5] Here also is where the delimitation of the potential explanatory role of truth is most vivid and problematic.

At its most basic, all of the various deflationary positions subscribe to the idea that the content of ‘is true’ is exhausted by, or inherited from, the contents to which it applies. (Brandom argues that a deflationism which takes ‘is true’ to behave predicatively misses a range of expressive features that may be accommodated by his anaphoric operator position, but the flatness is a constant.) Crucially, this flat characterisation is taken to be definitive or explicatory of truth via a notion of content in place independent of a conception of truth. Thus, in contradiction to the traditional wisdom stemming from Frege, deflationism deems truth unfit to play a central role in the understanding of content. In other words, deflationism implicitly commits itself to a certain true-free picture of content upon which deflationism might be grounded.[6] None of the papers articulate just what content should amount to on this deflationary view, but instead defer to previous work. The papers are revealing though, especially those of Brandom and Williams, in moving this issue to the front of the debate.

The third section is belligerently entitled ‘Deflationism Attacked’ and consists of papers by David, McGinn, Künne and van Inwagen that variously question the central claims of deflationism. The most effective paper here is Künne’s. He offers an ‘overload argument’ that the various deflationary positions are obliged to credit any possessor of the concept of truth with conceptual resources no speaker could possibly enjoy. The gist of the argument is that if truth is flat with the contents to which it applies, then one’s possessing the concept of truth appears to involve one’s possessing every content, i.e., every content that may be articulated on the left flank of an instance of ST. Most welcome here is Künne’s employment of this argument against the deflationary reading of generalisations: if truth generalisations give way to infinite conjunctions/disjunctions, then, it seems, we must have competence with the infinite contents covered. But this consequent is transparently false. Thus, the generalisations cannot be read as abbreviations of infinite lists. Künne goes on to ably show how this criticism is not vitiated by an idiolect-relative approach to deflationism. The point is disarmingly simple. It might be that our understanding of P is true is relative to our competence with P, i.e., the less determinate the content of P is for us, the less determinate is our understanding of what it is for P to be true. However, it makes no sense to think that our understanding of truth predications to quantifier noun phrases might be similarly relativised, for the relevant contents here are picked out by the NPs, they are not articulated at all, and there is no easy way to see how the contents covered by the NPs might be restricted to what one understands. For example, one can perfectly well judge that, say, most of what Bob said, which I didn’t understand, is true or every theorem of ZF set theory is true, but I only understand the basics. (These cases are analogous to the cancellations of supposed constitutive presuppositions, as in It’s true that the king of France isn’t bald, because there is no king of France.) Künne does not mention Grover’s prosententialism or Brandom’s variation on it, but his argument may be easily extended.[7]

Although derivative of Gupta’s 1993 argument[8], Künne’s paper is limpid and convincingly shows that attempts to evade the overload problem sink deflationism into a bog of increasingly unacceptable qualifications. The paper also serves to link the basic themes of the first two sections. On the one hand, the ST instances and the attendant ‘functions’ of the truth predicate may well be trivially acceptable, but if they are interpreted in the usual flat deflationary manner, then absurdity looms. On the other hand, what is required from deflationism is an account of content or perhaps the property of truth to evade this consequence, but, notwithstanding revisionary hyperbole typical of deflationists, this just means that deflationism is at best fundamentally incomplete. The other three papers of the section are usefully interpreted in this light.

David argues that the kind of constraints Horwich places on theories that may legitimately work in association with his minimal theory (the non-denumerable collection of the propositional instances of ST) result in an ensemble of theories much more bloated and unstable than result from letting truth be constitutive of our explanations of knowledge, meaning, logic, etc.

McGinn echoes the complaints of the correspondence theories of the first section, and suggests that while the deflationary paradigms are trivially acceptable, they are not best understood in a deflationary manner. McGinn argues for a “thick disquotationism”, which seeks to explain the deflationary paradigms by the thesis that truth is a peculiar self-effacing property: its conditions of application are given without appeal to any predicate that designates the property. He goes on to show why such a property is of importance to us, beyond the familiar expressive exigencies touted by deflationism. The paper is quite lovely.

van Inwagen’s paper is dedicated to showing that quantification over sentential position is meaningless. If this is right, then the deflationist cannot generalise any of the various deflationary schemata to offer an explicit or eliminative definition of truth. van Inwagen might well be right on quantification, but his background assumptions mischaracterise many deflationists. Horwich and Field, for instance, are quite explicit that the good of the truth predicate is precisely that it goes proxy for sentential substitutional quantification in the sense that it merely provides a predicate for objectual variables that would otherwise be stranded in clausal position, i.e., the predicate allows one to assert every instance of, say, If Bob said ‘p’, then p, as an objectual quantification. The point here, though, is that the predicate which saves the consequent ‘p’ is, for each instantiation of the variables, dischargable via ST. Of course, van Inwagen might find the proxy idea to be equally meaningless, but then he needs the kind of arguments presented in Künne to make the point stick, for the proxy idea does not syntactically involve quantification over sentential position. Nor are deflationists typically interested in explicit definitions, not Quine, Horwich, Field, Brandom, Grover, Williams, Soames, etc.

The fourth section consists of just two papers on Tarskian issues. The first is by Gupta on the inadequacy of Tarski’s Convention T to provide either necessary or sufficient conditions for the sense of our colloquial truth concept (although it is sufficient to fix the extension of ‘true-in-L’.) Gupta’s moral is that insofar as deflationism relies on Convention T, it is reciprocally inadequate. The second paper is by Hintikka on his campaign for independence friendly logic and its relation to truth.

Pace Gupta’s apparent understanding, most deflationists go out of their way to distance themselves from Tarski, certainly none of the key figures have anything approaching an uncritical acceptance of Convention T. Gupta’s basic point, however, is fundamental: if we seek to capture the sense of any concept in terms of that to which it applies, then we are bound to overload it. Gupta’s paper complements Künne’s. The force of Hintikka’s paper is harder to discern.

Hintikka suggests that our commonsensical notion of truth amounts to Skolem functions; or, better, P is true expresses the Skolem functions of P (Hintikka’s elliptical paper assumes a reader familiar with much technology beyond first-order logic.) To explain: Skolem functions are ways of representing existential quantifiers in the scope of universal ones. The representation is governed by the Skolem normal form theorem which says that every first-order quantifier formula is equivalent to a second-order prenex formula. Thus: e.g., ("x)("y)($z)[R(x, y, z)] iff ($f²)("x)("y)[R(x, y, f²(x, y))], where ‘f’ ranges over functions which take the values of the universally bound variables as arguments. This employment of Skolem functions is due to Henkin, and Hintikka takes the Skolem normal forms to be essential for the representation of the truth conditions of quantificational branching (scope independent) structures as found in natural language.[9] A truth predicate for an independence friendly first-order logic, then, amounts to (given established translation relations) the second-order expression of the existence of the Skolem functions for the sentences in the predicate’s range.

Hintikka’s argument is somewhat undermotivated, apart from the quite controversial status of branching quantification structures in natural language, of which Hintikka gives no indication.[10] Much more so than Gupta, Hintikka argues as if general conclusions about truth are forthcoming from specific results about the limitations of the Tarskian definitional format, specifically, that it cannot compositionally accommodate branching structures. But, as Hintikka well knows, independent quantificational structures are but an example of a wealth of semantic structure (including quantification generally) that is beyond fist-order regimentation, and IF logic appears to offer no better account of most of this structure.

The fifth and final section consists of ‘Alternative Approaches’ offered by Dummett, Puntel, Rosenberg, Walker and Wiggins. Rosenberg advances a careful and clear reappraisal of Peirce’s pragmatism. The paper’s focus is on answering the objection that truth cannot be characterised in terms of methodology, for sound methodology depends on an antecedent notion of truth. Rosenberg convincingly shows that it can, at least in Peirce’s own terms. Independent of whether one ultimately accepts the Peircean view that thought is for the production of “habits of action”, Rosenberg’s argument is very interesting in opening up what may be dubbed a cognitive conception of truth, one which understands the concept as an expression of the structure of thought, rather than as a metaphysical notion at which thought aims. In such a light may be understood Rosenberg’s parting shot that deflationism’s flatness is merely synchronously accurate, but fails to capture the concept’s diachronic regulative role that adds structure and self-appraisal to our thoughts. The paper should take its place among the recent Peircean revivalist work.[11]

Walker raises the traditional metaphysical/epistemological problem of how humans manage to get anything right at all, how it is that our rational principles ‘match’ the world. The problem, for sure, is a real one, but, while Walker’s discussion is lucid and historically interesting, it does not engage with the current discussion of the problem, precisely because Walker imagines that it is widely neglected. On the contrary, Field has been grappling with it and its relation to deflationism for the past decade or so; the present contributions of Grover and Brandom make it patent that they would reject the idea that the problem, if real, bears on truth at all. Further afield, Chomsky has long been concerned with precisely this issue (a variant of what he calls Plato’s problem), as has Fodor.[12] Again, neither are mentioned.

Puntel sketches a novel account of the role of the truth predicate as formative of an operator that (roughly) provides a “determinate status” as value to the otherwise “indeterminate” sentential arguments over which it ranges. The idea appears to me to be reminiscent of Frege’s turnstile, ‘d’, where the vertical stroke adds assertoric force to a content of possible judgment as marked by the horizontal stoke. Puntel seems to be dimly aware of the similarity, but registers force as pragmatics, which, indeed following Frege, is subservient to semantics. Yet it remains unclear just what Puntel means by “determinate” status and why a sentence without a truth predicate should in any way be indeterminate. That is, Puntel seems to be merely describing the platitudinous relation of truth to assertion in somewhat highfalutin terms. Puntel’s case is not helped by his bizarre insistence on the most derogatory descriptions of the work of Tarski and Quine.

The papers of Dummett and Wiggins advance what we may call the Frege view. In their different ways, both argue that, while truth is a primitive notion, it is best seen as essentially tied to the wider intentional profile of our communicative and meaningful practices. Dummett suggests that two approaches to truth need to be separated. On the one hand, we have what he calls ‘level’ theories which attempt to describe the semantic behaviour of the truth predicate within our language. On the other hand, there is the putative role of truth in a theory of meaning which targets linguistic behaviour in toto. From the optic of this distinction, we may think of deflationism as born in the sin of confusing the levels. Thus, following Frege, Dummett suggests that the propositional instances of ST (what Dummett calls a ‘p-level theory’) fail to be explicatory of the sense of the truth predicate because the propositional elements cannot be properly identified without an implicit grasp of the concept of truth that provides conditions for their individuation in assertion and thought; it is such an implicit grasp that a general theory of meaning must in part depict. So, deflationism goes deep into the explanatory red by construing an account of the behaviour of the truth predicate within language as if such an account rules out truth conditions entering into the explanation of that behaviour from outside of the language.

The deflationist might now protest, as Brandom and Williams do in their contributions, that truth cannot play a “global” explicatory role (Brandom’s term) precisely because its role within language is antecedently determined by the shape of content. But this, in itself, is to confuse levels again. As a description within the language, Dummett favours sentential instances of ST - a piecemeal (not global) ‘s-level theory’ - in that they reflect the semantic shift from a thought about a truth candidate to the thought expressed by that truth candidate. The claim here is that one can identify the sentential constituents and their composition independently of a notion of truth, and so the conditions for the predication of truth serve to articulate the content expressed by the sentence in a revelatory manner. This one cannot do with propositions. Yet Dummett’s point here is not that truth explicates meaning in the way which incurs deflationary criticism. For Dummett, if the ST instances offer a description of our grasp of the sense of ‘is true’, then they will not explain our grasp of the meaning of the embedded sentences because our grasp of the instances antecedently depends on an understanding of the articulated contents of their right flanks. Thus, Dummett concedes the deflationary point; indeed, he invented it. (This argument goes back to Dummett’s seminal 1959 article ‘Truth’.[13]) But again, what counts as an adequate description of the truth predicate’s behaviour at the level of the language does not perforce double as a theoretical explication at the level of content as such; at this level, a concept of truth is required to so much as identify propositions/contents.

Dummett’s discussion is typically sophisticated and repays study, the basic point, though, is that no deflationary theory of the behaviour of our truth predicate - and they are the only ones we have - is apt to usurp the explicatory role of truth to target content, for, as best as can be seen, truth is involved in the very identification of our thoughts as being items of assertion and communication. To think otherwise is to be confused about the appropriate levels, and to argue against an explicatory role for truth at the level of the language does nothing to militate against its apparent indispensability at the theoretical level.

Wiggins is similarly good at bringing into relief that the mere apparent triviality of the deflationary paradigms and the supposed attendant function of the truth predicate does not amount to a deflationism in any interesting sense. Truth, for Wiggins, remains a primitive dimension of evaluation, although one that may be variously elucidated by citing its relations to other ranges of concepts and interests. Wiggins appeals to Peirce as a paradigm of such elucidation in terms of the practice of inquiry.

Looking over the collection as a whole, we find that more or less all the contributors, save for the ‘official’ deflationists, explicitly commend the aspects of the truth predicate that encourage deflationism but also say that these aspects by themselves do not amount to any particularly interesting theory of truth, or otherwise fail to rule out substantive alternatives. The deflationists also recognise that various supplementary theories are required, although ones deemed to be strictly independent of the account of truth proper. What is perhaps not properly recognised by deflationists, either here or elsewhere, is that once it is accepted that a proprietary conception of truth cannot be unfurled from ST, then its instances effectively amount to data from which (inter alia) we may test candidate accounts of truth. Deflationism, in other words, has no kind of conceptual priority at all, it is on an equal footing with primitive property accounts (McGinn, Dummett, Wiggins, Davidson), correspondence theories (Alston, Devitt), pragmatist theories (Rosenberg, Misak), or other potential accounts (Hintikka, Puntel, Gupta). Worse, deflationism appears to be in somewhat of an invidious position, for now it must support the supposed ‘deflationary equivalencies’ in broader terms of meaning, metaphysics, cognitive capacity, or whatever. The typical anti-deflationist, on the other hand, has always said that truth is a complicated notion in as much as any worthwhile story of the concept would involve such robust associations with other domains, either constitutively or as exhibiting the explanatory role of a primitive concept. It is unclear if anything distinbctively deflationary would remain after the required supplementary theories are in place.

As if in riposte to this kind of dialectic, Williams’s paper offers two suggestions. (i) Deflationism should be the default position from which putative alternatives must distant themselves and (ii) that deflationism should restrict itself to a claim about the function of the truth predicate. In light of many of the arguments in this volume, these claims are highly questionable. The first one appears to be motivated by the thought that no property analysis in general is to be had, and so deflationism is the natural position vis-à-vis truth; consequently, deflationism ought to spurn the very idea of property metaphysics, and so it cannot be impugned or ‘deconstructed’ on such a basis. This stand might well serve the deflationist against those who would assume a correspondence conception, but it goes no way to answer the fundamental problem under consideration, for being an anti-deflationist in no sense depends on a commitment to an analysis of truth. The first anti-deflationist - Frege - explicitly rejected such a commitment, as do Wiggins, Dummett, Davidson, Rosenberg, Misak, McGinn…. Indeed, even correspondence theorists such as Alston and Devitt are careful not simply to assume their proprietary conception. Williams’s second claim is also proposed to side-step a Boghossian style deconstruction and is correlatively spurious: if deflationism is just a claim about the ‘function’ of the truth predicate, then, while it might well be the default position, it is only so in virtue of being acceptable to all. What is required is some grounding of the function that is suitably deflationary. Many of the authors in this volume have detailed arguments saying that no such grounding is available.

Notwithstanding these brisk remarks, it might be that deflationism can discharge the above burdens. However the debate turns out, the present collection should be welcomed by all.

Notes