Figure 1: A discourse status quiz.
Department of Philosophy
University of Sydney
Australia 2006
Hume is the St. Francis of modern metaphysics, the patron saint of ontological ascetics. The devout Humean takes the pinnacle of metaphysical virtue to be a world in which the only facts are the mundane first order physical facts about how things actually are--"a vast mosaic of local matters of particular fact, just one little thing and then another," as David Lewis puts it.[1] Hume's own ontological austerity derives in part from his empiricism, in part from the great medieval Fransciscan metaphysician, William of Ockham, and in part no doubt from canny inclinations of his native land. Whatever its original sources, however, the doctrine remains enormously influential. The Humean ideal continues to exert great force in contemporary philosophy.
However, like prosperous Franciscans many metaphysicians who pay lip service to the virtues of ontological economy have strayed from the true path. The Hume world has frequently been judged too cramped. Some lapsed Humeans extend the bare Humean structure to fit in modal facts; some extend it to fit in subjective experience or intentional mental states; others to fit in moral facts; and so on. The common theme is that a respectable metaphysics simply cannot survive on the bare regime that Hume prescribes.
Now the galling thing about a prosperous Franciscan is not the fact that he is not suffering the discomforts of poverty, or not this as such. It is his moral inconsistency, the fact that he claims virtue while failing to live according to his own professed conception of the virtuous life. There are two ways to challenge this inconsistency. One is to demonstrate to the Franciscan that he could make do with less, and hence establish that by his own lights he is not leading the pious life. This is in effect the strategy of those who dispute the need for particular extensions to the Hume world, arguing that the purpose such extensions serve can be met in some less extravagant manner.[2] The alternative deflationary strategy is more subtle. It is to accept the case for the extensions, for the Franciscan's additional consumption, but to point out that the consequence of accepting this is to undermine the minimalist conception of a virtuous lifestyle. In accepting that the Franciscan is fully entitled to satisfaction of the normal human appetites, in other words, we undercut his claim to abnormal piety.
It may not be obvious that there is any analogous move in the metaphysical case. In this paper I want to show that there is, and to suggest that this has profound consequences for our understanding of a range of contemporary metaphysical debates. Roughly, its effect is to undercut the distinction between various non-Humean forms of metaphysical realism and something akin to a Wittgensteinian linguistic pluralism. To the extent the distinction can be drawn, moreover, the latter is the default position. So not only is this form of pluralism an important and widely neglected option in a range of contemporary metaphysical debates; it actually has claim to be the pre-eminent option. In a sense I shall explain, it is the philosophical geodesic, the course from which no one is entitled to depart without good reason. The paper thus presents a challenge to the lapsed Humeans of contemporary metaphysics: Embrace worldly pluralism, or return to the pure faith, for there is no virtuous middle way!
The paper is in five main sections. Section I identifies our target species of pluralism, and distinguishes it from a more common species. In Section II I consider this target species in relation to a variety of anti-pluralist rivals, including in particular the monist doctrine which seems to be the intended position of the lapsed Humeans; and I draw attention to the central issue of the paper, namely as to whether there is actually a tenable distinction between this monist position and our target pluralism. Section III argues that despite the avowedly ontological nature of their concerns, these lapsed Humeans cannot avoid reliance on a certain semantic distinction--essentially, the distinction between descriptive and non-descriptive uses of language--and points out that this puts them at at least a prima facie disadvantage compared to the pluralists, who need no such distinction. Section IV then outlines a case for thinking that the required distinction cannot be drawn, and draws attention to some connections between the resulting sort of pluralism and certain other recent approaches to the same metaphysical topics. And in Section V I illustrate the advantages and character of such a pluralism with reference to the issue between realists and their instrumentalist critics; perhaps surprisingly, this rather Wittgensteinian pluralism turns out to provide an especially secure brand of realism.
One familiar kind of philosophical pluralism is exemplified by Quine's brand of ontological relativity, and perhaps in a different way by other forms of scientific relativism. Here the plurality consists in the possible existence of a range of alternative scientific world-views, each empirically adequate to more or less the same degree, and none, even in principle, having privileged claim to provide a "truer" description of the world. This form of pluralism itself comes in several varieties or sub-species, of course. The distinction between Quine's variety and Kuhnian scientific relativism may well be a deep one, for example.[3] However, for present purposes what matters is something that these views have in common, namely that the plurality they admit involves a range of different ways of doing the same kind of thing, of performing the same linguistic task. There may be many equally valid possible scientific world-views, but all of them are scientific world-views, and in that sense are on the same level of linguistic activity. In other words, this is what might appropriately be called horizontal pluralism.
Horizontal pluralism is far from confined to this scientific level. It is a product of certain kinds of relativism, and so may be found wherever these flourish. In ethics, for example, it is the familiar thesis that there is a range of equally coherent moral viewpoints, none objectively superior to any other. Why is this a case of horizontal pluralism? Again, because the plurality it envisages is confined to a single linguistic plane or level. Different moral systems are all nevertheless moral systems. They have something in common in virtue of which they may be counted to be different ways of performing the same linguistic task. It may be a very nice question how this something in common is to be properly characterised, but it must have an answer, if relativism is not to degenerate into the trivial point that the same words may mean different things for different people.
If these are cases of horizontal pluralism, what would be a vertical pluralism? It would be the view that philosophy should recognise an irreducible plurality of kinds of discourse--the moral as well as the scientific, for example. This is the species of pluralism with which we are going to be most concerned. (I shall mainly use the term discourse pluralism.) I want to show that it provides a natural if not initially a congenial home for the lapsed Humean.
The plurality of forms of discourse, or "language-games," is a prominent theme in the later Wittgenstein. This remark is typical: "We remain unconscious of the prodigious diversity of all the everyday language-games because the clothing of our language makes everything alike."[4] There also seems to be a strong element of discourse pluralism in the American pragmatist tradition, of which Goodman and Rorty are the most prominent recent representatives. True, Goodman and Rorty's pluralism is not exclusively vertical. It also contains a strong horizontal or relativistic element. However, the fact that the vertical component is significant in its own right is manifest in the contrast between the positions of Goodman or Rorty on the one hand, and Quine on the other. Quine at times seems close to a purely horizontal pluralism, to the view that all factual discourse is either eliminable or reducible to physical discourse. This view acknowledges that there might be alternative physical discourses--"alternative physical theories, insusceptible to adjudication," as Quine puts it[5]--but says that there is nothing else at any other level. And certainly Quine rejects Goodman's proposal to admit on an equal footing a multiplicity of further "world versions," such as those of art and music, saying that in his view "this sequence of worlds or versions founders in absurdity."[6]
More accurately, however, Quine himself is non-physicalist to the extent of accepting the existence of certain abstract objects, such as classes and numbers. This is not to say that he intends to be a discourse pluralist. On the contrary, he is critical of the proposal to regard the acceptance of such objects as in some sense radically unlike the acceptance of physical objects:
There are philosophers who stoutly maintain that 'true' said of logical or mathematical laws and 'true' said of weather predictions or suspects' confessions are two uses of an ambiguous term 'true'. There are philosophers who stoutly maintain that 'exists' said of numbers, classes and the like and 'exists' said of material objects are two uses of an ambiguous term 'exists'. What mainly baffles me is the stoutness of their maintenance. What can they possibly count as evidence? Why not view 'true' as unambiguous but very general, and recognize the difference between true logical laws and true confessions as a difference merely between logical laws and confessions? And correspondingly for existence?[7]
Later I want to argue that this point backfires. Discourse pluralism is the default position in this debate, and doesn't need to be defended by appealing to a claimed ambiguity in 'true' or 'exists'. On the contrary, it is the discourse pluralist's opponent who needs to appeal to some suitably "thick" or substantial unity in these notions--and Quine's preferred thin notions are not up to the task. Hence I want to suggest that Quine himself might best be recast as a discourse pluralist about abstract objects. He thus exemplifies what I shall argue to be the predicament of many would-be non-Humean realists in contemporary philosophy. (The striking thing about Quine's case is that his predicament stems from his own minimalism about truth and existence.)
The first task is to mark off discourse pluralism from a range of other ways of dealing with the same philosophical topics.
As a distinctive philosophical doctrine, discourse pluralism is perhaps best understood in terms of what it denies, in terms of the contrasts it tries to establish with other philosophical treatments of problematic topics. In this section I describe four such contrasts. I distinguish a pluralist treatment of a philosophical topic from four other approaches: reductionism, two forms of irrealism, and lastly a view I call additive monism, which is the intended position of non-Humean metaphysicians. (Later our main interest will be in the question whether an additive monist can resist reconstitution as a discourse pluralist.)
To establish the general framework, consider a problematic topic against the background of an unproblematic topic--morality against the background of natural science, say. Assume for simplicity that the status of the background topic is not at issue. Our first group of distinctions rest on different combinations of answers to two questions about the status of the problematic topic. First question: Is its (purported) subject matter distinct from that of the background discourse? Second question: Is the problematic topic fully legitimate, again by whatever standards may properly be applied to the background discourse? (What is meant by 'legitimate' here will become clearer as we proceed.) In terms of our example then: Is the subject matter of moral discourse distinct from that of natural science?[8] And is it a fully legitimate discourse, or does it in some way fall short?
Figure 1: A discourse status quiz.
Discourse pluralism answers "Yes" to both questions, and hence may be distinguished from views which answer "No" to one or other (and also from the less interesting view, which I won't mention further, which answers "No" to both). The first contrast is therefore with reductionism, which concedes autonomy in order to save legitimacy. (We are at the first level in the tree shown in Figure 1.) The reductionist agrees with the discourse pluralist about the legitimacy of moral discourse, but seeks to secure it by denying the ontological distinctness of its subject matter. On the contrary, the reductionist argues, morality is a branch of natural science--a fact which is obscured, in practice, simply by our usual ignorance of the appropriate identities between moral facts and natural facts. For the reductionist moral discourse is legitimate because it actually is natural discourse, albeit disguised natural discourse.[9] A pluralist wants both autonomy and legitimacy, whereas a reductionist is prepared to sacrifice the former in the interests of the latter. (The reductionist may not view it as a sacrifice, of course.)
Our next contrast is with two views which make the alternative sacrifice, conceding ground on legitimacy in order to preserve autonomy. They concede ground in very different ways, and we shall need to treat them separately. However, the fact that they both do so in some sense is a reason to deal with them under a common heading.
Irrealism thus comes in two main forms. The first is eliminativism, the approach famously exemplified in Mackie's error theory of moral discourse.[10] In effect, the eliminativist holds that a problematic area of discourse tries to be legitimate, as well as autonomous; but that it fails to connect with anything in reality, and so is systematically false. The problematic topic is thus legitimate in intent, but fails in the execution of that intent. Mackie notwithstanding, eliminativism has not been particularly popular among moral philosophers. There are more familiar cases elsewhere, however. It has recently become a prominent approach to intentional psychology, for example. And Quine has long been seen as an eliminativist about meaning (and hence at least in some respects about psychology).
The second form of irrealism--non-factualism, as I shall call it--is much more popular. It is the approach which preserves the autonomy of disputed topics by giving ground on their factual character. So whereas the eliminativist holds that moral discourse tries to be factual, but fails to connect, the non-factualist denies that moral claims are ever intended to be factual claims. They don't fail to connect, for their linguistic role is not to attempt to connect. What it is instead is a matter of some interest, of course. In the moral case the traditional options are the expressive or emotivist view, namely that moral judgements express certain sorts of evaluative psychological attitudes; and the prescriptivist view, which treats a moral assertion as something approaching a command.
Non-factualism is a very common doctrine. Apart from the moral case, the approach is also well known in application to aesthetic judgements, mathematical statements, theoretical sentences in science, secondary qualities, knowledge claims, psychology more generally, meaning ascriptions, indicative and subjunctive conditionals, and probabilistic, causal and modal judgements. (No doubt I have missed some.) Not always well appreciated are some of the basic philosophical commitments of any such position. It is clear that such a view is committed to the existence of a significant distinction between factual and non-factual uses of language. Perhaps not so obvious is the obligation that the non-factualist incurs as a result of this, namely to account for the fact that non-factual uses of language can present themselves as factual uses, at least superficially.
The contemporary philosopher who has done most to acknowledge and to try to meet this obligation is Simon Blackburn.[11] Drawing on Hume's idea that we project our attitudes and prejudices onto the world, and so see it as populated by seeming facts of our own construction, Blackburn has argued that the non-factualist can explain our conversing as if there really were such facts. He has tried to show that the projectivist is entitled to a notion of truth, and to the other trappings of a realist linguistic practice. Thus projectivism supports quasi-realism, as Blackburn calls it. Couched in these terms, the non-factualist's concessions concerning the legitimacy of a disputed topic are now very muted. To be sure, moral discourse (or whatever) is not really factual; but it has and is entitled to all the trappings of factuality, all the appearances of legitimacy. So although on this view there is only one real world, only one realm of genuine facts, our language quite properly works as if there were many.[12] If this is not already discourse pluralism, then what separates it is the availability of a substantial distinction between factual and non-factual uses of language. If that goes, non-factualism is no longer a distinct alternative to discourse pluralism.
So much for the distinction between discourse pluralism and the various views which differ from it in denying either the autonomy or the legitimacy of a problematic area of discourse. We now turn to a contrast not marked by a disagreement on either of these issues, namely the distinction between discourse pluralism and the doctrine I call additive monism. Additive monism agrees with discourse pluralism in rejecting reductionism and both forms of irrealism, accepting that multiple domains of discourse may each be autonomous and yet fully legitimate. The disagreement is only about how these separate domains are to be construed. The additive monist regards them as sub-domains of a single universe of facts--not a single physical universe, presumably, for that would be to concede the game to a physicalist reductionism, but single in some sense. What distinguishes additive monism from discourse pluralism is the claim that there is something that unifies the various autonomous discourses; what distinguishes it from reductive physicalism, the claim that whatever this unifying principle is, it is not that all facts are ultimately physical facts.
It is natural to wonder whether there a real distinction between discourse pluralism and additive monism, or whether the apparent issue rests on an empty metaphor. And if there isn't a real distinction, how is the single resulting position best described? These neglected questions are of enormous importance to a wide range of current philosophical debates. For additive monism seems the intended home of the lapsed Humeans with whom we began--of contemporary non-Humean approaches to such topics as diverse as causation, modality, the mental, abstract objects, morality, and so on. On the face of it, these accounts argue that there are more kinds of facts in the world than the Hume world admits--facts about causal relations, over and above facts about constant conjunction, for example. The proponents of these views do not think of themselves as Wittgensteinian pluralists, of course. But it is not clear that there is conceptual space for any other position. The additive monist needs to show that something unifies the various autonomous discourses. Failing this, pluralism is the default option--what the would-be monist must fall back to if the distinction cannot be maintained. Thus I want to argue that it is much harder not to be a Wittgensteinian pluralist than many people have assumed. Since a pluralism of this kind explicitly abandons Humean metaphysical minimalism, its effect here is deflationary--it undermines the Humean conception of metaphysical virtue.
The issue turns on the availability of a unifying principle. What might the additive monist appeal to at this point?
A first thought might be that supervenience will do the trick. Perhaps additive monism corresponds to the view that a problematic topic is supervenient on a non-problematic background (though not reducible to it); and discourse pluralism to the more extreme version of autonomy involved in denying both reducibility and supervenience between discourses. However, a little reflection shows that this won't do the job. There are counterexamples in both directions. Non-Humean metaphysicians commonly deny that facts about laws, causation and the like are supervenient on first order matters of particular fact.[13] Conversely, it is far from clear that a pluralist need deny supervenience. A nice counterexample may be extracted from Blackburn's discussion of moral supervenience.[14] Blackburn argues that his projectivist quasi-realism not only accounts for moral supervenience, but is especially well adapted to doing so. It is true that quasi-realism is not a form of discourse pluralism. But I think it differs mainly in assuming that some stronger form of realism is appropriate elsewhere, for (at least some) non-moral topics. So we would have a form of discourse pluralism if we abandoned this assumption, and thus extended quasi-realism "all the way down." At the same time we would still be entitled to Blackburn's account of moral supervenience, for this rests simply on the idea that moral discourse projects evaluative attitudes. Hence it seems that supervenience is independent of the issue between discourse pluralism and additive monism.
With supervenience out of the way, the next suggestion may be that additive monism is primarily a metaphysical doctrine, whereas discourse pluralism is primarily linguistic. The monist maintains that the autonomous domains of facts are all part of a single universe or metaphysical totality; whereas the pluralist seems to be denying the unity of factual language, opting instead for the view that a variety of different uses of language share certain superficial characteristics--Wittgenstein's common clothing of the diversity of language-games. (What are these characteristics? More on this in a moment.) But although this certainly distinguishes the two doctrines in terms of their natural characterisations, it does not exclude the possibility that they come to the same thing. After all, the pluralist may well take the view that metaphysics itself is a manifestation of the superficial clothing of language, and derives its apparent unity from this very source. Thus if metaphysics is essentially a concern with ontology, with "what there is," and this issue is seen in Quinean terms, then metaphysics and grammar become inseparable. Ontology is a matter of quantification, and this is precisely the sort of linguisitc feature that the pluralist might take to be an element of the superficial clothing of language, serving similar grammatical purposes in discourses whose underlying functions in language are widely varied. (Analogous remarks would apply to truth.)
Monists might object at this point that they have a more substantial ontological principle in mind, and that their dispute with discourse pluralism turns largely on the question of our entitlement to some such notion. I think that this is on the right lines, but that it leaves out an important part of the story. The dispute cannot be entirely ontological, but must have a semantic or linguistic dimension. For the metaphysical view requires that there be a non-trivial sense in which the various discourses to which it applies are all serving the same linguistic function. It requires that they are all descriptive in the same sense, that they share a common goal of truth, a common concern to portray the facts as they are. It is not enough for the monist that there be a unified world out there; it is also crucial that within each disputed part of language, statements stand in the same relation to the relevant part of the single world. Otherwise monism is trivial: it is easy to find a unified world to which every use of language relates in some sense. The monist requires that it always be the same sense. So monism needs a common semantic theory, as well as a unified world.
The monist might simply begin at this semantic level, of course, professing a concern with truth, the facts, or some such thing (again claiming to construe this semantic notion in a substantial way, incompatible with the pluralist's suggestion that these concepts belong to the superficial common layer of language). However, my point is that even if the commitment does not begin at this semantic level, it must get there in the end. In effect, the monist needs a distinction between factual discourse--the kind that does or is intended to describe a part of a metaphysical totality--and non-factual discourse, the kind that is not so intended.[15]
This obligation may seem to fall as much on the discourse pluralist as on the additive monist. However, I think there is an important asymmetry--the onus here lies predominantly with the monist. For diversity is obvious, being guaranteed by difference of subject matter (once reductionism is rejected). It is the monist's unity that calls for substantial argument. So not only does additive monism turn out to involve a linguistic doctrine after all, its monism requiring or entailing a linguistic characterisation; but it also turns out to be a significantly more committed linguistic doctrine than discourse pluralism. The monist must hold that the unity of factual discourse is more than skin deep (or clothing deep, in Wittgenstein's metaphor). Whereas the pluralist can be content with an easy-going linguistic "multifunctionalism"--with the view that under their common syntactic skin the roles of the various discourses in our lives may be as different as say those of our various internal organs.[16]
Additive monists thus incur a significant debt in the coin of semantic theory: they owe us a substantial characterisation of factual discourse. (We have already noted that the same is true of non-factualism.) I want to exploit this point to argue for the attractions of discourse pluralism over additive monism, or for that matter over its other non-reductionist rivals. In effect, I want to suggest that the debt thus incurred turns out to be impossible to repay in any of the acceptable currencies. We could accept payment in the debt's own currency, accepting the fact/non-fact distinction as an uncashed primitive. However, ordinary scientific economy recommends instead that we give preference to a view which needs no such primitive. Pluralism and perhaps a physicalist reductionism emerge as the economical alternatives.[17]
This line of argument depends on a claim that many readers may find surprising, namely that there isn't a readily available and well founded distinction between factual and non-factual uses of language. I have defended this claim at length elsewhere,[18] and in the next section I sketch some of the main points of that argument, and some related advantages of discourse pluralism. However I think the claim should really not be surprising to a contemporary philosophical audience, for in a sense it is a natural consequence of Quinean scepticism about meaning. Factuality is a semantic property, if anything. It is the property of "statementhood," or that property which is shared by all assertoric utterances. Thus it is a particular component of meaning, and Quine's general scepticism about meaning might already have led us to recognise the possibility that there is no such property, in any substantial sense.[19]
I have said that discourse pluralism and additive monism are distinguished by the fact that the former makes no appeal to--indeed, explicitly rejects--the assumption that there is a more than superficial category of fact-stating or descriptive uses of language. It may seem perverse to count this a point in the pluralist's favour. For isn't it obvious what such a category consists in? Fact-stating utterances are just those that have genuine truth-conditions, those that have assertoric force, those that express genuine beliefs about that world. Surely these notions are unproblematic?
The pluralist's response to this challenge is to draw attention to the possibility that these common sense intuitions might rely on nothing more than the superficial features of language, the clothing that covers a more fundamental diversity. Are the notions of truth, fact, assertion, belief, and so on foundational categories, inevitably central to any theoretical account of our use of language? Or are they mere products of language, categories thrown up by language itself, and not therefore presupposed by a proper explanatory theory of language? As a first step, the pluralist asks us simply to acknowledge that the latter answer is at least a conceptual possibility--that common sense is powerless to exclude it. Thus an appeal to common sense is ruled out by the consideration that common usage is the very object of theoretical enquiry here. All sides are agreed that ordinary usage exhibits a superficial unity between the discourses in question. The issue is whether this superficial unity is more than skin deep.
Perhaps surprisingly, it is the pluralist who has the better of the naturalistic high ground in this dispute. The issue concerns the structure of a complete naturalistic theory of language use. Roughly, it is the question whether such a naturalistic theory reveals a structural feature of language that may be regarded as constituting a fact-stating/non-fact-stating distinction. This doesn't mean that the pluralist must concede that all linguistic notions are ultimately naturalistic (which would be to embrace a form of reductionism that the pluralist is in general keen to avoid). The pluralist simply wants us to recognise the possibility that what can be said about language in naturalistic terms might reveal no trace of a substantial factual/non-factual distinction. As a naturalistic question, this not the sort of thing that can be decided a priori. So discourse pluralism is ultimately an empirical doctrine, albeit a highly theoretical one, and one that concerns the linguistic part of the natural world.
The onus lies with the additive monist to show not only that there is a substantial property of factuality underlying ordinary usage, but also that it has a particular distribution. The monist wants to show that the disputed discourses are unified by a common concern to "depict the world," or some such. The discourse pluralist is not denying that scientific utterances and moral utterances are alike in being meaningful speech acts, but simply that there is any more substantial sense in which they are both "statements of fact," or "descriptions of how things are." So a plausible strategy for the pluralist is to try argue that none of the usual attempts to characterise such a property is capable of settling the issue of distribution; or in particular of excluding the trivialising possibility that all discourse is factual discourse--i.e., that statementhood is a property guaranteed to be possessed by any utterance whatsoever. If a proposed analysis of statementhood cannot exclude even this possibility, it is unlikely to convince us that the monist has a defensible claim concerning the factual status of the discourses of interest.
This sceptical attempt to trivialise any proposed analysis of the notion of statementhood is the strategy I adopted in Facts and the Function of Truth. To illustrate, the most common suggestion is that factual discourse is distinguished by its relation to the notion of truth. Genuinely factual utterances are said to be those that are genuinely truth-bearing, or genuinely "aimed at truth," or some such. As several writers have pointed out, however, the notion of truth involved had better be capable of doing the work. One notion which obviously won't do the work is a redundancy or disquotational notion, which depends on nothing more than the standard equivalence principle: the fact that for any sentence P, 'P' and 'P is true' seem in some strong sense to say the same thing.[20] So if truth is to be invoked at this point, we need what Bernard Williams called a substantial theory of truth.[21] I argue in Facts and the Function of Truth that there is no such theory to be had. The contenders inevitably fail on one of two counts: either they are like the disquotational theory, in applying equally well to any sort of discourse; or they already assume a fact-stating/non-fact-stating distinction, in some other form.
One such alternative relies on the psychological distinction between beliefs and other kinds of propositional attitudes. This proposal originates in the kind of emotivism which grounds itself on Humean moral psychology, drawing a contrast between descriptive or belief-expressing language, on the one hand, and expressive or evaluative attitude-expressing language, on the other. The emotivist treatment of moral discourse has provided a model for parallel treatments of other topics. In the process it has given rise, at least implicitly, to a general theory of the nature of the distinction between factual and non-factual language: to the view that factual language expresses belief, whereas non-factual language expresses attitudes of other kinds. The trouble with this theory is that there is very little prospect of drawing the required psychological distinction, in advance of some other solution to the problem it is supposed to address.[22] We need a substantial notion of belief, but we won't get that until we have a substantial way of drawing the same distinction somewhere else.
The sceptical attempt to drive a conventional doctrine around a logical circle has a famous paradigm in modern philosophy, in Quine's own attack on the analytic/synthetic distinction. And as we have noted, the conclusion itself seems one that Quine might well have endorsed. Quine's irrealism about meaning seems very much at home with the view that there is no substantial unity to the factual, descriptive, or truth-bearing part of language as a whole; no single such semantic category, in any substantial sense. Indeed, this view might be seen as simply doing for semantics what Quine himself does for ontology: insisting that philosophy has no privileged second order vantage point, but must rather make do with the deliverances of the best first order theories, taken at face value.
At any rate, the effect of these sceptical arguments is to underline the economical advantages of discourse pluralism, by showing that a useful characterisation of factual discourse is not as readily to hand as we have tended to assume. But it might be objected that pluralism incurs costs of its own. For one thing, it owes us an account of what it regards as the superficial unity of language--for example, of the fact that all discourses apparently avail themselves of the same indicative syntax, and of the notion of truth. For another thing, we are entitled to be told in what the underlying diversity is supposed to consist--in what dimension the different discourses enjoy their vertical separations. Related to the latter point is the important question as to how we set the limits to pluralistic tolerance. Clearly there are some conceptual conflicts that call for a less tolerant response, for simply discarding one of the alternatives. The obvious thought is that this is appropriate when the theories concerned are attempting to occupy the same linguistic role, or level; but what does this mean, in general?
These are real obligations for a discourse pluralist. I think they should not be unwelcome, however. On the contrary, they provide a further opportunity for pluralism to display its comparative creditworthiness, by showing that it has the means to meet such conceptual debts. In Facts and the Function of Truth I tried to provide a suitable account of truth for these purposes--an account that has the potential to explain why diverse uses of language should be alike in being treated as "truth bearing." The basic idea is that despite the diversity of uses of language, different uses tend to be alike in this respect: within each use or discourse, there is some potential utility in noting and resolving disagreements between the members of a speech community. A notion of truth encourages a community to realise this potential. It does so by generating a social value which is negative when speakers disagree and positive when they agree. Disagreement thus becomes socially unstable, and argument, with its long-run benefits, is thereby encouraged.
This approach does not require that argument is always beneficial, of course. Like any evolutionary theory, it keeps its eyes on the long run. Nor, importantly for a pluralist, does it require that the kind of benefit always be the same. The pluralist can thus allow on the one hand that there is no uniform way in which each of the many language-games affects our well-being; and yet on the other that in all or most such games there is some benefit in argument, and hence in the availability of a notion of truth. In fact, I think the pluralist can do considerably more than this. For it turns out that our use of the notion of truth is not strictly uniform across the range of different domains of discourse. There are significant differences, underlying the predominant pattern. I think that pluralism can do much to account for these variations, by appealing to the different functions or linguistic roles of the different domains concerned. And only pluralism has the flexibility to do this: other approaches are tied too rigidly to the existence of a fact-stating/non-fact-stating dichotomy in language.
What are these discourse to discourse variations? They turn on the differential tendency of arguments to be resolved in such a way that neither participant is judged to have been initially mistaken--for example, the kind of thing that may happen when a disagreement about a probability turns out to have stemmed from the fact that the speakers concerned had access to different evidence. This kind of outcome to an argument manifests itself as a sudden reluctance to apply the notions of truth and falsity to the judgements concerned. Such no-fault disagreements occur for different reasons in different parts of language. To a surprising extent, however, it seems to be possible to explain where they do occur in terms of an understanding of the distinctive functional role of the type of discourse concerned, coupled with the general principle that the origins of truth and falsity lie in their role in encouraging useful argument. (In the process, incidentally, it is possible to explain many of the intuitions that have motivated non-factualism--but to do so without the non-factualist's problematic division of language into fact-stating and non-fact-stating categories.)
This approach has affinities with recent work of Crispin Wright's on what might be called the "fine structure" of truth.[23] Wright has distinguished a number of different components that may, but need not all, be characteristic of the use of truth in association with a particular area of discourse. In effect, he suggests that we may classify a subject matter according to which of the set of these characteristics we take its notion of truth to involve. As I understand it, his concern is mainly descriptive and taxonomic. However, I think there is some prospect that the structure thus discerned will turn out to be explicable as sketched above, in terms of some general account of the function of truth in language. Whether the best general account will be in the terms I outlined earlier remains to be seen. But if it is to be in keeping with Wright's program, I think it will have to be like my account in being explanatory rather than analytic: its focus will be not on the question "What is truth?," but on the question "Why do ordinary speakers have such a notion as truth?"
Interestingly, Wright too takes this stance on truth to support a non-committal metaphysics, at least as the natural fallback position. He characterises this position as an anti-realist one, in Dummett's sense, on the grounds that the minimal notion of truth it requires can be thought of as derived from assertibility. I disagree at this point, arguing that in practice the minimal notion of truth is already realist in this sense, for reasons to do with the origins of negation.[24] That issue aside, I take the following remark of Wright's to be in much the same vein as my emphasis on the economical advantages of discourse pluralism: "Anti-realism thus becomes the natural, initial position in any debate. It is the position from which we have to be shown that we ought to move. All the onus, everywhere, is on the realist."[25] Substitute 'pluralist' for 'anti-realist' and 'non-pluralist' for 'realist', and these are my sentiments exactly.
The project also has affinities with Blackburn's quasi-realism. The quasi-realist about moral discourse (for example) wants to argue that although moral judgements are not really factual, they are entitled to the trappings of factuality, including a respectable notion of truth. So although there is only one domain of genuine facts, it is quite proper on this view for ordinary usage (and truth in particular) to work as if there were many. As we noted earlier, what separates this from discourse pluralism is the quasi-realist's assumption that his project is a limited one, bounded in its application by the availability of a substantial distinction between factual and non-factual uses of language. If that distinction lapses, quasi-realism is no longer a distinct alternative to discourse pluralism. Discourse pluralism is again the default position. But so long as a more limited quasi-realism remains a live project, the quasi-realist's interest in explaining why certain discourses should usefully employ an "artificial" notion of truth will apparently coincide with the project described above.
In sum, there are considerable grounds for doubt as to whether a useful principle of semantic unity is available to an additive monist, other than as an ad hoc and costly theoretical primitive. And there are considerable grounds for optimism concerning the project to explain what superficial unity we find in language (and the limits of this unity) on foundations that only the discourse pluralist finds congenial. The case is not closed, of course, but it is the additive monist who is on the defensive.
At this point it may be helpful to emphasise that discourse pluralism is not an irrealist position. The pluralist accepts with all sincerity that there are moral states of affairs, possible worlds, numbers, or whatever. What he or she rejects is the additive monist's attempt to put a further metaphysical gloss on such existential claims. (That gloss turns out to depend on a semantic distinction of questionable standing, or so the pluralist argues.) Without the gloss, discourse pluralism sits quite happily with a non-metaphysical or "minimal" realism.[26]
Perhaps more surprisingly, this species of realism turns out to be particularly well protected against one of realism's traditional opponents. Discourse pluralism has a defence against an instrumentalist challenge, of a kind which continues to be pressed against realists in a number of branches of contemporary philosophy. This defence simply not available to the monist, for as we shall see, the antidote the pluralist is able to employ against instrumentalism works equally well against additive monism.
Instrumentalists prey on the Humean tendencies of the modern philosophical public, offering a simple path to ontological asceticism. To avoid the realist's ontological excesses, they tell us, we need simply treat our talk of such ontology as a convenient instrument. The talk works just as well when construed this way, and yet in ontological terms it is absolutely free. Thus vast ontological savings compared to realism, and at no extra cost elsewhere.
Of course, realists often respond that the offer is too good to be true. There is a cost elsewhere, albeit one that may not be obvious at first sight. What we lose under instrumentalism, the realist claims, is the ability to explain why the instrument works. The need for such an explanation thus becomes a debt that the instrumentalist cannot pay off.
This response is a good deal less effective in some cases than in others, however. Whatever its merits as an objection to instrumentalism about theoretical physics, it is likely to be of little use in defence of realism about possible worlds, for example. Here the realist will typically acknowledge that a proper explanation of why talk of non-actual worlds is useful can only appeal to the actual world--i.e., in effect, to acknowledge that such talk would be just as useful if there were no other worlds. Similar considerations might apply in the case of intentional psychology. Realists might well grant that an adequate explanation of why talk of beliefs and desires is useful could in principle be cast in terms of scientific theories making no reference to intentional states.
In cases such as this the realist needs a different strategy, and might be tempted to appeal to Quine. For has Quine not taught us that there is no more to the question whether there are possible worlds (say) than the matter as to whether such things are quantified over in the most serviceable philosophical theory of modality and related topics? And given that the instrumentalist does not deny us our talk of possible worlds (quantification included), don't we therefore have Quine's authority for rejecting the instrumentalist's claim that there are no such things? Doesn't the talk itself provide all the warrant we ever have for an existential claim?
However, it is important to appreciate that there are two ways to interpret this argument, resting on two quite different ways of interpreting Quine's views on ontology. On one reading the Quinean doctrine is effectively a principle of ontological quietism--the principle that there is no separate second order science of ontology, but simply the mundane business of existential quantification carried out by first order specialists in the course of their working lives. I shall come back to this: I think it does provide an objection to instrumentalism, but not the objection the metaphysical realist was hoping for. There seems to be a tendency to read the Quinean doctrine in quite a different way, however. Under this second reading the activities of first order specialists provide raw data to which the Quinean principle is then applied, yielding second order ontological conclusions. In other words, the fact that first order specialists quantify over the entities of a certain kind is held to constitute evidence that such entities actually exist. Far from dismissing the science of ontology, Quine's doctrine thus becomes the main instrument in the working ontologist's tool kit--a kind of "magic eye" for detecting otherwise invisible existents.
Had Quine provided us with such a magic eye it would indeed provide a powerful weapon against instrumentalists (or at least those instrumentalists not prepared to take issue with Quine). "See for yourself," the realist would say, inviting the instrumentalist to peer at reality through Quine's lens. But I think that to make this explicit is to make it obvious that Quine intends no such thing. The right reading of Quine is the quietist one.[27] Now this reading too counts against the instrumentalist. The instrumentalist requires that not all good theories involve ontological commitment (via their existential quantification); for it depends on whether the theory in question is genuinely descriptive or merely an instrument. As a quietist principle Quine's doctrine contends that there is no such further issue.
But does this favour realism? Yes, but only a realism of the discourse pluralist's minimal kind. For what the quietist principle denies us is something that is needed by additive monists, as well as by instrumentalists. Instrumentalism is a form of non-factualism. Like additive monism, though for a different reason, it thus depends on a substantial distinction between factual and non-factual uses of language. The monist needs an account of what keeps the domains of discourse together; the non-factualist an account of what sets them apart. The instrumentalist thus owes us an account of what is lacked by those domains of discourse that are judged "merely" instrumental, and hence of what is not lacked by those that pass the test. Without such an account, for one thing, there seems nothing to stop the above kind of argument for instrumentalism from going all the way down. All discourse would thus be construed as fictional discourse, and the contrast which gives point to instrumentalism would be lost. Modal and moral discourse would be on the same footing as discourse of any other kind, just as the pluralist and the additive monist both contend.
So the instrumentalist needs a substantial factual/non-factual distinction. Often this is supposed to be cashed in ontological terms. The instrumentalist is presented as denying the existence of a class of entities of a certain kind--possible worlds, or mathematical objects, for example. However, we have just encountered a reason for doubting that this can serve as the fundamental distinction. Construed à la Quine, ontological commitment is a more or less trivial consequence of serious theoretical commitment; even if there is always room for argument as to which ontology strikes the best ideological bargain. Hence instrumentalism cannot be primarily an ontological doctrine. The instrumentalist needs a prior distinction between full-blooded and merely fictional theoretical commitment. An ontological distinction would flow from such a distinction, but cannot provide it; not, at least, unless we are prepared to discard the Quinean principle that our only guide to ontology is the quantificational structure of accepted theory.
All of this applies equally to the additive monist. Quine's ontological quietism prevents the monist from casting the unifying principle he or she requires in ontological terms.[28] However, there is nothing to prevent a discourse pluralist from embracing the Quinean doctrine; the pluralist too will be interested in the most advantageous theoretical formulation of any given level of discourse, and may well see the choice in Quinean terms. The pluralist thus has an objection to instrumentalism which would be suicidal in the hands of an additive monist.
Discourse pluralism thus provides a particularly defensible form of realism--a position uniquely placed to exploit some of the hidden costs of irrealism. A corollary is that those who find themselves in the grip of realist intuitions about matters which are otherwise vulnerable to instrumentalist attack would do well to take pluralism seriously, for it may well be the only realism that they can have.
We began with two ways to respond to the moral inconsistency of a prosperous Franciscan. One response was simply to point out that he could well make do with less. I compared this to challenging the lapsed Humeans' case for particular metaphysical luxuries. This is what non-factualists do, in effect, in arguing that in order to make sense of say modal discourse, we don't need to assume a realm of modal facts; instead we may read the discourse non-descriptively, so that its interpretation does not require such facts. My main goal in this paper has been to offer an alternative deflationary challenge to non-Humean metaphysics. This compares to granting that the Franciscan has the same range of human needs as anyone else, and the same right to reasonable satisfaction of those needs; but insisting that in granting this, we overturn the idea that there is any special virtue in self-imposed poverty.
Discourse pluralism thus allows the lapsed Humeans their talk of possible worlds, modal facts or whatever, subject only to the modified Quinean requirement that such an ontology provides the most economical basis for the particular discourse in question (i.e., roughly, the basis that best enables that discourse to serve its intended function in human life.) What it withholds is the concession that such talk maps the bare objective structure of a single independent reality. On the contrary, we pluralists maintain, the idea that there is such a mapping rests on a theoretical error about language, namely the view that there is a single substantial category of descriptive or fact-stating discourse.[29] It follows that the Humean conception of metaphysical virtue was always misguided. In granting the lapsed Humeans their metaphysical comforts, we thus deny them their Humean metaphysical virtue; for we say that there has never been any such thing.
Pluralism thus takes the heat out of contemporary metaphysical debates about a very wide range of topics: modality, mathematics, moral discourse, conditionals, and many others. As a final example, a well-motivated pluralism might allow one to endorse something like Dennett's account of the role and origins of intentional psychology, without having to concern oneself as to whether the view amounts to instrumentalism about the mental.[30]
Some readers will see this supposed advantage as like that of theft over honest toil--or perhaps more accurately, of idleness over honest toil. I see it as rather the advantage of idleness over unnecessary and unproductive toil--the advantage of stopping work on a misconceived and unproductive philosophical seam. True, we cannot be certain that the seam will remain unproductive, but there is enough doubt about the matter for the wise course to be to consider some of the general issues on which its eventual productivity might depend. I have emphasised that these are naturalistic issues, in particular issues about language as a natural phenomenon. (Does it admit a significant descriptive/non-descriptive distinction, for example?) So they are hardly issues that the pluralist's likely opponents may justly ignore.
At the very least, we could then go back to work with a better idea of what we were looking for in the first place. Instead however we might want to explore other philosophical seams, other projects that would now seem more promising: perhaps the elaboration of the general descriptive/explanatory approach to truth; perhaps the broadly Kantian examination and classification of different domains of discourse, with reference to their role in human life[31]; perhaps the exploration of nice issues concerning the connections and disconnections between the various discourses. A different philosophy, certainly, but not an idle one.[32]