PART I
Chapter I
Terms of Reference
I have a simple proposal to make about a difficult matter. I should like to set down a reckoning of the most promising controversial philosophical options that have gained most in plausibility as we approach the end of the millennium, to explain their significance, and to speculate in their favor. The topic is an irresistible one, but it also deserves a straightforward airing. The options I have in mind are beseiged on every side. The complexity of the disputes they involve threatens to swamp the understanding of any audience of wide reading that lacks a specialist knowledge of philosophy itself. The matter is important, but it is scandalously difficult to collect and make explicit. The truth is, we are (I think) on the threshold of certain revolutionary possibilities. Imagine that it were possible—it is possible—to organize a set of program notes for the theorizing melodies of the entire symphony of thought that ranges from the time of the ancient Greeks to the present, isolating the themes that have become insistent as the century draws to a close and that favor the most novel and most radical possibilities of that entire opus.
There is no compendium of these ideas, and yet they have made their inexorable way to the very edge of collective awareness in our own time, against the dominant doctrines of more than two thousand years of continuous dispute in the West. They are ready to be collected, and they are already so clear that they are no longer arcane. In my opinion, they are the threshold resources of conceptual visions that will surely compare favorably in scope and power and intellectual generosity with the
greatest turns in thought that our common history has dared to claim.
I admit the proposal is immodest. But what if its theme were true? Even a small failure would then be worth the inning. Now, a radical conceptual innovation in the sense I have in mind cannot be such if its meaning can be easily detached from the social history in which it is causally embedded. It is easy enough to dream up "radical" notions. They cannot but be idle, even if coherent, if they fail to have entrenched themselves in the productive energies of a genuinely effective part of the human community. The greatest proof of this may be found in the career of Aristotle, self-exiled from Athens when the xenophobic citizens of that city accused him of impiety after the death of that other great Macedonian, Alexander; also, in the career of Immanuel Kant, correcting the clocks of thought in a new way in the streets of backwater Königsberg. Surely Aristotle and Kant revolutionized Western thought, but they could not have done so without the connivance of the currents of ordinary thinking that were capable of taking fire from their inventions without rising to the same magisterial heights.
Also, it cannot easily be shown that these two succeeded because they discovered a part of the timeless truths the race needed to know and has preserved ever since. No, they effected a turn in thinking when history was favorably disposed to receive it.[1] Doubtless, they shaped that receptive history, but they could not have done so singlehandedly, and time would never have sustained their innovations if they had not liberated possibilities of thought inchoate but ready for the potter's hand.
The extraordinary power of their invention rests with an amplitude of vision fitted to a good-sized history that was attracted to what seemed (and still seems) rewarding, perhaps even necessary; necessity, however, lies more in our need to remember the changing history of our best thinking than with the discovery of the actual invariant order of things: first, because the notion of the necessity of invariant order is itself now under the greatest conceptual challenge; and, second, because the particular visions of Aristotle and Kant cannot be said to have been the foundation stones of any still-secure science.
Philosophy, it is said, progresses by remembering its greatest "failures." These are the handful of encompassing visions that have organized all our systematic inquiries. But they have superseded one another sequentially and dialectially, enlarging and adjusting with each step the perceived field of coherent thought in the light of an unpredictable, irrepressibly changing history.
To see a new vision, you must live intently among the old ones—but
as a pilgrim. St. Augustine saw this clearly in the ruins of Rome, but the lesson did not require anything so grand as the assurance of eternal life. Furthermore, you must grasp the new vision at that moment of good fortune when time is kind enough to make the discovery of its emerging form an irresistible marker of the conceptual shape of the future. No one can make such a prophecy prove true. The themes that evolve beyond its first appearance will be as inventive as it itself is. Only a vision that conveys a compelling sense of its own absorptive power over time could possibly collect all such diverging and emerging novelty. None can guarantee its adequacy: if any could, it would not be the invention it pretended to be; if it were not, then its limitations would in time be exposed.
Time, of course, overtakes all conceptual creations. There are no systems of philosophical thought so clearly adequate for the needs of the race that they will have been universally and changelessly acknowledged to be such. In time, all require replacement. Still, we honor in the process those visions that have contributed to the best parts of our best thinking. The reason is simply that thinking has a history—indeed, it is a history or it is history. We cannot abandon these "best" theories, because what we take to be best about our present understanding has gained its status by contending with whatever, retrospectively, we once took to be have been the best parts of the philosophies of the past. There is an innocent conspiracy there, but it has about it a touch of self-fulfilling prophecy. In this sense, the masterful work of Aristotle and Kant—and, perhaps even more revealingly, the innovations of Descartes and Locke and Hegel in the self-conscious history of the modern world—confirms the fortuitous sense in which great turns in thought need not require the greatest perfection in thinking and in which each particular turn is bound to be succeeded by another that it cannot rightly anticipate.
Two ironies make themselves felt here. First, in promoting the idea of an incipient revolution, one cannot fail to identify the greatest strengths of the prevailing visions that are meant to topple; doing that, one may inadvertently guess at powers in the former that one supposes can be bested. Second, in advancing one's own vision, the history of thought will already have been reinterpreted in ways that are most favorable to the favorable reception of a new vision; in doing that, one risks a very deep circularity. Taken collectively, these are "civic" virtues. Here, individual fortunes count for nothing, although the defenders of particular prophecies may be subject to unpleasant reversals.
In any event, on any reasonable overview, philosophical questions
will be admitted to address three concerns at least: (i) what we take to be the nature of the real world, and why; (ii) what we take to be the extent of our ability to have knowledge of the real world, and why; and (iii), having answered those questions in a robust way, how we should conduct our lives, and why. Doubtless there are other questions that philosophers have raised, but these three concerns are the central ones of the tradition. No theory that does not address them can be seriously regarded as philosophically engaged.
We may add at once that these concerns are implicated in the posing of any determinate question about, say, the melting point of gold, or about whether the butler murdered the master, or about whether El Greco painted in the baroque manner, or about whether Tom believes that Cicero denounced Catiline, and so on. We say that philosophical questions are "implicated" in questions of these sorts because answering them effectively justifies us in supposing that we may, and must be able to, answer ulterior questions about the way the real world is, or about what knowledge is; or, if we insist that the inference is not justified, we must concede that we are then obliged to explain satisfactorily why it is not. Either way, we offer a philosophical answer. The maneuver (that is, the choice between answering directly and treating justifiably any such answer as conceptually improper) was long ago recognized by the youthful Aristotle in an all but lost work, Protepticus, as the principal way of actually doing philosophy. In effect, Aristotle says, we are bound to pursue philosophy even if we deny that we can! "Reality" and "knowledge," it seems, are not ordinary properties or states of any sort that may be added to what we know of gold's having a melting point or Tom's believing that Cicero denounced Catiline. They point instead to certain distinctive interpretations ("philosophical" theories) that we draw from whatever we may take to be true in the way statements about melting points and beliefs are true. Hence, the denial that there are genuine questions of these sorts is itself a form of the question being dismissed. The denial, we say, is self-defeating.
Here, a terminological convenience recommends itself. If we call questions of the first sort first-order questions (questions for first-order discourse )—broadly speaking, questions of fact, questions about the way the world is or appears to us to be, in terms of ordinary inspection, activity, scientific inquiry, and the like—then questions about what, in the most critical sense, we take it we should mean, as in speaking of reality and knowledge (implicated thus in our first-order inquiries), will be second-order questions (questions for second-order discourse ). Sim-
plistically put, first-order questions are questions about the way the world is, and second-order questions are questions about the right use of the concepts of reality and knowledge (or, what we should take to be the "nature" of reality and knowledge therein entailed) in the most responsible way we can manage. The concern of second-order philosophical questions (notice that not all second-order questions are philosophical: think for instance of abstracting the grammar said to be implicated in the ordinary use of our native language) is, as we now say, the legitimation (the reasoned or critical justification or defense) of an account of what reality and knowledge "are," or "should" be taken to be, under the condition of admitting a suitable run of first-order instances of actual knowledge.
I say: proceeding only thus is purely terminological, since it is open to us to reject the supposed convenience (and significance) of the idiom; although, in doing that, we should not have escaped (if Aristotle is right) the perceived philosophical entanglement. One additional qualification makes this quite clear—and problematic. Some would insist that the distinction between first-order and second-order questions is strongly disjunctive. I suggest that, whether this is so or not, it remains true that the distinction between first- and second-order questions is itself a second-order distinction. And so, the matter remains so far unsettled.
The reason the point deserves mention is that some contemporary thinkers believe it to be quite possible to save the inquiries of science (first-order questions) while, at the same time, repudiating the legitimative questions mentioned (the second-order questions). This is in fact the self-proclaimed theme of what is now fashionably called philosophical postmodernism:[2] the ability to eliminate legitimative questions without generating paradox. I contend only that any reasoned speculation as to why we should prefer one mode of inquiry over another, for the sake of an anticipated improvement in our discovery of the facts about any part of the world, is, by its very nature, a form of legitimative inquiry. Put another way, first-order inquiries without second-order inquiries cannot but be blind; second-order inquiries without first-order inquiries cannot but be empty; and the distinction between the two cannot but be a second-order distinction. This is already a considerable gain made with very little labor. It prepares the way for the principal work of philosophy. It is a philosophical gain. But it hardly settles what we should say about reality, knowledge, or conduct.
If an argument like the one just sketched did not hold, there would be an end to philosophy. If we may go on blithely collecting what we
take to constitute our sciences and our moral codes without ever stopping to reflect on our apparent good fortune, without ever having to consider the reasoned advantages of preferring this or that line of inquiry (Lavoisier's conjecture about the role of oxygen in combustion as opposed to Priestley's conjecture about the role of phlogiston), or of preferring this way of generalizing over experiments to that (applying Mill's method of difference or constructing a pertinently falsifiable claim in Popper's manner), then philosophy would indeed have no function. But if we cannot make sense of the supposed economy—and we cannot—if the intervening questions (about what to construe as knowledge and reality) do substantively and reasonably affect the direction and perceived outcome of our first-order inquiries, then philosophy is inescapable, but not for that reason cognitively "privileged" in any way at all. It is true that the very use of the paired terms "first-order" and "second-order" betrays a certain philosophical prejudice, namely, that science and moral judgment (for instance) are indeed first-order concerns and therefore implicate "second-order" questions of the sort sampled. But the denial of that conviction seems, as I say, distinctly paradoxical—self-referentially—in the sense Aristotle captured.
To ensure the merely terminological nature of the distinction being introduced, we must concede that the terms "first-order" and "second-order" have no bearing at all on the supposed logical or cognitive or legitimative priority of one over the other. The range of their respective applications is effectively inseparable. Every question of fact, we may suppose, implicitly signifies the relevance of a corresponding legitimative question, and no legitimative question has any point at all except relative to some admitted set of truths about the world. The division of professional labor may create the illusion that there is a strong disjunction between first-order and second-order queries: that one may be pursued without the other. I urge that this is false or cannot be demonstrated to be true. (It is part of the point of this primer to supply the sense in which that may be shown.) For the moment, I suggest only that you bear the matter in mind wherever the discussion turns to "legitimation" or "naturalism": for naturalism will be seen to be the dominant current of late analytic philosophy committed to the dismissal of legitimative questions or to their displacement by ordinary first-order causal questions. It is hard to gauge how radical such a program is, but it will occupy this primer in a surprising variety of ways. In fact, as in the work of Rorty and Davidson, naturalism and postmodernism sometimes converge. (My argument is opposed to both.)
It was, in fact, the presumption of an older tradition of philosophy—that held that questions of empirical science and of perennial philosophy could be separately pursued—that has led to the extravagance of what is nowadays called postmodernism. It is equally doubtful that the correction of that blunder could ever restore philosophy's presumed privilege or the disjunction tendered. (That is postmodernism's gain.) In any case, nothing is lost by the terminological caution I now offer; for, if philosophy does indeed have some cognitive advantage over the admitted resources of first-order inquiry, the champions of that view will surely make themselves heard and all of us may then decide the matter together. If such an advantage is shown, then what I have characterized as a terminological convenience will be much more than that. But there's the risk my claim admits. If one thinks of Einstein's reflections on the meaning of simultaneity in connection with relativity theory or of Bohr's reflections on the role of observation in the context of quantum physics, then it is clear at once that the burden of proof must rest with all those who mean to disjoin science and philosophy or to privilege either with respect to the other. That maneuver has become well-nigh unthinkable in our time.
May I, therefore, collect the gains we have made? I don't say I have proved the distinctions offered. But they look very promising, and they are as neutral as we may ever be able to afford—except in the sense that they are already opposed by thinkers of importance in the history of philosophy. We must, however, make a start somewhere. Perhaps any start is "hegemonic," as many now say (borrowing a term from Gramsci), in the plain sense that to catalog the world in this way is certainly to preclude its being cataloged in that way, in the same universe of discourse. (Such thinkers are often equivocally termed "postmodernists" or "poststructuralists";[3] but their worry, although not altogether pointless, is insuperable.) We cannot deny that any conceptual scheme excludes others in the very space in which it is actually applied, but that is not the same as claiming that a particular conceptual scheme is demonstrably not well suited to managing the legitimative question itself. To confuse the two issues is to opt for the irrelevance of thinking of any kind. It is a genuine form of conceptual anarchy well beyond a serious skepticism about what we could possibly know or say.
It is nonsense to suppose that categories could function otherwise, and it is useless (in the profoundest sense) to claim sans phrase that any catalog ruins every effort to legitimate our account of reality or knowledge or the right conduct of our lives. Particular challenges need to be
"reasoned," defended one by one, in the sense that some reasons may be seen to show compellingly that this way of cataloging the world but not that is indeed unsatisfactory. It is not helpful to add at once that, of course, that judgment is equally "hegemonic"—and that, really, no opinion of this (or any) sort can ever fail to be such. For, in each instance, serially as well as conjointly, each such challenge invites a pertinent dispute without ever implicating the old sense of privilege.
I was on the point of collecting all that has so far been shown—before I interrupted myself. I offer three findings:
(1.1) philosophy is concerned with the second-order legitimation of our notions of "reality," "knowledge," "right conduct," and the like, that are implicated in first-order discourse;
(1.2) the distinction between "first-order" and "second-order" discourse is terminological—neutral to every philosophical option;
(1.3) theorems (1.1) and (1.2) entail no cognitively privileged views of logic, knowledge, truth, meaning, reality, methods of inquiry, norms of conduct, or the like.
Theorems (1.1)–(1.3), I say, identify in the most generous and least quarrelsome way what philosophy is all about. They do so in the least informative way possible without being utterly vacuous. Any more explicit statement would risk skewing the issues I mean to pursue before they are even before us. (For instance, "legitimation" is meant to signify a form of second-order critique that is not, or at least need not be, "transcendental" in the Kantian sense. I shall come to this in time.) Nevertheless, these theorems already come to blows with the themes associated with postmodernism. Postmodernism, therefore, I now discount altogether (for example, in the work of Lyotard and Rorty), except in the sense of conceding that questions of would-be privilege regarding first- and second-order matters will indeed have to be aired. I do not dismiss poststructuralism for the same reasons, (i) because (preeminently in Foucault's work) it does not reject philosophical issues along postmodernist lines, and (ii) because, although the thesis regarding "hegemony" is preposterously grand, poststructuralism has other fish to fry, even with respect to the exclusionary function of given conceptual systems that, in effect, bear on the fortunes of theories subtended by theorem (1.3). (I shall return to the issue.) But, for the sake of a tidy
account, I say only that (1.1)–(1.3) serve a purely terminological function. If allowed, that amounts to a very nice gain.
I venture to add an interpretation of (1.1)–(1.3) that is substantive, that I cannot yet confirm, that I shall eventually confirm, and that is important for the entire argument that follows:
(1.4) the relationship between first-order and second-order (legitimative ) discourse is a function of the historicized nature of thinking.
Rightly understood, the validation of (1.4)[4] shows at a stroke the pointless fear of postmodernists regarding legitimative questions, namely, that their resolution must always devolve into Kantian or transcendental or apodictic arguments of some sort. I have misgivings about introducing (1.4) without suitable defense, but its eventual importance overrides that scruple; in any case, it will be satisfactorily secured in time (in part II). For the time being, you will have to guess at the meaning of "historicized" (and "privileged"). I shall draw no consequences from (1.4) until I have justified its use, but it belongs among the findings so far given, and it suggests an important version of what I should mean by "poststructuralism." (Actually, it adumbrates the entire lesson of this primer.)
Let me be clear about all this. The schema I introduce is meant to enable us to range over the entire history of Western philosophy without strain or undue distortion. I say "Western" philosophy because I am quite unable to judge to what extent what I have offered may affect the discussion of Hindu or Buddhist or Taoist or other Far Eastern philosophies, or Native American or African doctrines for that matter, or anything of the kind. I frankly cannot see how items (1.1)–(1.3) could adversely affect the analysis of these latter materials, but perhaps they can and perhaps I am blind to the danger. The qualification counts only as a scruple. I shall touch on it again at some point in the running account wherever the mention seems useful. I must however add here, before we go too far, another theorem (that is not quarrelsome in the manner of (1.4)), that I have already mentioned informally in passing, in explaining the sense of "legitimation," namely:
(1.5) the distinction between first- and second-order (legitimative) discourse is itself a second-order distinction.
Theorem (1.5), I should say, is in effect a corollary of (1.2)–(1.3) and defines the sense in which, in (1.1), philosophy is said to be implicated in first-order discourse.
On the argument so far sketched, I have said nothing as yet of a substantive nature regarding the radical themes I promised at the start. On the contrary, the various would-be canons of traditional philosophy may also claim to address the issues subtended (now) by (1.1)–(1.3) and (1.5), which, in accord with (1.4), I mean to pursue in a way opposed to the "canons." In short, I begin irenically. To examine matters thus is to have made provision for any number of confrontations between what the tradition has variously favored, from the Greeks to our own time, and the new options that (as I say) are taking form at the end of our century.
Chapter II
In Lieu of First Principles
There is no first principle of reality or knowledge or reason that must be settled before all other philosophical questions may be effectively answered. Certainly, there is no widespread agreement about what determinate form such a principle should take. There is also little prospect of agreeing on any first principle unless the real world has a discernibly changeless order. But if it had such a structure, then, in an obvious sense, that would provide the decisive first principle on which diverging philosophical policies could jointly claim to rely. Hence, short of establishing the facts of invariance—which innumerable philosophies have pursued in bewilderingly different ways—it would be a best counterstrategy to demonstrate why such a principle would be unlikely or impossible to confirm.
In actual fact, very nearly the entire literature of classical Greek philosophy—from the Presocratics through Aristotle and beyond—is committed to one version or another of this appealing (first) principle: necessarily, reality is invariantly structured and, when known, discernibly known to be such. The sole exceptions appear among the Sophists, notably Protagoras and Gorgias, although what they are reported to have said has been very harshly judged through the entire tradition as being incoherent or self-contradictory or paradoxical or self-defeating. In particular, Plato (in Theaetetus ) and Aristotle (in Metaphysics Book Gamma) argue that there is no coherent recovery possible of Protagoras's rejection of the principle just mentioned.
I have no intention of pursuing the issue in narrowly textual terms. I
have done so elsewhere. To pursue the matter would not conform with the modest purpose of my undertaking here; however, it would be a mistake to ignore altogether the advantage of airing the issue, since it bears directly on the fortunes of the radical possibilities I promised.
My thought is this. Aristotle's argument happens to supply the most explicit and the most powerful reason possible for claiming that any repudiation of the supposed principle just formulated would produce instant incoherence or self-contradiction. If Aristotle were right, it would pay to know it and stray no more. Nevertheless, Aristotle's argument fails (I claim)—fails utterly—and by that failure affords an opportunity for signaling an entirely new direction in which philosophy might proceed. (I do not say the defeat of Aristotle's argument entails the defeat of all would-be first principles, but it certainly shifts the burden of proof.)
The matter is complicated on the historical side but is altogether straightforward in terms of how the line of reasoning opposed to Aristotle's should go. Bear in mind that what I wish to draw attention to is not offered for textual or historical reasons primarily, but its effect on Aristotle's fate and on the record of the historical tradition cannot be ignored. For, Aristotle without the principle mentioned is no Aristotle, and the defeat or stalemate of the principle is—it is hard to soften the blow—a very strong (though not conclusive) indictment of the most salient, most authoritative thread of the entire tradition of Western philosophy. (That may not be believed.) So the economies of the contest intended are certainly worth noting, and the patience the argument requires may then be deemed justified.
The point of Aristotle's argument (I shall examine it more closely later in this chapter) is that the mere denial of the principle—I call its advocacy archism,[1] that is, any attempt to avoid conforming with it in making ordinary claims and judgments—instantly produces contradiction at some point in discourse because it must ultimately contradict (so it is claimed) an ineliminable part of what is true of whatever may be found in the real world. In short, Aristotle's argument is that the principle is conceptually necessary for all coherent, all cognitively qualified discourse. If true, his reasoning marks archism as the obvious first principle of all philosophy. But if he is mistaken, it follows not that the world lacks an invariant nature but rather that
(2.1) It is not in any way conceptually necessary that reality possess invariant structures or an invariant nature.
The essential thesis of Greek philosophy—that "what is real is changeless" (the doctrine shared by Parmenides, Plato, and Aristotle)—is, according to theorem (2.1), not itself a necessary truth. There's the pivot of the whole trajectory of Western thought. (I shall come to the pertinent details on Aristotle shortly.) The objection to archism is not an objection to the bare claim of invariance (qualified in whatever epistemic way we please: for instance, as apparent, approximate, perceived, inferred, or reasonable as a conjecture); it is an objection to its supposed conceptual unavoidability. Aristotle's argument holds quite simply that any departure from the archic doctrine is necessarily contradictory and that the principle of noncontradiction is the most certain of rational principles .
Archism, the denial of (2.1), is a modal principle, in that (i) invariance (of whatever kind is alleged) is said to be a necessary property of whatever is real (de re ) or (ii) adherence to such invariance is a necessary constraint on rational discourse or thought addressed to the analysis of reality (de dicto ). Mere empirical regularity—apparent invariance in perceptual terms, indicative "invariance" or "universality" (as I should say)—yields no modal claim stronger than the claim that, being actual, indicative invariance is, trivially (conceptually as well as "materially"), possible as far as nature is concerned.
Theorem (2.1) is, however, emphatically not that sort of modal claim itself; it is rather the upshot of a conceptual bet:that all such modal principles—archism, in the present context—fail of demonstration, fail of any proof that they are indeed necessary in the sense intended. Any alternate reading would produce an obvious paradox: the denial that there were necessary principles would instantly confute itself. Furthermore, if there were first principles, they would be principles of modal invariance. For any would-be "first" principle (regarding reality or knowledge of reality) would be a principle presupposed in all pertinent discourse but would not itself be subject similarly to any prior presupposition; that condition is itself a form of modal invariance.
This simply means that if (2.1) affords a coherent possibility, as indeed it does, then archism cannot but be false. Alternatively put, if (2.1) is admitted, then
(2.2) philosophy—a fortiori, science—may proceed, coherently, without first principles and without assuming the necessity of resting on first principles.
There is no direct argument here as yet—for instance, against Aristotle—only a hint of what remains to be shown. A principle I take to be
(i) a truth, (ii) general and exceptionless for a given domain, (iii) judged (on some theory) to be systematically implicated in a strong sense in confirming and disconfirming the pertinent truth-claims of that domain, for instance in an explanatory or legitimative way, hence (iv) usually ranked or normatively favored over other would-be principles. Principles need not claim modal invariance or necessity, but first principles cannot fail to. For instance, the golden rule may be a reasonable ("regulative") principle without being or without being viewed as necessarily true (that is, such that any violation of it would be intrinsically "irrational" or "incoherent"), and something like (but weaker than) archism may be a fair ("constitutive") conjecture about the way the world is without being taken in Aristotle's strong (modal) sense.
Of course, for perfectly trivial reasons, if (2.1) were true, it could not then also be true that reality could ever be known to have a necessarily invariant nature. In fact,
(2.3) if (2.1) is true, then archism is false: there are no first principles,
for archism holds that the denial of the principle that reality is changeless is itself necessarily self-contradictory. (I confess I find affinities here with St. Anselm's proof of God's existence, but the archic thesis is hardly as controversial as that.) For the moment, we need not dispute whether there is any pertinent sense in which the world exhibits an invariant structure: the important point is that discourse would not be obliged to assume it in order to save its logical virtue. To confirm (2.1) would give an initial impetus to the heterodox venture I am endorsing. The issue at stake is the modal operator, "necessarily " ( = "not possibly not," sans phrase: that is, conceptually—a fortiori, materially).
Before I turn to the supporting argument, let me say a word about the historical situation. No matter how subtle the issue may have become in recent philosophy, it would be fair to say that nearly the whole of Western thought has vigorously committed itself to the denial of (2.1) in some form or other. It's true that many of the obvious versions of that denial have been overthrown by this time. Certainly, the master theme of Aristotle's philosophy—its so-called essentialism[2] —is no longer thought to be necessarily true. But to admit only that is hardly to reject essentialism itself or to sustain (2.1). Arguing in the reverse direction, it may be claimed that to vindicate essentialism —if by that term we mean that natural phenomena have and may be discernibly known to have changeless natures—implicates archism. For the modal
claim—essentialism—itself presupposes a first principle of suitable modal strength. It is also worth mentioning that, although the history of philosophy has overwhelmingly favored the rejection of (2.1), now, for the first time in a strong and persistent way, the defense of (2.1) bids fair to dominate the rest of our millennium. The point of that reversal is an important part of the story I have promised.
Let me, then, for the sake of the intended final picture, label all those views that support (2.1) as an-archic philosophies. Alternatively put, archism holds that reality and reason are necessarily invariant; anarchism holds that reality may be a flux, that is, lacking invariant structure or lacking necessarily invariant structure, and that rational thought need not invoke such invariances. (Invariance, therefore, takes an "indicative" as well as a "modal" form: changelessness or necessary changelessness.) Furthermore, archism holds that knowledge—even the avoidance of contradiction—is impossible if (2.1) is true, and anarchism holds that the coherence and consistency of (any would-be) knowledge are not rendered impossible by accepting (2.1). The meaning of "knowledge" cannot be shared by archists and an-archists, then, if both admit that human inquirers are capable of knowledge.
It is remarkable not only that twenty-five hundred years of continuous philosophy have been largely archic but that philosophy is no longer reliably such. We cannot substitute chronicle for argument, of course, but it is hard to suppose that the growing contemporary support of (2.1) is merely wilful. Its advocacy suggests how to begin to mount a reasonable campaign in favor of the radical options I have been hinting at. The point of Aristotle's claim, for instance, is hardly local to his philosophy. It embodies a universal theme. It features an essential ingredient that our counterstrategy must address if it is to have any inning at all. The same theme, let it be noted, appears in the very different policies that link the work of Frege, the "early" Wittgenstein (of the Tractatus ), and the Postivists of the Vienna Circle, who, not unfairly, may be said to form the vanguard of the principal company of contemporary analytic philosophers who still repudiate (2.1). I take the modern and contemporary ways of opposing (2.1) to be, often, cryptic pronouncements of archism, sometimes misperceived by their advocates, even disguised at times. If the linkage between the ancients and the moderns may be endorsed without serious distortion, then, by a neat economy, in exposing the fatal weakness of Aristotle's version of the argument we may claim to have gained a more general march on the entire archic company.
That would make a very trim maneuver. It would set the stage for a
larger campaign. It would confirm the near-unity of the entire tradition of Western philosophy. It would raise an otherwise local skirmish to a planetary level. It would lay the ground for explaining the importance of the dispute intended for the entire range of science and inquiry committed—in terms of their own worries—to choosing between the archic and an-archic options. For example, it would not be difficult to guess that the analysis of what we should mean by a law of nature or a causal law, in physics or economics, could hardly fail to be affected by the fate of (2.1). Thus, for different reasons, Aristotle and Hempel would regard true science as impossible if (2.1) were true. Hence, I say, Hempel is a crypto-archist. Hempel regards the laws of nature as modally necessary. There can be no question that the quarrel focused by affirming (2.1) is inseparable from the legitimation of science itself. I shall return to the issue in a more pointed way when more of what is needed has been effectively assembled. I set it aside for the moment.
The immediate corollary of admitting the point of our quarrel is this:
(2.4) the structure of natural language and rational argument and the structure of the cognizable world "implicate" each other.
The intriguing thing is that (2.4) is a claim shared by Aristotle and the company (just mentioned) that includes Frege, (early) Wittgenstein, and the Postivists. It is (also) one I myself am fully prepared to endorse. In one form at least (to be clarified in a moment), it is absolutely central to all late twentieth-century disputes. That is, (2.4) may be shared by archists and an-archists alike. I don't mean that all an-archists explicitly accept (2.4) any more than archists. I say only that (2.4) need not separate them. Theorem (2.4) is a full-fledged principle (arche ) for the archists, because it is meant to be an abstraction from some more robust claim of modal invariance—for instance, Aristotle's essentialism or Wittgenstein's explicit correspondentism (in Tractatus ).
In the hands of the an-archists, on the other hand, (2.4) is not a first principle at all, but the upshot of an argument—a bet —that holds that, regardless of what analysis may determinately yield, there is no ineluctable policy for legitimating the analysis of reality, knowledge, language, argument, truth, or related notions in any way that (i) disjoins, in a "principled" way, the analysis of discourse, thought, reason, or knowledge from the analysis of reality, or (ii) requires that truths about the first accord with the logically prior independent truths about the second (or vice versa, for that matter). To hold (say) that what the analysis of
language and thought yields may be segregated in a principled way (even prioritized one way or another) is, I say, to favor a privileged thesis. But to say that what the analysis of either language or thought yields cannot be disjoined from what may be affirmed of the other is, as such, neither (or not yet either) archic nor (or) an-archic; on the second reading, the two "yields" implicate one another. Furthermore, where the archist makes a principled claim, the an-archist is content to show only that that claim fails to be confirmed as necessarily true. In that (negative) sense, the an-archist's position is a philosophical bet, a reasoned challenge to a would-be principle, not a modal principle at all: hence, not paradoxical. (But it is also not a mere Pascalian wager, an exercise of hope or will.)
In fact, the "privileged" reading is the nerve of Greek archism. (the meaning, for instance, of Aristotle's doctrine of nous [rational intuition]). It also appears (disguised) in reverse in Dummett's recent account of the relationship between semantics and metaphysics.[3] Hence, on the argument intended, we may move at once to a consequence of (2.4), namely,
(2.5) assigning legitimative priority to the right analysis of reality over knowledge or of knowledge over reality, or of logic over empirical science or science over logic, or the logical syntax of empirical language over the semantics of empirical inquiry or the reverse, or anything of the kind, is tantamount to legitimating some form of apriorism or cognitive privilege.
I say archists and an-archists may share the sense in which "privilege" is intended in (2.5). The archists embrace privilege and the an-archists reject it. It needs to be said that archists need not explicitly invoke privilege as they proceed. There is no such explicit claim in either the Aristotle of Metaphysics Gamma or the Wittgenstein of Tractatus, but the fact is that both claim that what they have discerned, in opposing relevant versions of (2.1), does indeed entail what, touching on the topics of (2.2)–(2.5), is necessary and invariant with regard to the real world. By privilege, then, I understand (i) that feature of certain of our cognitive powers or resources, which (ii) inquiry and experience do not adversely affect or distort (despite the contingencies of human existence and discourse), such that (iii) we possess (through their exercise) an assured ability to discern (a ) the real properties of the world as it is independent of our inquiries (objectivism ) or (b ) its necessary or invariant or universal
or merely "objective" structure prior to inquiry or (c ) what is necessarily or timelessly true of it even if the world is not "independent" (apriorism ), and/or such that (iv) that ability may yield indubitable knowledge of certain epistemically strategic parts of the world or at least the strong reliability of what, by such means, we take to be true of it (foundationalism ), or that (v) that ability may yield certain reflexively discerned, necessary, "constituting" structures that human understanding provides for the "constructed" order of the (phenomenal ) world we claim to know (transcendentalism ). It is not impossible, therefore, that transcendentalism and apriorism be jointly affirmed, without affirming any explicit foundationalism. This is, in fact, part of one standard reading of Kant. But, if so, then it is not paradoxical that, as an "empirical realist," Kant should be viewed as holding a thesis akin to objectivism (as Husserl charges).
Some terminological distinctions will be helpful here. By independent, I mean only that real condition of the world in virtue of which it possesses a determinate structure entirely apart from, entirely unaffected by, the intrinsic conditions of human knowledge, understanding, and inquiry. (By privilege, as I have said, I mean only that cognitive competence by which human inquirers may know (a ) the "independent" real world as such; or, if the world is, in some essential regard, constituted, structured, by the structure of our cognizing powers, (b ) what those constituting structures are.) I should add, since the matter will surface in various forms later in this primer, that the conjunction of these two notions (independence, privilege) constitutes what, by a term of art, I shall call the externalist theory of knowledge.[4] (Notice that "externalism" may be made to take either of two quite different forms—marked as (a ) and (b ) in the definition of "privilege." I should say the (a )-reading was the usual—indeed, the proper—one and the (b )-reading the illicit and distinctly paradoxical one. The (b )-reading, of course, is sometimes applied to Kant: that is, by permitting the (b )-reading of "privilege" to override strict independence. Thus construed, the plausibility of ascribing the (b )-reading to Kant rests with the fact that Kant reinterprets standard forms of seventeenth- and eighteenth-century externalism within the novel, non-externalist terms of his own apriorism [what I shall later call constructivism ].)
Externalism proper corresponds to what Dewey called the "spectator" theory of knowledge and world. The classic externalist target has been Descartes (or a Cartesian Locke), but, apart from the textual record, the preponderant part of the history of Western philosophy has
been externalist. (Even Kant's "empirical realism" has been construed along externalist lines (for instance, incipiently, by Moore), but disjoined as far as possible from Kant's "transcendental idealism." That, of course, is a misreading, but it is a misreading suggested by Kant's apriorism. Analytic philosophy tends to follow Moore in this.[5] The reading may be thought to be encouraged by such remarks as that in the preface to the second edition of the first Critique, in which Kant, reflecting on Galileo's and Torricelli's experiments, invokes "principles of judgment based on fixed laws" and speaks of "constraining nature to give answers ['not fictitiously' ascribed] to questions of reason's own determining.") Foundationalism I take to be a strong modal thesis. Hence, merely to posit cognitive foundations —perceptual sources, say—is not (yet) to support foundationalism. The relation between the two options is analogous to that between principles and first principles. (I return to these issues with ampler resources in chapter 9.)
I must risk a further consequence of (2.4), read in terms of (2.1) and viewed as an obvious corollary of (2.5):
(2.6) there are no de re necessities tout court, and there can be no separable de dicto necessities.
I mean that there are no formulable necessary connections or structures or regularities in the world that an alternative conceptual orientation could never coherently displace; any would-be formal, logical, syntactic, conceptual, or rational necessity is either vacuous (unless interpreted in accord with a determinate reading of matters regarding (2.5)) or (when so read) always open to being coherently displaced. One might reasonably suppose that Quine's opposition to any principled reading of the anayltic/synthetic distinction would incline him to favor (2.5)–(2.6).[6] That is by and large so, but there are also strains in Quine (for instance, his view of "holophrastic" sentences) that count against it—in particular, his strong preference for an externalist doctrine. (I return to Quine later.)
Another way of putting the same lesson is this:
(2.7) inquiry is symbiotized,[7]
that is, admitting theorems (2.5) and (2.1), (i) there can be no principled disjunction between cognizing subjects and cognized objects, every would-be inquiry into the "one" implicates an inquiry into the "other,"
and (ii) there can be no privileged knowledge of "either." Once we grant (2.1), there can be no archic account of separable knowledge (of the world) or separable world (as known). (I take this to expose the ultimate arbitrariness of Kant's apriorism. Kant supports a "constructed" world—but it is transcendentally constructed. That is, the cognizing subject is not itself constructed. Kant is, therefore, opposed to symbiosis.)[8] Hence,
(2.8) distinctions between the cognitional and the real are artifacts "implicated," by conjoint pairing, in our (an-archic) theories about knowledge and reality.
There are no unconditionally fixed truths about truth, reason, meaning, knowledge, selves, or world disjoined from one another. Of course, discourse speaks of all these matters determinately, and there seems no way of escaping that condition. I am not suggesting an alternative. But I do say that if (2.8) be admitted, then some extraordinarily powerful consequences follow. In particular:
(2.9) every disjunction between the mental and the physical, the subjective and the objective, the natural and the cultural, the private and the public, the perceptual and the explanatory, the theoretical and the practical, the syntactic and the semantic, the linguistic and the real, and similar "oppositional" pairs can be no more than a provisional distinction relativized to one symbiotized stance or another;
and
(2.10) what may be said of what belongs to the one pole in the oppositions (collected in (2.9)) is conceptually inseparable from what may be said of what belongs to the other pole.[9]
In a word, what are symbiotized are not more than relata in some ontic or epistemic space. The important thing is this: symbiosis precludes "privilege" of any sort. Relata, as such, lack any criterial or cognizing function. Hence, what are said to be symbiotized are also, in ontic and epistemic contexts, taken ("interpreted," as I shall later say) as individuatable denotata of some sort. (More needs to be said about this. I return to these distinctions in chapter 4.) Furthermore, oppositions like those given in (2.9) are, I should say, prejudiced,[10] in the sense that (i)
no particular oppositional pair is privileged in any archic way, (ii) the world and one's relation to the world will appear differently within different conceptual schemes, and (iii) the conceptual schemes we actually use are those we acquire (in some contingent way) as we become the competent agents of discourse (or of thought and action) that we do become as we grow up in our home societies. Any and all distinctions between "cognizers" and "cognized" in accord with (2.9) I shall call external, so that "externalism" supports counterpart ("externalist") distinctions, but on the condition of denying theorem (2.7).
There are perhaps two principal ways of opposing (2.6) for which the historical tradition offers evidence: in one, it is claimed that logical (or near-logical) principles like noncontradiction, excluded middle, and numerical identity are genuinely invariant, and that opposing them automatically risks contradiction or incoherence or both; in the other, it is claimed that certain concepts that we employ in understanding the meaning of what we predicate of the world already include, discernibly, or tend asymptotically toward, invariant constraints, and that (therefore) opposing them risks contradiction or incoherence. The first strategy belongs to the programs I've already mentioned—Aristotle's, Frege's, Wittgenstein's, the Positivists'; the second is more diffusely represented, but I think its greatest champion in modern times is Husserl working as a phenomenologist. (The second also appears in various versions of "progressivism," as in Peirce, Popper, and Habermas.) I intrude the substance of (2.6) merely to assure you that the argument intended has strong consequences that I shall not ignore.
A more manageable economy suggests itself. There are two senses in which the de re/de dicto distinction is employed in the philosophical tradition. In one, it is no more than a "stylistic" distinction (although with attendent potential felicities and infelicities): on that view, what, in general, may be formulated de re may also be formulated de dicto, and vice versa. This is akin to the sense in which Carnap speaks of the "material" and "formal" modes of discourse, which Quine is prepared to accept—although with a distinct preference for the de dicto ("semantic ascent").[11] In the other, the distinction is treated disjunctively: discourse about language is then not discourse about the world as such. On the first usage, it is clear that, except for convenience and stylistic felicity, there is no difference between what may be said in the one idiom and what may be said in the other. Hence, there is no essential difference between would-be de re and de dicto necessities (whether there are any such necessities). On the second usage, de re and de dicto necessities are
meant to be initially distinct (whether would-be necessities of the one sort may be suitably matched by necessities of the other). Hence, on the second usage, language, being "about" the world, is not "in" it. But if we think of language as an abstraction from actual linguistic practices, the disjunction is benign enough.
According to the archism Aristotle favors, the principle of noncontradiction (and that of excluded middle) may be taken to be necessary de dicto (that is, imposes a constraint on nous ) conformable with the (independent) de re necessity of the changelessness of reality. This is the sense in which Aristotle subscribes to (2.4) but not to (2.7): the structure of our theorizing reason (nous ), he says, matches the structure of reality, but they are separable (archism). By (2.7), the an-archist (i) admits no (strict) necessities de re or de dicto and (ii) denies any principled disjunction between de re and de dicto discourse. Nevertheless, the anarchist (iii) need not deny "apparent [indicative] necessities"—for instance, against Descartes's oddly weak conjecture that, given God's omnipotence, a "valley" need not necessitate a "mountain"—which provoked Leibniz's scorn for his illustrious contemporary.
By the same token, the an-archist need not oppose the sense in which Aristotle affirms the necessity of the principle of noncontradiction. He needs to reinterpret it in accord with his own account. In a spirit closer to that of this primer, Foucault admits the "historical a priori, " by which he means to admit conceptual necessities—"a priori, " but only relative to the historicized episteme (or discursive "regime") in which our particular conceptual resources are formed or "preformed." As Foucault realizes, these "necessities" ultimately depend on the contingencies of shifting epistemes . (This begins to give a sense of the promissory note collected at (1.4).)
We are now ready for Aristotle's archism. What Aristotle claims (in Metaphysics Gamma) is that you cannot help but contradict yourself if you restrict discourse about real things ("primary substances," as he calls them: ousiai ) to how such things appear or seem to us to be. (The argument is directed against Protagoras, but that is not essential.) Aristotle claims that to restrict discourse in this way yields a contradiction somewhere .
The modern reader is bound to be nonplussed, because he or she will not be able to see, in the option mentioned, any formal contradiction at all. It's only when one realizes that Aristotle means that, since "primary substances" have essential natures, the principle of noncontradiction must be applied in accord with that, that one will understand the charge
that the restriction of predicative discourse to the appearances of things produces contradiction "somewhere"; under Aristotle's constraint, one would then be talking of appearances only, one would be obliged to treat what is essential as accidental. The modern reader will be puzzled by Aristotle's claim that the contradiction arises here for reasons that, technically, are logical . He will not accept Aristotle's use of the term "logical." He will find metaphysical notions extraneously characterized as logical. (By logical, I mean, quite simply, any feature of language or thought said to be confined to formal, syntactic, semantically uninterpreted structures that yield valid arguments, or analogues of explicit arguments—as of interpreted sentences or terms treated as ordered premisses and conclusions. The validity of arguments is said to depend on uninterpreted argument forms; but arguments are always semantically interpreted sets of premisses and conclusions. Hence, argument forms are, effectively, always context-bound [bound to the range of interpreted sentences over which they range], in spite of appearances to the contrary.)
Aristotle is right, of course, to draw the conclusion he does: namely, that contradiction is bound to arise somewhere. But it is not obvious that his premiss regarding modal invariance is true. In fact, it is false ((2.1), (2.3)). Hence, it is not obvious that Aristotle has isolated the logical issue correctly. The important thing is that a rejection of Aristotle's full claim eliminates the threat of contradiction at that point but preserves the principle itself. The modern reader is likely to treat the logic of the matter as completely neutral to the interpreted (metaphysical) argument: he will certainly not side with Aristotle. But Aristotle is committed to supporting (2.4) and to denying (2.7).
What I urge instead, against both Aristotle and the modern habit of disconnecting logic and metaphysics, is that both (2.4) and (2.7) are worth defending and are genuinely defensible. By linking Aristotle and Frege in the way I have, the vindication of (2.5) serves to defeat as well the subtler versions of necessity and privilege that modern violations of (2.5) tend to obscure. It is not unreasonable to suppose that such important contemporaries as Quine, Goodman, and Davidson tend to camouflage their own violations of (2.5).
It is easy to show—by (2.1)—that we can detach the principles of contradiction and excluded middle from the fate of Aristotle's particular metaphysics, and, by a similar argument, that it is not necessary to suppose that real things have "essential" or fixed natures. No one has ever shown that the denial of essentialism is self-contradictory. (I shall characterize "essentialism" more fully in a moment.)
For my present purpose, the matter may be treated conditionally. If the essentialist premise may be denied without self-contradiction, then it would be wrong to insist, as Aristotle does, (i) that to restrict discourse to appearances alone (as Protagoras is said to have done) necessarily leads to self-contradiction, or (ii) that to concede that not all predicates need apply "bivalently" (in accord with excluded middle) yields the same disorder. Essentialism holds that (i) there are attributes particular things possess, without which, pertinently, they cannot be the particular things they are or, possessing which, they must be real particulars of the kinds they are, and that (ii) those attributes define the changeless "natures" of such things in accord with which they alone change and may change. Hence, essentialism is a strong modal notion: particulars cannot be real, have natures, or change except in accord with the necessary truth (de re ) that they are real in virtue of the changeless natures they possess. (I shall, in chapter 7, show how this line of argument confuses or conflates the quite different processes of "individuating" particulars and "identifying" or "reindentifying" them. To explicate the required distinction is to demonstrate how the fixity of "numbering" individual things does not as such entail the fixity of the "nature" of things thus numbered. The matter is not without its subtlety.) In any case, on the essentialist view, the status of noncontradiction and excluded middle is strengthened, since, on the argument, every predication must be reconciled with the essential nature of the referent in question.
The countermove against this thesis (of Aristotle's) is absurdly simple, for Aristotle nowhere shows that repudiating essentialism is incoherent or self-contradictory. He shows only—and correctly—that if essentialism is assumed, then conditions (i) and (ii) do indeed hold and may be made to hold as logical principles . He has not shown that the denial of this assumption also violates the principle of noncontradiction. Without having shown that, however, his entire argument collapses. For the issue is not whether we should suppose that things have essential natures, but whether it is logically necessary that they do—whether that is the only consistent policy possible . (Of course, it is not.) It is easy to see that (but not so easy to explain how), given the way "things" appear to us, the "natures" we assign "them" may be coherently assigned (without any fixity at all) solely on the strength of (their) changing appearances. (This agrees with what I have said about "foundations" and "principles." That is, the argument against essences and the modal necessity of noncontradiction is, as I've suggested, an argument against first principles.) I grant the conclusion may be delayed, but I claim it cannot be con-
vincingly overturned. The most important objection to it asks whether we can identify what we mean to make predications of if we do not individuate and identify our referents as particulars exhibiting essential natures. Indeed, that is Aristotle's intent, although Aristotle never investigates the modern question of how to identify and reidentify this or that particular as numerically distinct. He offers only a general solution to the puzzle of individuation. He never links the question of individuating things and identifying and reidentifying them numerically as one and the same or as different.
Particular things, after all, cannot be individuated or reidentified by reference to their "natures." Natures are, in principle, designated by general predicates, open to multiple instantiations. Hence, although the delaying question may be raised, it cannot serve Aristotle's purpose. Still, it must be met somewhere. For the moment, I postpone the contest and the reckoning. But by that same reckoning, I must add, as a promissory note, the following corollary of (2.1) and (2.6):
(2.11) among natural languages, reference is viable without essentialist assumptions.
(I shall return to the issue in the next chapter.)
I come finally to the principles of noncontradiction and excluded middle. They are obviously of very different sorts. For one thing, one supposes the principle of noncontradiction cannot be coherently denied. But if that is so, then, it seems, (2.6) must be false. On the other hand, the denial of excluded middle does seem to be straightforwardly possible: many-valued logics are not thought to be impossible.
The difficulty regarding noncontradiction is not a strenuous one: we need only introduce a distinction between uninterpreted sentential formulas and interpreted or meaningful sentences. The "necessity" ascribed to the principle of noncontradiction applies only to such "uninterpreted formulas" as "p and -p ." (This accords with Wittgenstein's formal treatment of contradiction and tautology in Tractatus .) Contradictions,we may say, are (interpreted) "sentences" (or token sentences used in arguments in speech-act contexts or such arguments), not (uninterpreted) "sentential formulas." The upshot is, the "necessity" of the (uninterpreted) principle of noncontradiction:that, necessarily, we cannot consistently jointly affirm and deny the same sentence, is now entirely benign. Alternatively put, there are no operative "contradictions" that do not depend on interpreting the terms of given sentences. But then, any
seeming contradiction can be offset by reinterpreting the substantive terms in question: for instance, as by reconciling the corpuscular and wave theories of light. In that sense, no (substantive) contradiction is, unconditionally, such. We may not care to pay the price of escaping contradiction, but that is an entirely different matter. Here, it seems more reasonable to side with Quine then with Aristotle (that is, with Quine's treatment of the analytic/synthetic distinction). In any case, the argument shows the sense in which Metaphysics Gamma may be resisted. (By interpreting, I mean here no more than construing sentences and the like as meaningful, having meaning, being assignable semantic import or content, being thus used in spontaneous discourse.)
Furthermore, if essentialism is no more a necessary truth than is the changelessness of reality ((2.1)), then we may say that the principle of excluded middle (not construed as a mere theorem): viz., that, "necessarily, for any predicate 'F ' and for any real denotatum 'a ', either 'F(a) ' or '-F(a) ' is true," is not an unconditionally necessary principle at all. Other philosophers, Dummett for instance,[12] while apparently denying the necessity of excluded middle for technical reasons, hold instead that tertium non datur is necessarily true. By tertium non datur, Dummett means: given that our sentences are suitably interpreted and brought to bear (in a testable manner) on the way the world is, "necessarily, it cannot obtain that neither 'Fa ' nor '-Fa ' is true." But Dummett, for one thing, restores the force of excluded middle (or bivalence), and, for a second, does so arbitrarily and without legitimating the modal operator. (That is, he fails to treat tertium non datur as a principle, not a theorem, as he had treated excluded middle. On my analysis, this is because Dummett is a "externalist": the finding bears on the important difference between Dummett and Putnam.)
Of course, all this is a preamble to the deeper question whether and at what price the heterodox doctrine of (2.1) can be globally defended. Only its bare eligibility has been ensured.
It needs to be said that that is a considerable achievement, however. It's not merely that the archic thesis has been defeated. It's also worth noting that we have made a productive beginning, philosophically, without philosophical presupposition —if, by "presupposition" in this strongest possible sense, we understand an initial affirmative commitment regarding first principles.
I suggest marking the novelty of this beginning by a sort of triangulation. Aristotle presupposed a suitable match ((2.4)) between the changelessness of reality and the power of cognizing reason (nous ) to
discern such fixity. Hume accepted that constraint on philosophy, in rejecting its defensibility in the version Descartes proposed, and his skepticism about the prospects of philosophy followed directly. Hegel accepted Hume's objection and so began the Phenomenology in a "presuppositionless" way (as Hegel himself suggests), that is, without any explicit first principle. Nevertheless, Hegel supposes that our very capacity to reflect on whatever appears to be genuine knowledge—wherever and however initiated—does harbor an intrinsic power to correct itself progressively, in the direction of ultimately discerning what, necessarily, knowing is ("absolute" knowing). Hence, although Hegel begins as we have (if I may speak thus) with the way things merely appear, he "presupposes" that the very structure of what is disclosed through the "necessary" sequence of reflecting on the way the world appears will, in the limit, capture the "necessary" match (or identity) between thought and that which is thought. (But he does not explain how he knows that this is so, or how he could confirm it, or even precisely what he means. I don't deny that Hegel affirms that the dialectical process of rational reflection is marked by "necessity." But the meaning of that necessity is a mystery; it is not even clear how it contrasts with contingency.)
In a most ingenious way, therefore, Hegel begins—not with an explicit first principle of Aristotle's sort but—with a first principle that affirms that, hidden in the process of reflection itself, there must be a (first) principle (operative, somehow, evolving through the successive stages of its own activity) adequate for discerning the necessary structure internal to the entire unfolding process. To view our puzzle in this light is to appreciate the utterly unguarded beginning of beginning with the flux. We begin with the flux—so we do begin. (We may even admit "principles.") We begin therefore, without presuppositions, without first principles: we elude the snares of Aristotle, Hume, and Hegel (if, indeed, that is Hegel's claim). Hence,
(2.12) an-archic philosophies are presuppositionless.
That is, they do not begin with unconditionally necessary truths: but they also do not claim to arrive at necessary truths (in the very different ways Kant, Hegel, Husserl, and Heidegger insist). There you have the unique distinction of Protagoras's early adherence to appearance and the flux: the target, jointly, of Plato's and Aristotle's insistence on adhering to the philosophical canon.
If, now, in the light of (2.12), we recall the distinction between "indicative" and "modal" invariances introduced earlier in this chapter, we can always pair an indicative mate with every archic distinction claimed: thus, admit (where wanted), "principles" but not "first principles," cognitive "sources" but not privileged "foundations," "uniformities" but not necessarily exceptionless "universals," "natures" but not "essences," and so on. This should obviate the need for repeated warnings about the "indicative" import of the distinctions that may be needed as we proceed. None are intended in the modal sense that archism favors.
There are other uses to which the analysis of noncontradiction, excluded middle, and numerical identity may be put. But they may be postponed for the time being. I have now shown the (provisional) viability of the an-archic policy. I have by no means confirmed it, but I have drawn attention to an unguarded and unconfirmed assumption in Aristotle's attack on the doctrine of the flux and on any idiom restricted to appearances, namely, that the seemingly formal principle of noncontradiction is actually dependent on the archic thesis itself—that the application of the principle to numbered ousiai entails the admission that particular things necessarily have a changeless nature. Here, I say only that the argument has not been supplied and that its denial is not obviously self-contradictory.
It is also to the point to remark that, for Peirce,[13] both the principle of noncontradiction and the principle of excluded middle fail to apply when the pertinent linguistic distinctions are not yet in play (for instance: when the predicate "human" obtains but not yet "male" and "female"). Later (in chapter 7), I shall conclude this part of the counterargument by demonstrating that the individuation of particular things can be coherently characterized in such a way that the conditions for reidentifying those particulars does not entail changeless natures. (The question is generally neglected.) In doing all that, I shall have brought the entire account into reasonable accord with construing the flux in a historicized way. So we have made a start. (Flux, I should say, is not a chaos, as Parmenides supposes, and it need not be historicized, as the example of Protagoras makes clear. But the full meaning of these possibilities must wait for part II.) We may draw the web a little more tightly, however, if we press (2.12) to yield:
(2.13) an-archic philosophies are neither externalist nor transcendental nor privileged.
Chapter III
Reference and Predication
The complexities of the line of reasoning I pursued in the preceding chapter obliged me to retreat to conditional arguments. The important claims I wished to establish were (2.1) and (2.4)—and (2.7)—but I had to settle for stopgap alternatives to (2.4). I drew out narrower corollaries that might have been conceded in their own terms. But those attenuated corollaries ((2.2), (2.3), (2.5), (2.6)) were restricted truths that did not really strengthen the larger claims, or else they ultimately depended on them. It hardly matters, since they have their own contribution to make. Still, it's natural to proceed in both directions: building conditionally toward the confirmation of (2.1) and (2.4) and drawing out entailments, partial or entire, that are advantageous to collect. It would do little good to try to establish the larger claims—particularly (2.4)—if the lesser linkages were not easily seen to hold, since it is hard to imagine a counterthesis to (2.1) more forceful than Aristotle's. For example, if de dicto and de re considerations were strongly disjoined—as in supposing, against (2.5), that the advocacy of a bivalent logic was a matter altogether independent of essentialism or other de re considerations (speaking in accord with Frege and Dummett, for instance)—then the argument being fashioned could be effectively challenged at square one. But it is very difficult to suppose that the assignment of truth-values to statements alleged to be about the way the world is is altogether independent of our views of the way the world is . This is especially clear if one admits that there is no formal reason for supposing that the bare conception of a many-valued logic (many-valued truth-values) is incoherent. What
other constraints on the choice of truth-values could there possibly be besides constraints assumed to be drawn from the nature of knowledge or the nature of the world or the nature of our knowledge or understanding of the world? (Bear in mind that considerations of consistency, coherence, relevance, and the like arise in many-valued logics as they do in bivalent logics.)
It is reasonable, therefore, to seek out independent claims that may outflank the threat of an ever-lengthening chain of conditional linkages that draw us further and further into side-issues. It would be very helpful if there were a maneuver or two that could confirm at least a part of what was needed to establish (2.4) firmly—the mutual implication of the structures of language and the structures of the world—without invoking the intervening conceptual dependencies I allowed to surface in the last chapter. As it happens, there are resources equal to the task. The situation is not desperate in the least.
The thing is this. All philosophical strategies, whether archic or anarchic (or however otherwise oriented), must come to terms with the salient conditions on which truth-bearing discourse depends. (Let me add, for convenience, that I shall use the somewhat unfamiliar term constative[1] —reminiscent of Austin's philosophical usage but not committed to his account—to signify the function of all speech acts by which we make statements or assertions or truth-claims or truth-like claims, or affirm or deny that something or other is the case. Its sense includes but is not restricted to that of the French verb "constater " used in giving testimony in court or declaring with assurance.) On any pertinent theory, the conditions in question will include conditions for the success of the "speech acts" of reference and predication: that is, the success of those linguistic acts by which (reference ) we identify something as the referent of our discourse (that about which we are speaking) and by which (predication ) we predicate what we mean to affirm of it. Reference and predication are peculiarly difficult to analyze but they cannot be dismissed. They are as essential to speech as anything that could be named. (Nevertheless, some suppose that it is barely possible to conceive of a formal alternative to referential and predicative languages: for instance, mereologically organized languages, languages that capture what we capture in the usual way but [now] by devices for part/whole relations that permit us to retire both reference and predication. My own intuition is that such a language would, for humans, be completely parasitic on the resources they would replace.)
In particular, it is uncompelling, I claim, to characterize reference and
predication in purely linguistic or logical or syntactic terms—opposing (2.4). Nevertheless, that is the way they have been standardly addressed in the strong analytic tradition that includes Frege, Russell, Wittgenstein, Quine, Davidson, and Kripke. They cannot be adequately treated in de dicto terms without attending to their de re complications (that is, without attending to the import of their actually being uttered; used in performing [token] speech acts in the real world.)
Alternatively put:
(3.1) the analysis of natural languages is not an autonomous discipline; as actual utterance, actual activity, language is an inseparable part of the world it represents or an abstraction from it.
I believe (3.1) may be directly confirmed if reference and predication can be shown to be impossible to analyze without addressing the indissoluble linkage between their de dicto and de re features, and, I claim, that can be shown in the strongest possible way. It is, of course, already implicated in the doctrine of symbiosis ((2.7)–(2.9)). Hence, if (3.1) can be independently confirmed, then (2.4) and (2.7) will be considerably strengthened as well. So a good deal rides on probing (3.1) by way of reference and predication.
Consider reference. Speaking unguardedly, reference conveys a speaker's intention to affirm or deny that this or that is true of what he or she effectively singles out (in the world) as the referent about which one affirms or denies what one affirms or denies. Reference, then, cannot be altogether independent of predication. Furthermore, reference is intended in a cognitively robust way: that is, it is "uttered" in order to single out effectively—as numerically distinct and reidentifiable—that (the "referent") about which the speaker is speaking. This is at least the core notion: referents, let us say, are the grammatical or "logical" accusatives of acts of reference, identified as such without regard to "ontic" issues. Once we see matters this way, it is extraordinarily important (and easy) to grasp that
(3.2) reference is inherently context-bound, not definable syntactically, impossible to retire (without loss) in favor of any merely predicative resource.
This is a powerful claim. If sustained, it would completely undermine any purely "extensional" theory of the truth-bearing capacity of sen-
tences used constatively (for instance, the fashionable disquotational theory of truth). That would be a grand prize, though the linkage is hardly obvious at the moment. (The full connection needs still to be laid out. I come to disquotation toward the end of the next chapter.) Let me say again: I take the "constative" function to be satisfied by certain speech acts, by the use of sentences or propositions—by "uttering" which (the speech acts), sentences or propositions are made "truth-bearing," are (themselves) dependently "uttered," are made hospitable for ascriptions of truth and falsity in the world we inhabit.
You must appreciate that, in the current analytic practice that treats mere sentences or propositions as truth-bearing—or, treats affirmation and denial ("utterance") as no more than the enactment or instantiation of the occasion on which truth-values may be rightly assigned particular sentences —utterances (the token speech acts performed by individual agents) become little more than external "operators." They may also be replaced, it is said, by a fuller specification of the (token) sentences they operate "on," or the relations the sentences enter into. (That is, the separable uttered sentences rather than the more complex token speech-act utterances.) For instance, in Davidson's account, it may merely specify the conditions of time, speaker, referential context, and the like, in which the sense of the operative sentence is made more determinate.
Effectively, such qualifications are construed as semantic ingredients of the token sentences in question (or features of the relations into which they enter), whereas I favor distinguishing between the speech-act context in which token sentences are constatively "uttered" and those sentences. Furthermore, I say, the "context" of utterance, or the uttered "speech act," is linguistically pertinent in ways that cannot be recovered from any would-be extensional analysis of uttered sentences (as in Quine or Davidson or Tarski). (I take this to have been conclusively demonstrated by Strawson's analysis of Russell's account of denoting. The matter bears directly on the supposed adequacy of any formal model of the extensional syntax of natural language promised in Principia Mathematica and meant to be improved in later logics.)
Featuring speech acts rather than sentences has the advantage of making explicit (i) the conceptual connection between the assignment of truth-values and the context of utterance, and (ii) the indissoluble link between constative acts and the still-unanalyzed nonlinguistic conditions of life within an actual community of apt speakers. My point, very simply, is that, by moving from sentence to speech act, by insisting on context rather than ignoring it, by denying the autonomy of language rather
than affirming it, by noting the way the nonlinguistic conditions of discourse bear on linguistic meaning itself, we prepare the ground for a complete acceptance of the thesis of symbiosis ((2.7)). Davidson's strategy[2] gives the impression that context is a negligible complication, one easily retired in principle. The syntactic structure of truth-bearing sentences, Davidson feels, may be crisply and adequately supplied: that is part of the motivation of his appeal to Tarski's formal treatment of truth. (It accounts for his being attracted, at least initially, to the disquotational theory of truth and, more pointedly, to the prospect of retiring reference.) I, on the other hand, cannot see how context can be neutralized for natural languages as used—in cognitively effective ways ((3.2)). Davidson does not show how it is possible. (Tarski does not share Davidson's conviction.) Reference, I say, affords a decisive test case against externalism.
The essential difficulty confronting treating reference in terms of sentences rather than of speech acts conveying speaker's intentions is this: a speaker may intend to single out a unique referent (for whatever he means to predicate of it), but he need not know which (if any) predicates are uniquely instantiated by his referent; those who account for reference predicatively (Russell and Quine, of course) must be able to give an account of reference in predicative terms, but all such accounts are either epistemically vacuous or not demonstrably adequate. In effect, this means that strategies of the second sort always fall back to whatever enables the first to succeed. The logical structure of our sentences is a function of the speech acts in which they actually function.
In the standard notational practice that links, say, Frege and Russell (and Searle, more surprisingly), the act of assertion or utterance is said to be adequately represented by the (external ) assertion sign (or operator ) "
erators—declarative, interrogative, imperative—could be "externally" applied to the same "propositions" [the "phrastic" element] affecting thereby only the relevance of [external] compliance with orders [given], answering questions [asked], assenting to or dissenting from claims [made]! The serious point of quarreling with this usage is that it fails to address the matter of the bearing of utterance and speech-act context ["neustic" aspects, in Hare's idiom] on the sense of what is uttered [the sentences]. This is just what an adequate theory of reference should make clear, though, as we shall see, it can do so only at a considerable price.)
In different ways, Frege and Russell hit on the idea that we could "improve" on the logical treatment of denotation by introducing logically proper names, names that, taken singly, would unequivocally single out one and only one "thing" in the world (one denotatum ). Frege's and Russell's strategies[3] for accomplishing this were at once hopelessly inoperative in cognitive terms and syntactically inexplicit for the purpose intended. The formal adequacy of their idealized notations depended on substantive (epistemic) considerations they could not represent in a purely formal way and they could not satisfactorily defend philosophically. (Frege's account is epistemically inexplicit, a mere promissory note; Russell's is "solipsistic," in the sense that Russell acknowledges that no one other than the speaker could possibly know, "by acquaintance," the denotatum of another's "logically proper names.") In natural-language contexts, it is pointless to ignore the cognitional role of reference; and, in a formal logic, it would be a disaster to make the syntax of sentences depend on the solipsistic intentions of speakers. (The problem goes back at least to Meister Eckhart's rejecting the naming of God and Duns Scotus's attempt to capture the individuum ineffabile .)
When Strawson offered his well-known criticism of Russell's "On Denoting," he ineluctably confirmed as well that the referents of our speech acts (in effect, what has to be specified by the referring act itself) could not be successfully captured by the supposed denotata of any sentences or propositions we might affirm. There is, in fact, only one possible strategy (if it is indeed possible) by which that simplification could have been achieved: namely, a strategy, by the use of general predicates ("indefinite descriptions"), by which we could effectively and uniquely individuate everything in the universe.
As it happens, that grand strategy has been laid out in a pretty way by Quine: Quine recommends replacing all proper names, definite descriptions, indexical terms, and the like by uniquely denoting or uniquely
designating general predicates. Of course, Quine is right in supposing that that would effectively retire reference as a necessary part of the grammar of speech acts invoked by natural-language users. But there is a difficulty. Quine's maneuver was no more than formal: it never addressed the cognitive side of reference (in effect, the issues raised by (2.4) and (2.7)). There are, however, no referring acts that are not cognitively intended. Quine pretends to analyze the formal structure of sentences used in referential contexts, but he has no way of demonstrating that that is possible.
Suppose, Quine argued, the name "Socrates" denotes Socrates, and suppose Socrates may be singled out in principle by some unique general property (captured by an "indefinite description": call it "socratizing"). Well then: install uniquely denoting expressions everywhere (or retire the labor the referential use of various expressions is meant to achieve). At one stroke, proper names and reference could then be retired in principle.
But of course, reference is a speech act (embedded in other acts) intended in a robust cognitive sense —intended, that is, to single out this or that in the actual world . Quine's "solution" assures us neither that there are at hand any such uniquely individating predicates as "socratizing" nor that humans (as opposed to God) could ever recognize them to be such. You cannot know that "socratizing" singles out the one we call "Socrates," unless you also know that "platonizing" singles out the one we call "Plato" and so on for every particular thing there is. (One grave difficulty in this regard, affecting Quine's account, is that, on Quine's own theory, there cannot be a single operative sense in which, speaking of the denumerably many things there "are," we may claim to speak of "everything there is.")
Leibniz had already considered the matter. In effect, he concluded (at least in his fifth letter to Samuel Clarke) that there was no purely logical barrier[4] against supposing that, for anygeneral predicables, if any were instantiated at all, they might always be multiply instantiated in principle. Leibniz felt, however, that a benevolent God would never allow it. Hence, he invoked the principle of sufficient reason to rescue the claim, but denied the adequacy of appealing only to the principle of noncontradiction. What Quine needs, precisely, is a demonstration to the effect that any serious opposing of his recommendation is instantly self-contradictory. He never offers such an argument. He never would have. No such argument is known to be successful. (Clearly, it would be profoundly archic.)
Just so. Unless human speakers can justifiably claim to possess God's knowledge of the world, it is impossible to retire reference altogether. Reference, we say, is invariably context-bound, incapable of being managed in any context-free way (extensionally ), whether by some formal syntax fitted to denotata and compliant sentences or by some cognitive rule applied to the range of application of all possible general predicates. The project is doomed, therefore, and (3.2) is true. (For the time being, I shall treat discourse as behaving extensionally if it is context-free, that is, not "context-bound" ((3.2))—if at least reference can be made entirely determinate by the mere use of a finite string of sentences.)
But if this is so, then an important theorem falls out as a consequence:
(3.3) the context in which "utterances" are made the bearers of truth-values is not a merely "external" locale in which sentences are rightly assigned such values.
This goes equally against the views of Quine and Chomsky. The critical proof rests with reference itself. To speak referringly (as the Leibnizian concession shows) cannot be captured by syntactic, semantic, or any other features of sentences: the meaning of what is uttered (both referentially and predicatively), "implicates" the collective practices of the enabling society in which assertion itself functions ((3.1), (3.3)). The meaning of what is (thus) uttered, the abstracted sentence itself, can only be assigned from the vantage of the speech-act context in which it is (thus) uttered. This yields an important corollary of (3.3):
(3.4) natural-language sentences are abstractions from speech-act contexts.
Theorems (3.3) and (3.4) explicate the sense of (3.1): language cannot support an autonomous inquiry; the meaning and meaningful structure of linguistic utterances cannot be disjoined at any point from the embedding consensual practices of the societies in which they natively function. The complexity of reference confirms this. The predicative intent of referring expressions cannot be made completely explicit linguistically—in principle . Whatever, in practice, is assigned as the uniquely individuating sense of referring expressions is assigned with the understanding that it is meant to mark off a referent that cannot be identified solely in virtue of whatever predicates may be supplied . The intention of a referring act is indeed to single out a unique referent. But if Quine's
formal program fails in epistemic terms (as it must), then the apt speakers of natural languages understand that the very success of their referring intentions depends on the supporting consensus of their society's more fundamental practices—in terms of which the contextual import of sentences thus "uttered" is effectively settled. This, of course, also supplies the wedge for replacing Aristotle's archism, for, now, it becomes possible to consider how particulars may be individuated as the unique particulars they are—without invoking fixed natures: successful reference is neutral to predicative uniqueness and provides the logical basis for the relevance of any predication at all. (In the idiom I shall introduce later, "individuation" takes priority over "reidentifiability" and makes it possible: reidentification presupposes individuation. Kripkean "rigid designators"[5] cannot retire speakers' intentions and cannot function successfully in causal terms alone.)
Referential success cannot possibly be ensured by the sole use of denumerably many predicates. For, if Quine's program fails, there cannot be any adequate predicative replacement. In practice, every predicative clue must be interpreted in accord with its corresponding referential intent, and referential intent must itself be interpreted in accord with the consensual practices of the enabling society in which it obtains. You may take this to be the import of Wittgenstein's marvelous notion of "forms of life" (Lebensformen ),[6] to which Strawson's and Austin's studies of the complexities of speech-act contexts effectively incline (although that was hardly their intent).
Wittgenstein's account is itself little more than an abstraction from a much richer and more adequate account of the processes of societal life (that he does not explore); but what he does offer is enough to offset the opposing picture of language I have sketched. Ultimately, Wittgenstein (in Investigations ) is an opponent of Frege and Russell, whom, at the time of the Tractatus, he so much admired. I shall, therefore, adopt Wittgenstein's notion of Lebensformen as signifying at least (i) the collective nature of the practices, both linguistic and nonlinguistic (lingual, let us say, linguistically informed but not specifically linguistic), that inform any and all effective constative acts, and (ii) the deeper "lingual" context of the consensual practices of apt speakers in which linguistic acts are indissolubly embedded—and because of which such acts succeed. In a word, admitting Lebensformen spells doom for all purely extensional treatments of natural language.
Let me acknowledge, therefore, that I coopt Wittgenstein's term of art as a term of art of my own. It serves to show, among other things, that
if, as is true, individual referents are linguistically "ineffable" when treated predicatively, then the whole of language is linguistically ineffable as well; for, as the argument of this chapter shows, language is lingual.[7] Once you see that in terms of my reading of Lebensformen, you see, for instance, that Kripke's initially skeptical reading of Wittgenstein's treatment of conforming with a "practice" (as "following a rule") as implicating that there is a rule that is followed (in thus conforming) is irrelevant, ruled out, committed to precisely what Wittgenstein obviates by admitting Lebensformen: namely, that language is a determinate system for which the constitutive and regulative rules can always, in principle, be formulated.
It is certainly reasonable to admit that reference and predication are normally successful. If they were not, linguistic communication would utterly founder. It cannot be supposed, therefore, that the failure of ordinary reference to yield to Quine's or Russell's or Frege's analysis—as by introducing logically proper names or replacing referents (and denotata ) by unique predicates, strategies that are themselves notably susceptible of extensional treatment—signifies a serious defect of natural languages. But if that is so, then, it is certainly reasonable to maintain:
(3.5) reference presupposes (but not criterially ) the consensual lingual practices of viable societies.
(By criterial, I mean that feature of discourse in which specific conditions, or marks, may be stipulated, that govern, adequately, the correct application of terms to the things of the world. Such "criteria" may be necessary, sufficient, necessary and sufficient, typical, or employed in looser ways. My only point is that speaking merely of Lebensformen is never "criterial.")
This is a deliberately bland formulation of a radical thesis. What it implies is that every merely formal treatment of natural language is inherently tethered to the encompassing collective life of particular societies of apt speakers whose (own) successful practices cannot be vouchsafed by formal means alone and cannot be confined to linguistic practices alone. I venture to add that (3.5) indirectly signifies that reference has a historical (or historied) dimension, since the consensual practices that inform effective speech acts (Lebensformen ) change, historically, as a result of actual ongoing life. But I have not laid the ground for a firm acknowledgement of this. (I am working toward it, of course.)
One begins to see the more radical possibilities of the line of argument
being pursued. Two important conclusions may be drawn from it alone:
(3.6) the meaning and syntax of sentences depend on the context of their token utterances;
and
(3.7) the context of utterance cannot be analyzed solely from the vantage of analyzing sentences or a language abstractly construed.
We are touching on the historical nature of human existence, but only tentatively. We are not yet prepared to pursue the matter. But it is clear that reference is bound to be a practice peculiarly sensitive to the historically changing nature of a society's Lebensformen . (It may be usefully remarked that Wittgenstein acknowledges the open-ended nature of ongoing societal practices, but he has almost nothing to say about the structure of such practices in terms of an understanding of history. So the puzzle deepens.)
I should warn, however, that some will find my use of Wittgenstein's notion of Lebensform(en) highhanded.[8] Certainly, the term "Lebensform " occurs only infrequently in the Investigations and (I believe) only once in the plural. Wittgenstein does not really explain his use of the term satisfactorily. Garver, for one, believes that Wittgenstein held that there was only one "form of life" for humans, not many (although we might speak of different "forms of life" for different creatures). According to this view, Wittgenstein (i) did not mean to treat the human Lebensform in essentialist terms, (ii) did mean to accommodate the variety of cultural modes of life, (iii) acknowledged that one could not state the defining marks of the human Lebensform, and (iv) treated it at once as "animal" and "transcendental"—meaning by the latter term perhaps that, whatever the variety of cultural modes, they are all constrained (but not in any Kantian sense) by a natural animal endowment that cannot be fully captured, descriptively, by whatever may be the contingent manifestations of this or that society. I mention all this to mark the scruple of my appearing to usurp Wittgenstein's term by (i') using it in the plural, (ii') displacing its "transcendental" function (without denying the intended insight), (iii') applying it to enculturation, and (iv') construing it in historicized terms. If Wittgenstein would have opposed my irreverent use, I confess I cannot see what he would have offered as the defeating argument—apart from textual fidelity. The important thing is
that Wittgenstein does not use his notion dynamically, does not explicate the relation between the cultural and the biological, does not account for the emergence of persons as culturally apt agents, does not reconcile the collective and the individual. In adopting "Lebensformen " as a term of art, I deliberately extend its use to all that, and I believe it captures what is most promising in Wittgenstein's own notion. But I do not claim textual fidelity.
Let me take advantage of having intruded the matter of history. I believe we shall eventually come to see the reasonableness of affirming that
(3.8) referring acts are inherently entrenched in the historically changing life of societies of apt speakers; the recovery of their meaning (or intent or successful function) cannot be disjoined from an understanding of their society's practices.
I cannot strengthen (3.8) at this point in the argument without going very far afield. I introduce it somewhat arbitrarily, but I do so in order to keep certain ulterior objectives before us. We are, you remember, bent on understanding certain radical claims (and the logic of those claims) now gathering strength at the end of our millennium. Theorem (3.8) is as good a clue as any about their common theme. But we cannot yet confirm it. (Its lesson will resurface in chapters 8 and 9.) For the moment, I may at least say that by entrenched I mean that feature of a natural language (I shall call it "historicized," later) in accord with which the effectiveness of speech acts depends on their being uttered in a lebensformlich context, by agents whose own linguistic competence presupposes and entails the mastery of a language. (Here, "entrenched" lacks criterial force altogether, but it allows lebensformlich and historied qualification. I may perhaps take the occasion to say also that I treat "lebenformlich " as an English adjective: I drop the inflectional endings.) An aggregate's sharing of such lebensformlich practices, viewed with an eye to the success of reference and predication, I call consensual. (Consensus need not be "criterial" but the criterial must be consensually apt.)
What I have shown in effect is that language cannot rightly be treated as an autonomous discipline ((3.1))—since reference is already lingual ((3.5)); that is, that the successful functioning of natural languages can be accounted for only in terms of the inseparability of speech acts from the lingual lebensformlich acts of an apt society of speaking agents. Once that much is admitted, we cannot fail to see that it is but a step to
historicize the lebensformlich practices in question. There would then be two decisive limitations to any "extensional" treatment of language: (i) its being embedded in lingual ways, which affects the meaning of sentences actually used; (ii) its meaning being historicized, in virtue of being lebensformlich . (The "extensional" treatment of language is tantamount to its being treated as autonomous.) In short, the discovery that reference is lingual is a discovery of decisive importance: it challenges in the most profound way every facet of analytic philosophy that relies on an extensional treatment of language—as, notably, in Quine, in the Wittgenstein of the Tractatus (properly confined), in Carnap, and in Frege. The beauty of the argument is that it does not rely on historicity but prepares the ground for its admission. Hence, by featuring the analysis of reference as I have, we see at a stroke that the most influential and admired features of Anglo-American analytic philosophy are placed in serious question. I should perhaps add that the analysis of reference may fairly be taken to be the single most strategic site for challenging the strongest claims of analytic philosophy. For instance, all the supposed virtues of "naturalism" (or what I shall later call "naturalizing," following Quine's usage) fail if language cannot be shown to be autonomous in the manner sketched.
Turn, now, to predication. There is an obvious connection between reference and predication: we cannot successfully predicate what is true of anything if we cannot fix the identity of what we are talking about. But if what has been said of reference is valid, then, admitting that it cannot be effectively regimented in the "extensional" way, neither can predicative acts. Bear in mind that by an extensional analysis I mean (thus far at least) either or both of two general strategies: (i) an analysis (conformable with the syntactic constraints of first-order logics like Principia ) of the structure of sentences (used in suitable speech acts) by which the denotata of our discourse may be specified in a cognitively effective way as the unique denotata they are; or (ii) a semantic analysis of our predicative terms (whether as replacements or not for referential expressions, but restricted to predicative position), such that we may effectively fix the extension (or range of application) of our most important general predicates. (The resources of reference and predication implicate one another.)
I should add: (iii) that (as with Dummett and perhaps the early Tarski) the supposed disjunction between sentence and context of utterance or between the underlying logical syntax of language and the interpreted sentences we normally utter encourages a certain confidence in programs
(i) and (ii) even where programs in accord with (i) and (ii) are not actively pursued. Strategy (i) may be illustrated by Frege's notion of uniquely denoting names or Russell's notion of "logically proper names" keyed to the "sense-content" of a direct "acquaintance" with the world, and strategy (ii) may be illustrated by Davidson's notion of the conditions for ascribing truth to sentences in accord with his reading of Tarski's semantic conception of truth), or by Fodor's account in which complex predicates may be composed (without remainder) from simple predicates and in which the extensions of the most important predicates are "known" in a "nativist" or hardwired way.[9]
A fully adequate extensional account of reference would require some suitable union of strategies (i) and (ii), possibly along Quine's lines. But that is not our concern, since, on the foregoing argument, the project is doomed from the start. The point I wish to salvage is simply that, if the linkage mentioned holds, then our inability to treat reference extensionally cannot fail to infect predication as well.
I hasten to add (iv): that "extensionality," or an "extensional" analysis, may be confined syntactically to uninterpreted "sentential formulas"—hence, to the rules of first-order logic. But, by itself, that would contribute little or nothing to our understanding of philosophical logic —that is, the analysis of those questions, in particular, that bear on reference, predication, individuation, numerical identity, truth, meaning, de re and de dicto modalities ((2.4)). For, we would have to be able to show that the analysis of sentential formulas was, pertinently, prior to the analysis of interpreted sentences, but that is precisely what I was objecting to in opposing the modern logician's tendency to disjoin logic and natural language—in discussing Aristotle's use of the principles of noncontradiction and excluded middle (in chapter 2). I agree with Aristotle about the indissoluble connection between logic and language, but to admit that is not yet to admit Aristotle's archism.
There are deeper puzzles confronting the extensional treatment of predicates. They have to do with resolving the famous problem of universals . (By universals, I intend abstract denumerable entities that (i) are the denotata of general predicates and (ii) exist. I oppose the venerable thesis that there are universals or that they are needed in an adequate philosophy; but the argument must wait.) The problem is as old as (even older than) the philosophies of Plato and Aristotle. The issue takes the form of a triple question (notably routinized in the Middle Ages). Given the use of general terms, are there, one asks, "real generals"[10] in the world? (The phrasing is favored by Peirce—and answered by Peirce in
the affirmative.) Secondly, what relationship holds between the general predicates of our language and the actual attributes of independently existing things? And finally, on what grounds may we claim to know the truth about either or both of these matters?
The history of the question is a disaster. The medievals offered three lines of argument: realism regarding universals, nominalism, and conceptualism. By the first (realism ) is intended all those theories that maintain that general terms (for instance, "blue," "round," "hirsute," "baroque," "just," "beautiful," "human," "canine," "female"—or, better, some reasoned selection therefrom) do account for, by way of universals, or possibly even designate, the real general attributes of things. Peirce, who followed to some extent (and "corrected") Duns Scotus's subtle account, tried to recover a strong "realist" claim for the sciences, by wedding the analysis of general predicables to the realism of the laws of nature. (He took "generals" to be real, but he denied that universals exist .) A modern attack on all such realisms—directed particularly against Aristotle and inductivists—was mounted, for associated reasons, by Popper, who believed that, inasmuch as nature was unfathomably deep, human inquirers could not reasonably suppose their own nomological conjectures ever satisfactorily captured the underlying structure of the real world. Popper might have answered, affirmatively, the first of the three questions posed and, more cautiously (along "realist" lines of Peirce's sort), the second and third.
The trouble with the realist reading of "universals" rests with the plain fact that the cognition of "real generals" cannot be made to function in any criterially serious way when fitted to what are usually called "universals." It's all very well saying that our general terms correspond to real universals, but how could any such claim be confirmed? Plato's dialogues (particularly Republic and Parmenides ) have collected all the difficulties known to infect the thesis, without ever explicitly providing for their assured resolution; Aristotle's solution rests, quite scandalously, on the notion that, by exercising its native power prior to the work of "science" or episteme (which ultimately depends upon it), pure mind (nous )[11] simply internalizes (by brute intuition) a "noetic" grasp of the independent changeless ("universal") structures of things. Of course, contemporary views of the constructed or idealized standing of lawlike universals tend to ignore any pointedly perceptual or intuited analysis of universals. (Russell was committed to a kind of Platonism.) In a word, theories about scientific laws tend to universalize their scope, but they do not characteristically address the specifically epistemic condition un-
der which "universals " (or "real generals") can rightly be said to be actually discerned. Straightforwardly, then: the problem about general predicates and real universals is the analogue of that other vexed problem, the one about "propositions" and their corresponding "facts."
Nominalism fails outright, if by "nominalism" is meant the thesis that we do indeed use general terms cognitively, in spite of the fact that there are no general traits in the real world and no general concepts in the mind either.[12] The trouble with nominalism is simply that, on its own assumption, it cannot account for the smooth, spontaneous (massively successful) extension of general predicates in natural-language use—beyond any postulated paradigms of use or original convention. For, if any "nominalist" convention functioned smoothly within a society's practices (without new conventions being made at every moment of application), then, of course, some sort of "realism" (regarding general predicables) would be vindicated; if it failed, because (on the hypothesis) there was nothing truly "general" in the world or in thought to invoke, then nominalism could not possibly account for the plain successes (cognitive, remember) of science and ordinary linguistic communication. One sees the difficulty, for instance, in Goodman's well-known nominalism, which, although uncompromising, has nothing to say about the cognizing process itself—unless to deny that there are any "general traits" that could be discerned in any two compared things that could then be discerned as well in an independent third.
Conceptualism (a subtle medieval compromise that Peirce assigns to Duns Scotus) is also troublesome because it leans in the direction of nominalism, supposing that universal terms answer to universal concepts in the mind but not (perhaps) to any general structures in nature itself. And yet, it also leans in the direction of an attenuated realism, since the smooth functioning of such terms among chance aggregates of apt native speakers demands an explanation. Thus construed, conceptualism requires that there be a structure common (enough) to the concepts of each of us to support a valid and viable use of general terms, but it is hard to see that that could be convincingly separated from the thesis that there are "real generals" in nature. (Scotus inclines in the latter direction. The original thesis arose in a different way, earlier in the Middle Ages, in Roscelin and Abelard. But it suggests the sense in which Goodman cannot escape an implicit conceptualism.)
The upshot is that the classic answers to the problem of universals fail. One canny way of putting the difficulty is this: the standard answers suppose that there must be an externalist relation of some sort (in effect,
the realist view), a correspondence (or other relation) between our terms and the structure of independent nature (or, "what there is" or "what is real"), but it is impossible to prise our terms apart from that connection in order to test whether it holds or not ((2.4)). There is no operative way to answer the question. The difficulty is analogous (as I say) to that between true propositions and real facts: the "facts" to which true propositions "correspond" are identified only by means of those propositions, and vice versa. On the argument, we cannot "exit" from language.[13] We can never "examine" the would-be correspondence—or any putatively weaker relation, for that matter—between what our propositions "mean" and the "intelligible" facts to which they are said to conform. The argument is compelling. (I return to the matter of a realist reading of "external" relations in chapter 4. Notice, please, the change of term.)
Nevertheless, it is not the last word. For one thing, the strong "correspondence" theory—the one so masterfully developed in Wittgenstein's Tractatus —appears to be completely unworkable in epistemic terms, because of the very paradoxes Wittgenstein himself uncovered. But if the notion of a correspondence between two independent orders were rejected, then it would be open to us to consider entirely different solutions to the "problem of universals" (in effect, the puzzle of predication). The fact is that, in terms of the history of modern philosophy, nearly all forms of that sort of realism that requires a correspondence theory of truth are pretty well rejected. In the Tractatus, Wittgenstein himself supplies no correspondentist theory of cognition. None, it seems, could escape the paradox of "picturing" the world.
In this very mild sense, nearly all of modern philosophy is post-Kantian —that is, holds that, eschewing all cognitive privilege (hence, Kant's own apriorism), there is no principled epistemic disjunction between an independent actual world and the world we experience and claim to know . Call this thesis symbiosis ((2.7)), since it is, as such, neutral regarding archic and an-archic denials of the correspondence claim. Symbiosis signifies (i) a holism within which the "external" relation between cognizing subject and cognized object is first posited, and (ii) a holism, therefore, the antecedent structure of which (whatever that may be conjectured to be) cannot be discerned by way of the external relationship thus posteriorly posited. Later, I shall characterize that relationship as "constituted," "constructed," even subject to "historicized" reconstitution—for the moment, however, those refinements are not needed. Nevertheless, it is to the point to emphasize that, although Kant thinks
of the real world as "constituted" as an intelligible world—in some way by the native powers of cognizing subjects—he does not think of the external relation between subject and world as itself originally posited within some prior holist "space." In that sense, Kant is not committed to—opposes, in fact—the doctrine of symbiosis. (It is precisely in that sense that analytic philosophy has found a convergence between Kant's "empirical realism" and its own externalism, although Kant of course is no externalist. Hegel, by contrast, accepts a form of symbiosis, although it is [notoriously] difficult to say what he understands by the structure of any external relationship between subject and object.)
Nothing much depends on the epithet "post-Kantian," except to link a conceptual strategy to the actual history of philosophy. I use it in a deliberately thin way; for, of course, Kant was quite sanguine about the inescapability of a strong archic substitute for correspondence. In particular, he believed that an invariant structure could be apodictically imputed to the contingencies of the phenomenal world (the work of the "synthetic a priori "). But it is entirely open to us—since (2.1) is not demonstrably incoherent or self-contradictory (hence, is valid)—to theorize that human cognitive capacities (within the symbiosis just tendered) cannot be shown to possess invariant structures, may in fact be subject to radical historical variability in the way of conceptual understanding. (This goes against Kant.)
"Symbiosis" (in the "post-Kantian" sense) signifies, therefore, no more than the denial of any principled disjunction between "the way the world is" and "the way the world appears to us to be" (in effect: Hegel's thesis in the Phenomenology )—between what (illicitly) would have been called "noumena" and "phenomena" in Kant's idiom, or between "metaphysical realism" and "internal realism" in Putnam's, or indeed between the "subjective" and "objective" elements of the phenomenal world.[14] (I shall return to the analysis of "subjects" and "objects.") I have, in effect, been arguing that a careful analysis of reference and predication—a fortiori, of individuation and numerical identity—confirms that
(3.9) the success of reference and predication signifies that, in natural-language contexts, they are indissolubly symbiotized.
This, of course, is tantamount to saying:
(3.10) no principled disjunction between "alethic," "epistemic," and "ontic" discourse can be legitimated.
Theorems (3.9) and (3.10) may be read as corollaries of (3.1). (I examine the "alethic"-"epistemic"-"ontic" distinction more closely in chapter 4. Here, it is enough to note that it is symbiotized.)
It may be convenient to add that what I have called the "externalist" view also appears classically as pre-Kantian (treating figures like Descartes and Locke as exemplars). What is instructive is that, as with Kant himself, strict "Kantians" are hardly different from pre-Kantians wherever they insist on the archic or the privileged or an asymptotic approach to the final fixities of reality.[15] (Apel and Habermas, for instance, are excellent specimens of the kind of convergence I have in mind. But so are Peirce and Popper.) Epistemically construed, the correspondence theory of truth is, by definition, an externalist theory matching constative claims and the real nature of the independent world: hence, it cannot help but presume some form of cognitive privilege (as indeed both Descartes and Locke betray). It might be argued, in this regard, that Moore's notorious "Refutation of Idealism" (that is, the refutation of a condition on which symbiosis depends) is designed to show how a correct reading of Kant's first Critique confirms that it too is ultimately committed to an analogue of a form of externalism.[16] My point is not that this is a preposterous claim (though it is indeed preposterous), it is rather that there is some justice in what Moore alleges, since, within a symbiotized conception (which neither Moore nor Kant would espouse), Kant is committed to reclaiming what holds invariantly (transcendentally) of what is empirically external to individual cognizing agents. This is precisely what contemporary Kantians fail to acknowledge—or defend. (What, we may ask, is the Kantian subject?)
The charm of this line of argument lies in the fact that the theorems just mentioned are as nearly presuppositionless as is humanly possible. That is, they are very difficult to resist, on any reasonable review of the operative complexities of natural-language reference and natural-language predication. One has only to reflect that theorizing (instead) about the nature of truth—the point of the correspondence theory—cannot be "presuppositionless" in the same sense. This is why disputes about the "disquotational" theory of truth—in Quine, in Davidson, in Putnam, in Kuhn (in effect)—cannot be disconnected from substantive views about the way the world is or about the way we come to have knowledge of it.
When, for instance, Davidson and Putnam claim that Tarski's semantic conception (of truth) is "philosophically neutral," they can only mean what they say if Tarski's account is read disquotationally (in effect:
if the predicate "true" is judged to be logically otiose).[17] But, for one thing, Tarski's extensionalism can hardly be merely disquotational; and, for another, the disquotational theory presupposes, although it does not affirm—not philosophical neutrality, but—a very confident belief that a massive part of the core of our ordinary truth-claims cannot possibly be mistaken (which, of course, is Davidson's view), or cannot possibly be meaningfully subjected to epistemic doubts (which is Quine's view in speaking of "holophrastic" sentences). This, then, neatly confirms the sense in which I earlier suggested that the solution to the puzzles of chapter 2 could be secured by turning to examine the puzzles of reference and predication. (I return to disquotation in chapter 4.)
You see, therefore, the strategy that begins to collect along the lines of (3.8). For the sake of explicit clarity, I draw the obvious conclusion:
(3.11) the an-archic reading of symbiosis is coherent and congruent with the general features of reference and predication.
I cannot justify pursuing the strategy just yet. Still, it does allow me to draw out the analogy with what has turned up regarding reference; that is:
(3.12) predicative acts are "entrenched" in the historically changing life of societies of apt speakers.
For our present purpose, the analogy is the important thing. I cannot yet explain what an adequate account of (3.8) and (3.12) require, or, indeed, the full meaning of "entrenched." But you may already see that a proper answer would involve an an-archic interpretation of symbiosis along historicized lines. Nothing is lost, however, by getting clear about the argumentative linkages being recommended. On the contrary, it helps to allay reasonable worries about the possible arbitrariness or drift of the strategy I am pursuing. I am, in effect, pursuing and illustrating at one and the same time the argumentative strategies open to an-archic philosophies.
Short of insisting on the specifically historicist cast of (3.8) and (3.12)—which, I concede, has radical implications—we can now confirm, for predication (viewed as a speech act), the counterpart of (3.5):
(3.13) predication presupposes (but not criterially) the consensual practices of viable societies;
and, of course, trivially:
(3.14) predication is inherently context-bound.
The sole point I wish to press just now is that the resolution of the problem of universals, as far as the an-archic program is concerned, requires that it be located within the space of symbiosis and in accord with (3.8) and (3.12). But thus far at least, there seem to be no strenuous objections against attempting to do that. On the contrary, what I have shown (or better: what I have sketched with an eye to showing) is that an archic resolution of the problem must be defective, that the classic answers to the problem of universals presuppose some strong form of correspondence or surrogate relation, and that, if it is to be resolved at all, the problem of universals will require some option along an-archic lines. That is,
(3.15) "real generals" are necessary for the successful practice of reference and predication.
That I take to be the direct upshot of our analysis of reference and predication. (By "real generals " [Peirce's term], I mean predicables (i) that admit of multiple instantiation but (ii), although real, do not exist . So real generals are not universals . The full force of these distinctions must wait for chapter 6. To say that "generals" are real is not yet to supply the sense in which they are real. That is certainly a reasonable caveat.)
On the argument thus far given, we may conclude that
(3.16) the structure of context-bound discourse cannot be specified in linguistic terms alone.
By context-bound I mean (adding to the sense of (3.2)) that (i) the meaning of sentences is a function of their speech-act contexts ((3.3)–(3.4)), and (ii) their contexts of utterances are not merely linguistic and are inseparable from the world—they form a part of the world ((2.4), (3.1)). Alternatively put,
(3.17) speech acts are lingual, not merely linguistic.
Theorem (3.17) leads directly to a very strong finding, namely, that
(3.18) the communicative success of both reference and predication is lingual; that is to say, incapable of being managed syntactically, algorithmically, criterially, or in any related way.
Reference and predication are intrinsically informal. A great deal of the labor of twentieth-century analytic philosophy has assumed that this was simply false. Theorems (3.16)–(3.18) explicitly return us to (3.1). (I return to the matter of context in chapter 8.) But what is becoming clearer is the sense in which the an-archic strategy means to give up every possible form of privilege and invariance, means to fall back to characterizing the world in ways adequate for whatever we discern in first-order discourse, in ways that approach the limit of presuppositionless discourse. In effect, we have now stalemated the archist and the externalist (and, earlier, the postmodernist).
Chapter IV
Truth-Values
The principal claims I have so far confirmed are: (2.1), to the effect that real invariance is not a necessary presupposition of every coherent account of the world or (a fortiori ) of our knowledge of the world; and (3.1)—or, alternatively, (3.9) or (3.12)—to the effect that the analysis of language cannot be separated from, thus is symbiotized with, the analysis of the world (which it is said to represent). The rest of what has so far been confirmed is, however important, of lesser interest.
If we reflect on these two claims only, certain additional, quite powerful theorems fall into place. For one,
(4.1) there is no necessity of any kind for restricting constative discourse to bivalent truth-values.[1]
If (2.1) had been false, then, as Aristotle correctly observed, a strong bivalence would have been unavoidable: every well-formed assertion would have had to be, disjunctively, either true or false. One could perhaps have postponed the inevitable—as Dummett attempts to do along so-called intuitionistic lines—by holding that the principle of excluded middle need not (yet) apply to syntactically well-formed statements that cannot (yet) be shown to be decidable, that is, such that determinate procedures for confirming their truth or falsity by finite operations may be explicitly formulated or are at hand. But, although Dummett thus retired excluded middle as a necessary and universally operative principle (that is, the principle that, for any "a " and for any "F, " "Fa " is
necessarily either true or false), he automatically restored bivalent values wherever decidability obtained—that is, he installed tertium non datur unconditionally (the principle that, on the condition of satisfying decidability conditions, it is then not possible that "Fa " be neither true nor false).
The important point is that Dummett resolved the question of bivalent values by some deep assumption about the way the world is (or must be), without explaining just how he arrived at that reckoning or how it justified (legitimated) precluding (or necessarily precluding) making a free choice (in this or that domain) of some array of many-valued truth-values or truth-like values that depart from tertium non datur . For example, one might have denied that "true" and "false" are exhaustive and exclusive alternatives; one might have dismissed "true" and kept "false" and then introduced a continuum of intermediary (many-valued) truth- or truth-like-values ("reasonable," "plausible," "apt," and the like) that could still be treated as ranging over "decidable" claims (open to being validated or invalidated) in ways strongly analogous with the procedure replaced. Constative practices in accord with such a policy might be very much favored in interpreting artworks (interpreting Hamlet, say), and whether they could be extended to other sectors of inquiry convincingly would surely remain an open (and entirely reasonable) question. Alternatively, one might reasonably claim that, in opposing excluded middle, Dummett already invoked an externalist assumption on the basis of which he was obliged to confirm tertium non datur . That is, he had already construed decidability in terms of a bivalent logic. Hence, in a real sense, Dummett never opposed excluded middle on the strength of a principled choice regarding truth-values.
If we ask ourselves why we should take (4.1) to be true or false, we see that there is nothing in the bare meaning of the predicates "true" and "false" that entails that affirming (4.1) could not fail to be contradictory or paradoxical or the like. The only way to resist (4.1) effectively is to require that the norms of truth should conform to what we suppose the actual structure of reality could (or must) support, and then insist that that entails the necessity of bivalent values.
That, you will remember, was the central issue of chapter 2. We were concerned there to answer the question of whether reality was necessarily invariant or necessarily possessed invariant structures (Aristotle's question). Here we are concerned rather with options regarding the assignment of truth-values to different ranges of constative discourse. There, we were occupied primarily with ontic (as well as epistemic) mat-
ters; here, we are more concerned with alethic (and epistemic) matters. I had in fact offered (2.4)–(2.10), which concede in various ways the (symbiotized) inseparability of what I am now identifying as "ontic" and "alethic"; and I had already explicitly arrived at (3.10), regarding the conceptual linkage between alethic, epistemic, and ontic distinctions. That is tantamount to affirming (4.1), for it is now impossible to claim that bivalence (either excluded middle or tertium non datur ) is necessary solely on the strength of the meaning of, or formal relations between, the terms "true" and "false." The acceptance of (2.1) leads directly to accepting (4.1). Hence, in chapter 2, we had already anticipated (4.1): after all, the denial of (4.1) is essential to Aristotle's argument. It is also essential to any "Fregean" program, and, in the present setting, it constitutes an archic (and "privileged") maneuver. Put another way: Dummett's "principled" objection to excluded middle has, contrary to appearances, absolutely nothing to do with any specifically alethic commitment: it affords no more than a negative scruple (decidability) imposed on whatever we offer in the way of an epistemically responsible account of any part of inquiry. (In this sense, Putnam and Dummett were arguing at cross-purposes in opposing "metaphysical realism.") This is also why Dummett's endorsement of tertium non datur is similarly "unprincipled."
The argument is a modest one, but it is also of considerable power. For, if (2.4) and (4.1) are conceded, then it is also true that
(4.2) the choice of truth-values (or truth-like values) assigned, as a matter of policy or principle, to any sector of inquiry is a function, under symbiosis, of what we take to be the nature of the domain in question.
For example, if, as Peirce affirms, reality is in some measure objectively "vague" and "indeterminate," then (as Peirce argued) noncontradiction may not apply at every point of constative discourse (as when, say, we use notoriously vague predicates—"bald"), and excluded middle will not apply everywhere (as when we use relatively indeterminate predicates—"human," say, where this has not yet been made sufficiently determinate to permit us to decide, from the sense of "human," whether "male" and "female" apply disjunctively). Alternately, in interpretive contexts (as remarked), we are sometimes prepared to concede (as in reading Hamlet ) that there may be valid interpretations of a given artwork that are incompatible with one another (on a bivalent logic) yet
separately compatible with the work in question: we are not prepared to dismiss one or the other merely because they cannot be reconciled in a single interpretation. Art, we suppose, may be of such a nature that such tolerance counts as a reasonable policy. The essential point is not the mere coherence of the policy but the reasoning by which it may be recommended or opposed.
For my present purpose, it is important to see that the judgment that we should or should not subscribe to bivalent values depends entirely on our characterization of the field surveyed. We may hold that interpretive judgments (of the sort in question) are not "incompatible" but only incongruent (let us say), in the sense that (i) they would be incompatible or yield contradictories on a bivalent logic but not now (although they could not of course be integrated in a single perspicuous interpretation—except trivially by a conjunction of disjuncts), if we replaced or supplemented bivalence with a suitably formed "many-valued" logic, and (ii), in fact, a many-valued logic suits the field of inquiry better than a straightforward bivalence.
On the argument, the retreat from bivalent values is not merely an "epistemic" consideration but an "ontic" one as well—that bears on the "alethic" issue in accord with (3.10). By alethic, I mean matters that concern only the choice, assigned meaning, and formal constraints on the use of truth- and truth-like values; by epistemic, I mean matters that concern pertinent evidence or grounds for ascribing truth-values to some sector of inquiry; and, by ontic, I mean matters that concern the putative structure and attributes of some part of the world. These distinctions systematically implicate one another and cannot be disjoined in principle without invoking privilege ((3.10)).
It comes as a surprise that the necessity of abiding by bivalent truth-values may be so easily set aside. The dominant view in the history of philosophy—from Aristotle, say, to Frege and Dummett—is that either departing from excluded middle produces self-contradiction somewhere in our discourse (Aristotle) or that, where a departure seems possible and imminent, one should find an alternative way of characterizing the alethic features suited to the field. For example, although Quine's notorious doctrine of the "indeterminacy of translation" appears to repudiate excluded middle, Quine himself demurs—quite arbitrarily it seems (perhaps also inconsistently), for he insists that, there, is "no fact of the matter"[2] —in a sense that does not produce truth-value gaps. Stalnaker begins his instructive discussion of conditionals by affirming (in accord with Frege) that there are just two truth-values, True and
False. But to the question, "What are they? What are these truth-values?" Stalnaker replies (avoiding Frege's odd answer), "It doesn't matter; all that matters is that there are just two of them." Be that as it may, once we accept (2.1), (4.1) follows; once we admit that, it follows as a consequence that
(4.3) many-valued logics need not be incoherent.
Of course, many philosophers would have been willing to concede (4.3) straight off, without any of the plodding caution I have preferred. But it is not true that they would normally have been willing to admit the coherence of relativism —or the coherence of relativism as a consequence of applying (4.3) in accord with (ontic) interpretations of "the way the world is." But relativism requires a many-valued logic. Since it is my intention to support the viability of relativism in a robust way, I regard it as an advantage to have approached the matter in the conservative way I have. Furthermore, admitting (4.2) and applying it to particular domains of inquiry, for instance literary criticism and interpretation, we move quite straightforwardly from (4.3) to the important finding:
(4.4) relativistic logics need not be incoherent.
I mean by a relativistic logic a many-valued logic that, relative to some interpreted domain of discourse, admits, as compatible, those constative claims that would (but not now) be incompatibles on a bivalent logic (whether in accord with excluded middle or tertium non datur ) or would yield contradictory propositions. The resultant propositions may (now) be termed incongruent —signifying, by that, what has just been remarked, plus a willingness to treat such propositions as "objectively" pertinent to (an understanding of) the domain in question. I have no doubt that the concession would serve very nicely in many other sectors of inquiry: in history for example, in the human sciences, among explanatory theories in high-level physics, in linguistic analysis, in philosophy itself. But I shall forego pursuing these extensions here. (I return to the question of objectivity in part II.)
I must add, however, that
(4.5) bivalent and many-valued logics are not inherently incompatible with one another.
For example, probabilistic truth-values (probabilizing truth-value assignments to evidence at hand) are standardly admitted to be compatible with bivalence: the first are said (as by Hempel) to be epistemically weak in some pertinent respect—so that, in the epistemic limit, where the probability of particular claims is either "1" or "0," bivalence obtains.[3] (I deny, however, that relativism takes probabilistic values in anything like this sense.) Relativistic logics do not behave conformably, for, where a relativistic logic is invoked, the bivalent values (or at least the value "true") are (is) retired and no longer apply (applies). "True" and "false" are, then, no longer (need no longer be) treated symmetrically (in effect, bivalently, as exhaustively and exclusively assignable to constative claims). As a consequence, (4.5) is not violated—if the range of application of the pertinent bivalent and many-valued values are suitably segregated. (Call these considerations, grounds of relevance. ) Hence,
(4.6) affirming (4.5) does not, as such, produce a self-referential paradox.[4]
Everyone has heard that to regard relativism as true is self-contradictory, but that is now seen to be a hasty finding: if (4.2) is conceded, then
(4.7) truth-value policies may be treated piecemeal, fitted to one domain or sub-domain or another as needed.
Let me spell this out a little more neatly. A relativistic logic I define as a logic that, for a given domain or range of inquiry: (i) replaces (alethically) a bivalent logic with a "many-valued logic," not a merely "three-valued" logic, one in which a third value ("indeterminate") is added to an otherwise bivalent pair ("true," "false"); (ii) can validate "incongruent" claims; (iii) treats truth and falsity asymmetrically, so that claims may be disconfirmed, as false, although (within the range defined) truth no longer obtains; (iv) replaces the value "true" with "many-valued" values ("plausible," "apt," "reasonable," and the like) that may, if wanted, be graded; (v) need not, if relevantly segregated, be incompatible with a bivalent logic; and (vi) can accommodate considerations of consistency, coherence, contradiction, and the like. (I return to "grading" in the final chapter.)
Relativism is a controversial matter, of course. The history of philosophy has long supposed that relativism is intrinsically incoherent. Hence, (4.4) is decidedly provocative. I take that theorem to be a par-
ticularly strategic finding, essential to the radical options of the philosophical currents of the end of our century. I have for that reason taken care to derive (4.4) from the most conservative sources. That has its tactical advantages, but I need to assure you that (4.4) is not a trick—and that it accommodates (in a fair way) the persistent philosophical conviction that relativism is incoherent and, at the same time, salvages a pertinent form of relativism that is demonstrably not incoherent.
The answer lies in a benign equivocation. There are really two distinct forms of relativism, viewed in "alethic" terms or viewed in terms of the connection between "alethic" and "epistemic" considerations. The two are distinguished by the simplest of devices: the objectionable form holds that "true" means "true-in-L, " where "L " signifies some well-demarcated language among disjunctively many languages, or, some "world" ("W ")—some "sector of inquiry," some "historical horizon," or some such context; the viable form, in contrast, unconditionally rejects definitions of that sort. The viable form treats "true" as having the same (alethic) sense it has in nonrelativistic logics, or, simply, as being not relationally defined. The objectionable form (call it relationalism, since it defines truth relationally) cannot overcome the self-disabling paradox: viz., that, on the use of its own truth-predicate, it is impossible to identify the many different "relationalized" uses of "true" that should then obtain. That, in effect, is the heart of the argument Plato has Socrates advance against Protagoras in Theaetetus, and that is the same argument upon which contemporary opponents of relativism insist (for example, Putnam).
It's of no concern here to redeem Protagoras against the ancient argument, though it can be done. The important thing is that the viable form of relativism (call it robust relativism, for tendentious reasons) defines truth in a conventional way but insists (i) on retiring, in particular sectors of inquiry, bivalent values that accord with excluded middle and tertium non datur, (ii) on replacing such values with a many-valued logic that countenances "incongruent" judgments, and (iii) on reconciling such a logic with a bivalent logic, by not allowing the ranges of their respective application to overlap (by restricting their relevance ).
In short, in accord with the viable form of relativism, "true" simply means "true"! Truth-claims remain subject to "epistemic" and "ontic" constraints. "True" means "true", but, whether relativistically construed or not, the assignment of truth-values remains subject to the vagaries of "L " and "W " and so on. That is, "relationalists" hold (by self-referential paradox ) that "true" means "true-in-Lk "; "robust relativists"
hold that "true" simply means "true" as it standardly does in bivalent logics. Relationalists define "true" (alethically) in terms of the epistemic peculiarities said to obtain in "Lk "; robust relativists oppose any such maneuver. Robust relativists insist that bivalence must be replaced, in given domains, for epistemic and ontic reasons bearing on the peculiarities of "Lk " and "Wk " and so on. They do not dispute the (alethic) sense of "true," only the choice of truth-values and the conceptual rationale for it.
There's all the difference in the world between the two options. The conditions for ascribing truth-values to interpreted claims about the world cannot ever be decided on "purely" logical grounds. There are no possible such grounds except self-contradiction, self-referential paradox, and the like. And these, as we have seen (distinguishing between interpreted "sentences" and uninterpreted "sentential formulas") yield no pertinent objections.
We may now bring the matter to bear on the epistemic complications. Fortunately, the issue returns us to the themes of chapter 3, without yet venturing any of the more radical doctrines I mean still to support. In chapter 3 I showed that reference and predication were (i) ubiquitous in constative discourse, (ii) undeniably viable and effective in natural-language use, (iii) inseparable, in being effective, from the consensual practices of particular societies, and (iv) incapable of being explained in terms of invariant rules or principles or algorithms or the like, or in exclusively linguistic terms, or in any way in which the supporting consensual practices function criterially. I construed the lesson of (i)–(iv) as explicating part of the meaning of "Lebensformen " (taken as a term of art—in effect, (3.17)).
If, now, we reflect on our findings regarding reference and predication in the context of (2.1), the denial that reality must possess invariant structures, and (3.1), the denial that the structure of language may be examined independently of the structure of the world (and vice versa), then it should be clear at once that
(4.8) language and world form an indissoluble symbiosis: the "world" is "languaged," and "language," is "worlded"; effectively, the analysis of the world and the analysis of language are one and the same.
We cannot, of course, fail to distinguish "world" and "language" in actual discourse: reference and predication presuppose the viability of doing so. For similar reasons, it must be that
(4.9) cognizing "subjects" and cognized world (or cognized "objects") are indissolubly symbiotized.
By parity of reasoning, we cannot fail to distinguish between speaking "subjects" and discursible "objects" within the space of our referential and predicative acts; the success of such acts presupposes and entails some operative ("external") relationship between what is being referred to and who, speaking, is speaking referringly. But if we admit that, then we come to a surprisingly powerful finding:
(4.10) the relationship between (speaking) subjects and (discursible) objects, and between (worlded) language and (languaged) world, although "implicated" in viable natural-language practices, is an artifact of, and discernible only within, the indissoluble symbiosis of the "two."
A perspicuous consequence of (4.10), then, is this:
(4.11) the "external" relationship of subject and object and of language and world are artifacts of their "internal" relationship.[ 5]
Theorem (4.11) may now be fairly read as "Hegelian" or at least "post-Kantian," or simply as defining symbiosis. But if you concede (4.11), you see at once the arbitrariness of Dummett's treatment of tertium non datur: it cannot possibly be regarded (by Dummett) as an alethic principle of any sort, it must be some sort of dependent theorem defended (at best) by a principled ontic policy that is as yet nowhere vindicated. For example, if you accept (4.11), which defines symbiosis, it becomes impossible to insist that a bivalent logic is alethically inescapable.
By an external relationship, I now mean one in which, as in the dyadic relationship "aRb, " "a " and "b " may be individuated, identified, and characterized independently of the relationship "R "; by an internal relationship, I mean one in which, say, "a " and "b " are no more than relata incapable of being individuated independently of "R " or of being characterized in any way independently of how they are characterized in "R ." It is, for instance, one of the notorious defects of Francophone structuralism (originated by Saussure) that the "elements" of a structuralist binarism are no more than relata . (This is in fact an important clue to Derrida's merciless "deconstruction" of Rousseau, Saussure, and Lévi-Strauss.) Symbiosis, as in (4.8)–(4.9), is an "internal relationship";
but since it plays (as such) no criterial role, it enjoys an entirely benign standing. It serves holistically: that is, only to call into question any privileged or archic reading of the "external" relationship between the polar "elements" (or surrogates) of (4.10). The conceptual weakness of structuralism lies, precisely, (i) in its treating what is real and intelligible only as relata (signifiant/signifié ), and (ii) in direct epistemic terms. (That move is self-defeating—whether in structuralist or Hegelian terms.) In accord with this reading, symbiosis—that is, theorems (4.8)–(4.11)—precludes the separability of realism and idealism. Realism concerns constative discourse addressed in criterially effective ways to (what are taken to be) the "external" things of the world; idealism addresses the subjective conditions relative to which alone the other obtains. Realism without idealism, we may say, is privileged; idealism without realism is epistemically irresponsible.[6] (I take this to be the import of (3.10).)
Idealism, therefore, may be symbiotized (as among the post-Kantian Idealists), but it also need not be (as it is not, in Kant). In Kant, the cognizing powers of human subjects are not thought to be constituted in the same sense in which the intelligible structure of the cognizable world is. That is what I take to be undefended in Kant, to be, indeed, indefensible: it avoids the initial holism of symbiosis.
Theorem (4.11) is quite radical. One sees this, for instance, if one considers how the denial of "privilege" and the admission of "historicity" bear on its meaning. (Of course, the full import of denying "privilege" and of affirming "historicity" has yet to be supplied, but, in pursuing the an-archic vision, I have already dismissed cognitive privilege as archic, and, in defeating Aristotle's archic paradigm, I have shifted the burden of proof to the champions of the archic vision.) If, therefore, (4.11) is read an-archically, one cannot resist admitting that
(4.12) the analysis of symbiosis cannot yield any privileged account of language and world or of cognizing subject and cognized object.
In short,
(4.13) symbiosis never functions criterially or epistimically.
Any inquiry that admits (4.12) but violates (4.13) is self-defeating. Theorems (4.12)–(4.13) are quite enough to defeat Kant for instance—that is, read as an archic theorist. For Kant holds that we are epistemically confined to the contingencies of the phenomenal world (noumena, after
all, are unknowable for Kant). Nevertheless, postulating that the phenomenal world lacks all cognitive privilege, Kant gymnastically manages to elicit an extraordinary number of synthetic a priori truths regulating the phenomenal world ("experience") from the side of cognizing subjects. He cannot possibly succeed in this, on his own assumptions (for instance, those of the preface to the second edition of the first Critique ). If, alternatively, he opposes (4.12)–(4.13) insofar as he elicits (reflexively) various necessary conditions of perception, understanding, and reason, he must also (in some sense) deny that the noumenal world is unknowable and, what comes to the same thing, that we confront the phenomenal world holistically (without a priori resources). Q.E.D.
It was in just this sense that I suggested, in the previous chapter, that late twentieth-century (continental) philosophy is essentially "post-Kantian" rather than fully "Kantian": it subscribes to symbiosis, but, in doing that, it abandons privilege. In the terms just introduced, Kant restricts ("mere") empirical realism (the range of cognizable "objects") so that it cannot arrive at ideal(ist) invariances, but he does not correspondingly restrict pretensions regarding the subjective or idealist sources of apodictic, transcendental invariances to the same holist contingencies. Kant cheats in this regard. (As a result, Rorty hurries, mistakenly, to the postmodernist conclusion.) Kant should have confessed that he could not possibly know what constraining conditions tacitly affected his conjectures about apparent a priori truths drawn transcendentally from experience.
In reaching (4.11), we have come to an elegant resolution of one of the deepest puzzles of philosophy. For, I am prepared to argue, if we deny cognitive privilege tout court, we thereby subscribe to some form of the principle of internal relations (specified above (4.11)); even though, if we also admit reference and predication, we cannot deny that the relation between "subject" and "object" must be an "external" one that is not (as such) subject to the peculiar logic of "internal relations." The trick is turned by denying any epistemic or criterial function to "internal relations."
The strong analytic criticism of Hegel and Bradley in this regard is largely due to Russell's perceptiveness. Nevertheless, the champions of internal relations (implicitly opposing Russell, who is an externalist and a frank advocate of cognitive privilege) have indeed grasped the important fact that their own insistence precluded cognitive privilege of every kind. It is a curious fact that, through a good part of the analytic tradition, which (rightly) opposes the doctrine of internal relations (con-
strued epistemically, as in Bradley), some form or other of cognitive privilege is often found to lurk. That privilege may, for instance, take the form of prioritizing "semantics" over "metaphysics" (as Dummett does), or it may take the form of openly advocating privileged sense-data (as the English empiricists have done) or "protocol sentences" (as Carnap had done before Neurath's correction),[7] or it may take the form of an a priori presumption that intentionality à la Brentano may be eliminated in any realist account of the world (as Quine has insisted). You begin to see, therefore, the (negative) epistemic import of holism and symbiosis. (I shall look more pointedly at this in a moment. I have already made clear, however, that, in treating Lebensformen as a term of art, I employ it holistically and symbiotically—and will, as I have also signaled, use it in a historicized way.)
These last maneuvers gain credibility only on the assumption that no version of the doctrine of "internal relations" is tenable. What I have tried to do, therefore, is provide a sense in which that doctrine (i) does not lead to the disastrous consequences Russell rightly warned of and (ii) itself exposes the equally disastrous uses to which construing the relationship of subject and object as "externalist" has been put.
Arguments in these two directions are clearly asymmetrical: against exclusively internal relations, the argument identifies a self-defeating (epistemic) paradox; against exclusively external ("externalist") relations, it warns about a drift toward cognitive privilege and modal invariance. The first simply shows that reference and predication are effectively impossible if the "external" (epistemic) relationship between subject and object is not preserved. The second shows equally compellingly that merely preserving that relationship cannot by itself ensure that the relata thereby related remain (or must remain) invariantly structured in any epistemically pertinent or productive sense. It is surely in the work of Kant and Hegel that this two-sided puzzle is effectively laid out and then threatened with incoherence. I may say that I find in Foucault the late heritage of both, formulated in a way that deliberately risks the puzzle's insolubility but, in doing that, also defines its most radical possibilities in historicist terms.[8] I take this to be characteristic of poststructuralism. Furthermore, I take theorem (4.11) to suggest a basis for reconciling the whole of analytic and continental philosophy: on the analytic side, it subverts all externalism; on the continental side, it supports the subversive work of all poststructuralism.
Beyond that, I say only that, by treating symbiosis holistically, not criterially, not in any way that plays an epistemic role, I believe I have
supplied the essential clue for solving the puzzle of internal relations without privileging external relations. In that sense, every viable philosophy of the new century will have to come to terms with the "poststructuralist" challenge, which signifies the symbiosis of realism and idealism—most radically in explicitly historicist terms. Reference to "external" relations concerns epistemic matters on their (epistemically) effective side; reference to "internal" relations concerns the holist and symbiotized conditions (not themselves criterial or epistemic) under which the other functions. There is no paradox there.
Certain late tendencies in our century, particularly in continental Europe, have, I should say, strongly favored a reading of (4.11) that is not open to the charge of privilege. In this sense, (4.11) yields:
(4.14) the world (or, a world) is a text (is "texted");[9]
and
(4.15) constative discourse "interprets" the world by "parsing" its holist space; in doing that it constitutes "the world" a text, posits distinct cognizing subjects and cognizable objects.
The conjoint import of (4.13) and (4.14) signifies that: (i) all distinctions regarding particular things obtain only within the holist space of "the world" (or, the "universe of discourse"), regarding which there cannot (yet) be any privileged posit of "what there is" (distributively); (ii) holism itself does not preclude discursive practices regarding things (thus) distributed in the "world"; (iii) every discursive practice presupposes and entails an interpretive or categorical scheme (or schemes) in terms of which the "whole" world is thought to be distributively "ordered" (is parsed, is made a "text") in the ways that discourse affirms or could affirm; and (iv) no interpretive scheme is privileged as such.
Taken together, (4.14) and (4.15) formally resolve the paradox of internal and external relations—but they do so only formally. They lead, however, to:
(4.16) referents, predicables, and truth-values are artifacts internal to the "interpreted text" that is the intelligible world.
By a text, I mean a symbiotized but individuated referent apt for interpretations. (I realize how easily the term may be abused. I shall take due
care.) Here, it may be useful to invoke something akin to Spinoza's distinction between natura naturans and natura naturata . I say (heuristically) that, within the symbiotized world, conceptual schemes "interpret" (naturans ) that undifferentiated "holist" ensemble; in doing that they constitute that ensemble, a "text" apt for further "interpretation" (naturata ). Interpretation naturans "parses" the first; interpretation naturata discursively addresses what is thereby parsed within the symbiosis of the first. In this sense, we may take a leaf from Barthes and affirm that:
(4.17) texts are the contexts of other texts.
That is, contexts are simply "texted" worlds (or parts of worlds) in which individuated texts are themselves identified. Hence, "worlds" are, alternatively, texts or the contexts of texts. (You must bear in mind the way in which (4.17) implicates (4.9).) I risk saying therefore—although I fear I shall be misunderstood—that Heidegger's Being and Time and Quine's Word and Object converge, from altogether different directions, on theorems (4.14)–(4.17).
The grandest "internalist" version of (4.16) is probably Hegel's invention (the self-discovery of evolving Geist ), and the leanest "internalist" version may be very close to what Quine allows to fall under the constraints of his "analytical hypotheses." Both doctrines are dubious but for different reasons. Similarly, the grandest "externalist" version of (4.16) appears in Wittgenstein's Tractatus, which is notable because it invokes the correspondence theory of truth in an unqualified way but without reference to the epistemic conditions of first-order truths; the leanest "externalist" version (if that is what it is) may, eccentrically, be claimed by Rorty, in the sense that, for Rorty, constative discourse simply gets on with its affirmations and denials, without ever troubling about the nature of truth or the legitimation of truth-claims.[10] (This is the import of Rorty's rejecting interpretive "tertia ".) Both doctrines are paradoxical but for different reasons.
It may be helpful to distinguish carefully between truth and truth-values . Truth-values, I should say, are (i) various sets of special predicates (bivalent, many-valued, possibly differing in other ways as well) that (ii), on a theory of truth, are assigned their proper denotata (sentences, propositions, statements, truth-claims, sentences used in constative acts, or the like), such that (iii) those bearers are ranked or graded in accord with the regulative function affirmed by the theory of truth invoked. Truth, on the other hand, is (i) a would-be relation between
the bearers of truth-values and appropriately segregated parts of the world (facts, states of affairs, or the like) which (ii) those bearers (sentences, say) claim to represent in some normative way (as by representation or correspondence or mapping or by favored forms of coherence or verisimilitude or the like), and which (iii) may or may not be independently confirmable. "Truth-values," therefore, are specified alethically but have their use only in epistemic and ontic contexts, whereas "truth," construed as genuinely regulative, designates an appraisive condition at once epistemically and ontically pertinent that normatively governs the ascription of truth-values. This accords pretty well with Aquinas's classic formula, and even with William James's, although it is easy to see that the formulation I have just given may actually be completely vacuous.[11]
The important thing is this: assuming symbiosis and rejecting cognitive privilege,
(4.18) truth can have no independent ("externalist") regulative function vis-à-vis the assignment of truth-values.
We can, of course, still "regulate" truth-claims ("internally"). Once we characterize truth-values and truth, theorem (4.18) is the clear corollary of (4.10)–(4.11). I should say truth was intransparent or transparent (derivatively: those parts of the world truth-claims are "about" are "intransparent" or "transparent"), depending on our complicitously acceding (or not) to (4.17). Hence, recalling the issues of chapter 1, if we act in accord with (4.18), then
(4.19) truth and legitimation are intransparent.
This is precisely what those drawn to the disquotational theory of truth had feared.[12] Not being able to ensure an extensional treatment of natural language by way of an analysis of truth directly, they construed truth as otiose (disquotationally, as in Quine) or replaced its epistemic function by causal means (by "naturalizing" truth, as I shall say, as Davidson does).
These distinctions help to confirm a number of additional findings. For one, reading theorems (4.18)–(4.19) in accord with (4.10)–(4.12),
(4.20) truth and legitimation are second-order artifacts of our symbiotized world;
that is, they are part of the "text" that is "the world." The postmodernists (Rorty, Lyotard) assumed that either truth and legitimation were completely "transparent" or else philosophy should be abandoned, but their judgment is an extravagance, and also a non sequitur . Under "intransparency," the reasoned legitimation of what to take as truth, knowledge, reality, confirmation, and the like increases in importance as the hazards of life and inquiry increase. In this sense, our theories of truth continue to function "regulatively," although they can no longer do so in an archic way. Furthermore, to speak constatively is to have constituted "the world" a text: truth and legitimation are as internal to that constituted world as any other objective "part." (Foucauldian "genealogy" would make this clear.)
This explains the sense in which, for instance, Putnam's insistence on the regulative function of truth (on the side of "internal realism") as against "idealism" (championed by Kuhn and Feyerabend) or against "metaphysical realism" (as in Putnam's quarrel with Dummett) is a complete nonstarter. Once privilege is abandoned and symbiosis accepted, there can be no principled difference among these doctrines as far as the "regulative" function of truth is concerned (naturata ). Furthermore, once symbiosis is admitted and/or privilege denied, then, in the spirit of (4.14)–(4.15),
(4.21) truth and legitimation are interpretive.
In the same sense (although it will have to be spelled out a little more carefully):
(4.22) the disquotational theory of truth is false, being incompatible with (4.21).
Theorem (4.22) is a powerful finding, which leads to the issues of the next chapter. (I shall come to its full meaning before closing the present chapter.) For one thing, Quine may be seen to have trapped himself in something close to a contradiction: he subscribes to the "disquotational theory" of truth, but he also holds that truth-claims (for the most part) fall under our "analytical hypotheses" (our local interpretive schemata, in accord with which we parse our holist world and, relative to which, "indeterminacy of translation" cannot be overcome). This is to argue at cross-purposes—not inconsistently, it's true, but Quine would have us believe that he eschews all forms of cognitive privilege. For another
thing, Davidson opposes the disquotational reading of Tarski's theory (which he claims to adopt and "supplement"), although he also opposes (as does Rorty, for entirely different reasons) interpretive tertia . Quine salvages "transparency" (against his own holism) by disjoining sentence and term and reserving some regulative function of truth along transparent lines ("stimulus-meaning"). Davidson favors transparency for the massive core of our ordinary beliefs (in a somewhat Wittgensteinian manner), but he does not specify a source of epistemic privilege for distributed beliefs. (Illicitly, Davidson thus exceeds Wittgenstein's holism: he commits the fallacy of division.)
I judge that Davidson rejects disquotation because he fears (rightly) that a theory of truth cannot fail to be interpretive—as, for instance, in supporting correspondence. Rorty rejects the regulative function of truth altogether and, hence, theories of truth, legitimation, and interpretive tertia regarding the structure of the world and philosophy itself. But he cannot mean what he says, in the plain sense that he respects too much the success of the sciences.[13] The work of the sciences obviously involves reasoned changes on legitimative questions.
I must clear up some terminological matters here lest the argument not be understood. By interpretive tertia (the term is introduced by Rorty and is shared, at least implicitly, by Davidson), I mean (i) alternative categorical schemes by which the holism (the intransparency) of our symbiotized world is parsed in such a way as to permit constative acts or truth-claims, or (ii) schemes that, one way or another, mediate, as such, what are linked within the truth-relation itself ((4.15)). Paradoxically, in a period dominated by admissions of intransparency, merely to offer a theory of truth that rejects tertia will appear to invoke some other tertium —as with disquotation and correspondence. (This may explain Davidson's most recent pronouncements on truth and language.)
By holist, I mean the philosophical effect of adverting to the inclusive "universe of discourse," but without reference (or grounds for making reference) to any particular truth-claims regarding what may be found in "the world" (thus, one speaks in a "holist" way about "the world"). To parse (the term is Quine's) is simply to provide a determinate conceptual scheme (one of Quine's "analytical hypotheses") among various options, facilitating particular truth-claims;[14] such claims are distributed. The intelligible world thus constituted (naturans ) I call (by a term of art) a "text" (or "texted world"). My quarrel with Quine lies in this: against what he explicitly says, admitting sentences is as much an exercise of "parsing" as admitting terms —the one cannot obtain without
the other. There are no sentences without terms, or terms without sentences. (Quine denies this.) This explains why Quine's holism is not entirely consistent, why he is committed to some form of privilege (as in his account of "holophrastic" sentences). Furthermore, once the "distributed"/"holist" contrast is adopted, any parsing (privileged or not) will be seen to be "interpretive"—hence Davidson's worry. All this follows directly from a strict adherence to the rigors of symbiosis.
Finally, by disquotational theory of truth, I mean that theory that accepts as its paradigm (disputatiously attributed to Tarski) the formula: 'Snow is white' is true if and only if snow is white. To ascribe truth, it is said (notably by Quine, who accepts the formulation), one merely "cancels the quotation marks." My own view is that the disquotational theory is either utterly vacuous (in the form just given), privileged (if taken as correct as far as it goes, as on Davidson's reading of Tarski), or else flatly false (since, admitting intransparency, some interpretive tertium will be needed). The objective of these strenuous (but ineffectual) moves against the legitimation of truth (a fortiori, the legitimation of truth-value assignments)—on which Quine and Davidson agree—is simply to endorse what has come to be called "naturalizing." I believe that program fails utterly, and I turn to it in the following chapter.
I draw one final consequence from all this:
(4.23) attributions of truth and falsity cannot be epistemically regulated in any way that escapes the constraints of symbiosis and holism.
Ascriptions of truth and knowledge, therefore, cannot fail to be (artifactual) posits made and sustained within our lebensformlich practices. (That alone defeats the disquotational thesis—and, more significantly, "naturalism.") There is no other way to avoid privilege. But to say so is hardly to disallow whatever realism our sciences are said to support.
If, now, you take a moment to reflect on the running argument, you will see that the inherent informality of reference and predication quite naturally invites a certain conceptual tolerance regarding alethic choices and that the admission of symbiosis confirms the reasonableness of considering the selective use of a many-valued logic. Conjectures along these lines permit us to question the strong sense of systematic closure often favored in analytic philosophies (for instance, in applying Tarski's conception of truth to natural languages or precluding Brentano's notion of intentionality in parsing reality). The power of these maneuvers is greatly strengthened by the fact that they are originally favored for rea-
sons internal to canonical treatments of reference and predication and subject and object. That is, that power does not depend in any way on first invoking historicity or intentionality. On the contrary, the argument leads us to see the reasonableness of admitting them in the first place.
In a word, the disquotational theory of truth (or any of a family of related maneuvers) is rightly confined to alethic constraints regarding the use of truth-values but illicitly draws epistemic and/or ontic consequences regarding truth. That, I suggest, is one of the generic strategies favored by so-called "naturalisms."
Chapter V
Epistemic Competence
If what has been claimed in the preceding chapter is conceded, particularly (4.10) and (4.11), which affirm that subject and object and language and world are symbiotized artifacts and that the "external" relationship between these artifacts is specified only within a more inclusive "internal" space, then, as remarked,
(5.1) ascriptions of knowledge are assigned within the same "texted world" in which subject and object (and language and world) are externally related.
At one stroke, therefore, all claims of privilege are subverted, without disallowing the cognitive assessment of particular beliefs and judgments. As we approach the end of the century, one sees fewer and fewer explicit claims of cognitive privilege. It's no accident, for instance, that Aristotle's insistence on reality as structured in an invariant way is matched (in the same theory) by the epistemic ("externalist") claim that the rational soul in man, the highest competence of mind (nous ), natively possesses the power (noesis ) to discern without distortion (to intuit) the changeless universal forms of independent things. You will hardly find a comparable claim in any late twentieth-century philosophy not already explicitly committed to Aristotle (or, with significantly reduced conceptual resources, committed to Kant or Husserl or some such figure). You must look for subtler marks of privilege.
Sense-data (or related) theories of the sort that once dominated Brit-
ish empiricist philosophies in the early part of the twentieth century have almost completely died out—those that in a variety of forms attracted Russell and Moore and Price and Ayer and, in the United States, Lewis and Chisholm.[1] These are the paradigms (although hardly the only specimens) of what has come to be called foundationalism: the theory of knowledge that segregates beliefs (perceptions, sensory experiences, and the like) into an ordered two-story (externalist) world of indubitable or relatively certain cognitive episodes of one distinct (privileged) sort and all those other beliefs the reliability of which ultimately depends on their appropriate linkage to the putatively "foundational" ones. Wittgenstein showed compellingly (particularly against Moore) that expressions like "I know . . . " are not (i) reports of cognitively privileged (psychological) states, nor (ii), in conveying a sense of "subjective" certainty (as Wittgenstein puts it), expressions that link certainty with the sure confirmation of truth or actual knowledge.[2] Wittgenstein, therefore, exposes (in On Certainty ) Moore's penchant for joining, in a systematic way, a naive realism, a commonsense form of the correspondence theory, and a strong version of the view that subject and object form only an externalist relationship. (Moore, therefore, is a "pre-Kantian" well after the fact.)
That Moore and Russell (and Carnap and Schlick and Ayer) were attracted to foundationalism (as well as to something like the doctrine just sketched) helps to locate the radical turn in Wittgenstein's own thought that pits him against himself as the author of the Tractatus (although he does not offer, in the earlier manuscript, any explicitly epistemic version of the foundationalist theory).
It is a fact, nevertheless, that the Positivists, who believed that their own adherence to the epistemic privilege they assigned so-called "protocol sentences" (mistakenly, as Neurath, among their own number showed), also believed that that doctrine conformed—as closely as possible—with the severe constraints on atomic sentences the Tractatus imposed. That required a gymnastic leap. They studied the Tractatus as a working text regarding the foundations of science: with an eye to the mutual fit between the formal syntax of its linguistic model and correspondence theory and their own quite different conception of empirical science and the foundational sources of scientific knowledge.[3] Wittgenstein's oblique assumption of cognitive privilege in Tractatus depends, however, entirely on the so-called picture theory relating true propositions and facts formulated in an unyielding extensional form. (There is very little of empirical science in Tractatus; there was meant to be none.
Ramsey ultimately convinced Wittgenstein that the empirical science that remained was disastrous to his project.)
The curious thing is that, in Tractatus, Wittgenstein does not directly address epistemic questions, even though his (archic) theory certainly entails some inexplicit form of cognitive privilege. Once we grasp the protean possibilities of privilege, we begin to see how epistemic claims may masquerade as merely alethic or ontic or in an even more unlikely disguise. This explains, for example, why earlier I construed Dummett's prioritizing of semantics over metaphysics as an archic thesis incorporating a camouflaged form of privilege ((2.5)). A similar charge may be laid against Frege's sanguine views regarding bivalence and an extensionally ideal language, but Frege's program was more straightforward than Dummett's.
Certainly, (5.1) goes some distance toward exposing important versions of privilege that have exercised late twentieth-century critics—views that are "foundationalist" in spirit although not cast in explicitly epistemic terms. Two such obscure doctrines address what, in the jargon of our time (in Europe at least), have come to be called the "originary" and the "totalizing."[4] (These, you may guess, are hardly perceived by "analytic" philosophers to be kin to the views of figures like Moore and Carnap. But they are, and the failure to discern that fact signifies a misperception.)
Thus, consider that—in a sense remotely related to what was featured in the Tractatus but not committed to any explicitly extensional program and not committed to an exclusively linguistic form of "picturing"—it is supposed (among certain continental European philosophies) that there must be some "original" epistemic (entirely "externalist") relationship between the language and thought of cognizing subjects and the independent structure of cognized objects, suited to ensuring a secure ("foundational," or, better, "foundationalist") basis for every form of knowledge. On the argument, the originary is that mythical point at which language and thought (possibly even prelinguistic thought) is supposed to have "first" confronted the world in the correspondence sense. The notion certainly appears in Lévi-Strauss's journal entries. The tireless (possibly tiresome) exposé of that search for the "originary" constitutes a good part of the motivation for Derrida's early, distinctly "analytic" attack on Saussure's structuralism and Husserl's phenomenology (and what Derrida takes to be the effective convergence between the two.) It's the straightforward key, in fact, to Derrida's deliberately florid program of déconstruction and interminable pursuit of "différance, "
"radical alterity," "supplémentation, " and similar exotica. (The same target, however, may be found in Descartes, Kant, and Husserl.)
Very simply put, Derrida's thesis holds that Saussure's binarism (of "signifiant " and "signifié ") and Husserl's phenomenological analysis of the constituting function of "subjectivity" in the formation of our concepts ineluctably presuppose some form of the "originary." (Derrida is probably not entirely accurate in his account of Saussure.) Furthermore, Derrida argues, if that be admitted, then, under the condition that there cannot be an "originary" (origin) at which thinking first comes to picture the independent world, it must be similarly impossible that the categories the mind supplies for the systematic analysis of reality could ever include "all possible" categories or could discern any fixity in any network of the categories with which we understand our world. Our categories cannot count on being adequate at any time for all the contingencies of evolving experience. One cannot presume to arrive at any (genuinely) totalized schema—one that in principle would never require supplémentation, wherein new distinctions (and the inevitable reinterpretation of the old ones) would dawn (unpredictably) from time to time. One cannot then escape the radical contingency of the contexts of conceptual understanding. One cannot discern the "totalized" context of all conceptual schemes. (Deconstruction is the insinuated and linguistically parasitic—but not explicitly or independently argued or analytic—exposé of "totalizing" and the assumption of the "originary.")
The matter is complicated, Derrida supposes, by the fact that conceptual schemes are said (by the structuralists) to be inherently "relational" (in the sense sketched in the previous chapter: except that, for the structuralists, that relationship is, now, fatally, epistemically prior to other forms of cognition). The complication affects Saussure's (and Lévi-Strauss's) structuralism, Derrida affirms, since signifiant and signifié are explicitly defined as the (epistemically pertinent) relata of all signes within a "closed system" of signs; something similar is said to be true as well of phenomenology, since the would-be discovery of the ("eidetic") invariances of our formative ("subjective") concepts are inseparable from the deep contingencies (historical and preformative) that mark the unfathomed originating milieu within which we presume to fix the conceptual invariances we claim to discern. This is the same theme, drawn from Derrida's objections to Husserl's philosophical program, that appears (in other forms) in the critiques of Husserl offered by Heidegger, Merleau-Ponty, and Sartre. But you must appreciate that what I have said about the holist nature of the phenomenal world (in
the preceding chapter) affects—by way of "internal relations"—the fortunes of "totalizing" as well. In analytic philosophy, this appears preeminently in Kuhn's acknowledging the historicity of science.
If privilege may take the attenuated forms mentioned, then it becomes instructive to insist that
(5.2) all conceptual schemes are interpretive tertia, artifacts of prior interpretation,
in the strong sense (i) that conceptual schemes cannot be originary or totalized, and (ii) that explicit conceptual schemes reflect some deeper tacit precondition that we cannot completely fathom. (By artifacts, I mean what, by a term of art, I have in the preceding chapter called texts, the products or posits of productive or interpretive human powers applied within a holist space. I extend the term, of course, to whatever is symbiotized in realist/idealist terms.)
This makes for a very pretty irony.[5] Husserl, we may now say, despite having exposed what he took to be Kant's philosophical naiveté, neglected (in his turn) to consider the (symbiotized) preconditions of his own epistemic certitude; he thereby failed to grasp the deep contingency of his phenomenological labors and the impossibility of treating them as more "foundational" (again, "subjective" or "idealist") than Kant's (or any other similarly) "naturalistic" reflection. Derrida actually has all this in mind when he adds to the presumptions mentioned a third form of privilege, the "apodictic," which, in a sense, is inseparable from the other two. The apodictic signifies (particularly in Husserl) the cognitive certainty of grasping (or the asymptotic certainty of approaching closer to grasping "eidetically") the originary and/or the totalized, in terms of which some favored conceptual scheme may (in the appropriate limit) be legitimated as necessarily invariant. It would not be entirely amiss to see in all this an alien analogue of the very thin foundationalism the British sense-data theorists and the Positivists espoused. There is, therefore, a surprising convergence (which critics have not altogether missed) between the Wittgenstein of On Certainty and the Derrida of Of Grammatology .
If we collect these sprawling patterns effectively, we cannot fail to see that (5.1) and (5.2) lead to the strong finding:
(5.3) all discourse is, epistemically, context-bound, contingently preformed, unfathomed to some extent, endogenously impenetrable, in-
transparent, requiring interpretation, unable to support claims regarding the originary or the totalized or the apodictic.
Theorem (5.3) is a firm part of the "therapeutic" philosophies practiced by (the "later") Wittgenstein and Derrida. The Anglo-American and continental European traditions converge here, however bizarrely: both Wittgenstein and Derrida mount a sustained attack on those philosophies that presume to eliminate the complex epistemic relevance of context . There's the thread that runs from the correspondence theory of Wittenstein's Tractatus (which Wittgenstein himself repudiates) and the unrelenting extensionalism of Quine's Word and Object (which Quine cannot quite make coherent) to Saussure and Husserl, the targets of Derridean deconstruction.
I believe an entire family of theorems may be drawn from this comparison, which, I remind you, has already been adumbrated in (2.5)—which, against Dummett's policy of prioritizing semantics over metaphysics, opposes any hierarchized privilege. If so, then the following makes a reasonable beginning:
(5.4) in neither epistemic nor ontic terms can realism and idealism be disjoined in a principled way.
I might have derived (5.4) directly from (4.10) and (4.11), which I recalled at the beginning of this chapter, but we have it now in the more articulated space in which epistemic distinctions are indissolubly linked to alethic and ontic ones. Theorem (5.4) signifies that most of the going philosophies of our own age are probably (or may be said to be) "naturalistic" (to use a term intended in a derogatory way by Husserl, as signifying [for him] a presumption of one-sided [and unguarded] privilege, and in a rather complimentary spirit by Quine [who speaks more usually of "naturalizing" than of "naturalistic"], as signifying [for him] the demystification of the merely arbitrary forms of second-order legitimation). But such philosophies may also be said to be "phenomenological" (to use a term intended in a privileged way by Husserl, as signifying [for him] the correct approach to an apodictic assurance regarding our conceptual resources, and [alternatively] intended in an opposed way by Hegel, as signifying [for him] the impossibility of privilege itself, the endless unfolding of novel historical experience within the movement of which we try to grasp the coherence of our changing conceptual resources).
By naturalism (also, by "naturalizing " philosophies), I (now) mean (as a term of art) no more than what, epistemically, corresponds to the ontic reading of the "external" or "externalist" relationship between cognizing subject and cognized object, in terms of which some form of "realism" and objectivity may be defended. By phenomenology, I (now) mean (as a term of art) no more than what, epistemically, corresponds to the ontic reading of the "internal" (holist) relationship between cognizing subject and cognized object, in terms of which some account of the contribution of "subjectivity" to the realism of our intelligible world may be defended. Furthermore, by realism, I mean the ontic complement of "naturalism," whether or not privileged (whether merely "external" or fully "externalist"), and by idealism, I mean the ontic theory of the "constructive" relation between subject and object, whether or not strengthened by positing privileged reflexive powers (as, differently, by Kant and Husserl) addressed to the "subjective" conditions of that same relation. On my argument, this accounts for what is illicit in Kant's and Husserl's idealisms. (It needs also to be said that the history of such terms as "realism" and "idealism" precludes any useful point in insisting on assigning them a univocal sense, even as terms of art.)
These complex notions are now definable in a strikingly symmetrical and simple way. We may go further; we may say:
(5.5) the analysis of "external" and "internal" relations in any domain presupposes the prior use of determinate distinctions of the same kind;
and
(5.6) naturalistic and phenomenological analyses are indissolubly symbiotized parts of a single inquiry.
If we take these terms in a sufficiently generous way—in accord with (5.5), that is, not in a privileged way—then "naturalism" and "phenomenology" will be seen to signify the symbiotized halves of a single coherent epistemic theory (not admitted to be such by Kant or Husserl) that the an-archist requires.[6] They are (in this respect) the epistemic counterparts of the (similarly unprivileged) ontic theories that, as "realism" and "idealism," contribute to a viable an-archic account of the intelligible world. (I shall say that an-archic, but not archic, theories favor both a "constructivist" and a symbiotized account of world and
knowledge. Neither Kant nor Husserl—nor sense-datum theorists, to include a latent comparison at which I have been hinting—admit symbiosis. But it will take some time to set the stage for admitting "constructionism.")
These definitions of "realism" and "idealism" and of "naturalism" and "phenomenology" are too tight, too spare, to be of much use without considerable elaboration. They cannot, for instance, be of much use in working out the difference between "Husserlian" and "Hegelian" phenomenology. But they are not intended as a mere classificatory convenience: they help to isolate the inseparable dialectical connection between "internal" and "external" relations and between "epistemic" and "ontic" inquiries. They serve to clarify the insuperable conceptual contingencies that result from adhering—scrupulously—to the "holist" limitations of symbiosis and intransparency, and they show how the resultant theories remain coherent.
There is at least one further dimension implicit in this conceptual knot that needs to be acknowledged—if ever we are to bring the entire sweep of Western philosophy into one vision: that of the historicity of thinking . I take the thesis to be the principal postulate of this primer, as I say. Nevertheless, in an extraordinary way, it now appears that the running argument will lead us to it as to an implicated theorem. In fact, it is hardly farfetched to insist that, in a post-Hegelian world, it is quite unlikely (in fact it is strange) that any serious critic of Kantianism—Husserl, particularly—should have questioned the sources of Kant's apriorism without invoking the formative contingencies of cultural history. But it is a plain fact that Husserl never really concedes the point, rather, he sometimes thinks he has but overrides its "transient" complications.
More pertinently, the argument signifies that I have now confirmed—by way of (5.5)–(5.6)—the unmistakable mutual dependence (and halting convergence) between the entire Anglo-American and continental European philosophical traditions. I have done this by drawing directly on the distinctive resources of each. This yields a fair sense of the self-impoverishing tendencies of late twentieth-century philosophy. It is hardly a caricature to say that Anglo-American philosophy is largely wedded to the analysis of the "external" or "externalist" relationship between subject and object, and continental European philosophy to its "internal" relationship or the privileged internalist analogue of externalism. This helps to grasp how disconcerting Kuhn's immensely important (but weakly limned and ultimately abortive) intuition must have
been when it first dawned on the local quarrels of the Positivists and Popperians, and how peculiarly unresponsive and unguarded Habermas's recovery of Kantianism must now appear when viewed in the light of its own Marxist, phenomenological, hermeneutic, and Frankfurt-Critical origins.[7] I take both episodes to confirm the sense of a loss of balanced conceptual resources resulting from a disjunctive division of conceptual labor. Kuhn exposes (in effect) the illicit "externalist" presumptions in the realism of the Positivists and Popperians; Habermas attempts to legitimate the "external" empirical inquiries of the Kantians (which risk invoking privilege, as we have seen) by his own illicit "internalist" claims regarding the would-be invariant norms of reason.
In the process, I have collected a number of specimen strategies of cognitive privilege: (i) all forms of foundationalism (sense-data theories; commonsense minima, in Moore's sense; Aristotle's noetic intuition of the essential forms of things); (ii) a priori alethic commitments on truth-values (bivalence, in Aristotle's and Frege's sense; tertium non datur, in Dummett's); (iii) correspondence as an a priori constraint on epistemic and ontic theories (as in the Tractatus ); the a priori presumption in favor of extensionalism (as in Quine's and Davidson's philosophies); (iv) any principled priority assigned to alethic, epistemic, or ontic claims separated from conditions affecting the others (as in Dummett's prioritizing of semantics over metaphysics or Chisholm's insistence on self-evident experience); (v) all versions of originary, totalizing, and apodictic claims (in the sense exposed by Derrida); (vi) all denials of the symbiosis of realism and idealism or of naturalism and phenomenology (as, respectively, in Davidson and Husserl) or of language and world or of thought and language; (vii) all principled disjunctions between analytic and synthetic truths or their sources (as in Kant and Carnap); the alleged discovery of unconditional de re and de dicto necessities or invariances (as in Aristotle); and (viii) all forms of second-order legitimation that deny or fail to concede that they are themselves constrained by the first-order claims in which they are embedded (as in Kant and Husserl). These make for a tidy windfall.
There is perhaps one additional entry to this last set of theorems that it will prove useful not to ignore, namely:
(5.7) there are no unconditionally necessary, universal, transhistorical, exceptionless interpreted structures or uniformities discernible in nature or thought.
All this may be taken to be the import of the symbiosis mentioned in (4.10) and (4.11)—with which I began this chapter—of the holism they subtend, and of course (2.1). (But that is not to disallow "indicative" or transhistorical "universals," as I acknowledge in chapter 2. Theorem (5.7) is a "bet," not a necessary truth itself.) The argument comes to this, finally:
(5.8) symbiosis, denied any and all forms of cognitive privilege, subtends some inchoate holism relative to which (only and tacitly) our (distributed) first- and second-order claims (criterially) obtain.
By holism, I now mean no more than the posit of an undifferentiated conceptual space in which (and within which alone) all distributed (realist) truth-claims may, in principle and under conditions not thereby specified, be processed.[8]
I hope you will not regard it as an extravagance if I suggest that, as I now see matters, theorem (5.8) affords a key to a useful rapprochement between Western and certain Far-Eastern philosophies. I take (5.8) to draw the common thread of Anaxagoras's Apeiron (or, better, an improvement on what Anaxagoras appears to have intended), the Tao of classical Chinese thought, and the notion of sunyatta in Buddhist philosophy.
I can put this in a better way: I stand by (5.8) in the context of the whole of Western philosophy, even its most bizarre expressions (as in Nietzsche and Derrida); I also believe that (5.8) is not altogether irreconcilable with the central themes of Taoism and Buddhism, but I am not competent to pursue the matter with any feeling of textual accuracy. I also think there is nothing determinate to say about what "holism" collects: it is no more than the "unity" of "the universe of discourse" effected, as many have remarked, as a sheer artifact of what is the philosophical analogue of negative theology. That "unity" is perhaps what Wittgenstein marked as "mysticism" at the end of the Tractatus . (I confess I see that lesson as akin to what would result from applying (4.18)—that is, the relativity of speaking of "text" and "context"—to the "universe.")
Once matters are put this way, it becomes quite clear that a productive philosophy will be concerned with the conditions under which determinate truth-claims are drawn out of an initial holism. Certainly, if we concede holism and symbiosis, and if we admit that science and related inquiries are concerned with determinate truth-claims, we cannot
reasonably deny that we must decide just what to take (i) as the mark of truth and falsity and (ii) as the mark of reasonable policies for maximizing the discovery of significant truths. But that—(i) and (ii) together—is what is meant by legitimation. Hence, at the very least,
(5.9) second-order legitimation precludes transcendentalism,
as it does the disquotational theory of truth; and
(5.10) truth and the legitimation of truth are symbiotized.
By transcendentalism, I mean (adding to what I have offered earlier) no more than (second-order) legitimative discourse that depends on and entails some form of cognitive privilege, some violation of (5.8)—Kant's own synthetic a priori truths for instance (on the interpretation I have offered)—and some violation of (5.1). Taken together, (5.7)–(5.10) defeat philosophical postmodernism . Thus, Lyotard, acknowledging (in effect) the force of (5.7) but not wishing to support transcendentalism, rejects the seemingly obvious fact that science is occupied with truth-claims ((5.8))—in effect, Lyotard rejects the reasonableness of (5.9). Rorty rejects transcendentalism, but then, admitting the work of science and the force of (5.7), he unaccountably rejects legitimation as well—he rejects (5.10). Postmodernism, then, signifes (i) the rejection of either (5.9) or (5.10), (ii) the rejection of all forms of cognitive privilege, and (iii) as a consequence, the rejection of the viability and philosophical fruitfulness of legitimation itself. It is for these reasons I regard postmodernism as incoherent. A variation on the theme of Kant's famous dictum (mentioned in chapter 1) seems very much in order: first-order truths without second-order legitimation are blind; second-order legitimation without first-order truths is empty. (Something similar holds for naturalism and phenomenology—and for the same reasons.)
Two further theorems recommend themselves. For one:
(5.11) preference for one legitimative policy over another is a function of the first-order truths we are prepared to accept;
and
(5.12) legitimation is itself an "empirical" or "indicative" undertaking concerned only with rational "adequation."
There you have the straightforward reversal of the Kantian ideal of transcendental reasoning as well as the defeat of postmodernism. The reason is the same: reason is itself a normative artifact of legitimative inquiry and, as such, affects our sense of the competence of human "subjects." We cannot reasonably abandon the (second-order) question, "How is science possible [conceptually]?"; but we also cannot pursue it in any archic way. The argument rests on two entirely straightforward considerations: (i) that first- and second-order (constative) acts are parts of a single indivisible discourse; and (ii) that the external and internal relations between subject and object are artifactual distinctions within one inseparably unified epistemic and ontic theory. (Notice: you may say that Kant is not an apriorist, in the sense that he offers his conjectures about the a priori conditions of knowledge by way of "phenomenal" reflection. But if that is so, then, of course, his determinate findings are tacitly, and impenetrably, constrained—and logically weakened—by the preconditions of such conjectures. For example, his conjectures about Euclidean geometry and causal determinism prove to be false. Wherever Kant's claims are determinate enough to lead to such would-be necessities, his conjectures fail to vindicate themselves; otherwise, they behave in ways similar to that in which, as we have seen, Aristotle's claim about the principle of noncontradiction behaves. This bears, for instance, on the conceptual status of the pure intuitions of time and space and the categories of the understanding.)
I regard (i)–(ii) as minimal conditions of rational discourse, dialectically supported by the entire foregoing argument. In claiming that legitimation is empirical, I mean only that it is ineluctably constrained by, and relativized to, first-order ("indicative") discourse. In claiming that it is adequational, I mean only that it is concerned to theorize about the conceptual relationship (or fit) between constative discourse and the cognized world (within the terms of symbiosis): it is second-order, there, in the plain sense that it considers the ("rational") regulation of first-order ascriptions of truth-values by way of policies about what best to construe as truth and how best to maximize the corpus of truth. By rational, then, in the way of constative inquiry, I mean that normative feature of inquiry in virtue of which it adheres to the empirical and adequational aspects of legitimation. (I shall return to the "adequational" in chapter 10.)
On the argument, truth itself (but not legitimation) has no explanatory function. There you have the defeat of so-called "naturalizing" theories of knowledge—Quine's, preeminently—that seek to construe epis-
temology as a subdiscipline within (empirical) psychology. For, although truth is not explanatory, its ascription must be explained and defended. That is the upshot of (5.2). To admit that much is to defeat disquotation and naturalism at a stroke. We should perhaps put this in the form of an explicit theorem:
(5.13) naturalized epistemologies are untenable, because truth is not disquotational and because legitimation, although explanatory, is not explanatory in the causal sense.
I regard (5.13) as one of the most strategic findings of this primer, simply because nearly all current analytic philosophies are drawn to "naturalizing" epistemologies—or, similarly, to naturalizing morality, intentionality, methodology, rationality, or philosophy itself. (I shall come to the moral analogue in chapter 11.)
The seminal expression of all these programs is surely the one offered by Quine in "Epistemology Naturalized." But, as already remarked, Quine favors the disquotational theory of truth; he either means to retire legitimative questions altogether or to naturalize them. In effect, Quine construes (or reduces) second-order inquiries as (or to) first-order inquiries. For his part, Davidson assumes (without argument) the reliability of some modally necessary connection between first- and second-order inquiry: what he calls "supervenience." (I return to "supervenience" in chapter 9. I mention Davidson, only in passing, only to flag the variety of "naturalisms." Nothing hangs on supervenience as yet: it simply marks another way of denying relevance to legitimative questions.) But disquotation won't do, as I have already remarked, once "intransparency" is admitted; naturalizing legitimation is unworkable if its explanatory function concerns what is "rational" (in the sense specified) rather than what is merely "causally ordered." For, what is "rational" answers to what "rationalization" yields—some favorable warrant or appraisal. Quine does not pursue the matter, but his influence (in this regard) has been extraordinary. Nowhere, I may add, do theorists like Quine, Davidson, or Rorty demonstrate that legitimation is explanatory in the causal sense, or may be replaced by causal explanation, or may be abandoned because it is not causal in nature. That is a remarkable oversight.
If, then, (5.13) holds—as I believe it does—then a very large part of contemporary analytic philosophy is simply misdirected. At the very least, there is a contest here that cannot be ignored and that depends on
the upshot of the two considerations—(i)–(ii)—listed just a moment ago. I think the meanings of "rational" and "rationalization" (that is, explanations of why this or that policy or action is "rational") are clear enough, but I shall postpone a closer study of the matter.
One other (last) topic that suggests itself concerns the famous "definition" of knowledge, namely: that knowledge is "justified true belief" (JTB ).[9] The formula is surely the single most general epistemological finding in all of Anglo-American philosophy. It is not quite the same as Plato's formula in Theaetetus: first, because, on a literal reading of the dialogue, belief and knowledge are associated (by Socrates) with entirely disjoint domains; second, because, again literally, the confirmation of knowledge in Theaetetus depends on archic and privileged grounds. But, Plato aside, there is a lot that is unpromising about the formula. For one thing, it is intended in a criterial way; for another, it is usually read in the externalist way. But, if we abandon transparency, then (as already argued) truth can serve no criterial function. (This was the trouble, you remember, with Putnam's insisting on truth's "regulative" role—truth as a Grenzbegriff —within the terms of his "internal realism.")
Clearly, the attraction of the disquotational theory of truth (as in Quine's account) is just that its adoption seems to facilitate the naturalizing of epistemology: "rationalization" (as a form of explanation) seems suddenly to be replaceable by first-order "causal explanation."[10] (The point of the maneuver, of course, lies with the fact that, in the JTB theory, "T" is not indexed to any would-be cognizing agent: as such, it defies the benign purpose of causal replacement. Hence, it must be retired!) Still, I know of no sustained account that makes the case convincingly. It is not irrelevant to mention in this regard that, in a deservedly famous early paper, Davidson did indeed attempt to demonstrate that "explanation by reasons" (rationalization) was a "species" of causal explanation. Unfortunately, the argument was a non sequitur: from the fact that reasons (the "having " of reasons) may be a cause (of an action), nothing follows regarding whether explanations by reasons are or are not a species of causal explanation.
Two attractive options hang in the balance: one, that rationalization may be replaced by causal explanation; the other, that rationalization just is a form of causal explanation. If, as I surmise, both fail, then we are bound to conclude that naturalized epistemologies fail ((5.13)), and
(5.14) legitimation is a form of rationalization,
where, by rationalization is meant (i) explanation by reasons, (ii) in accord not with causes (though reasons may be causes) but (iii) with some model of "rationality" (of what it is to be "rational" in the way of belief and understanding). Since, however, rationality in the way of belief cannot fail to have "adequational" pretentions, we must add that rationalization in epistemic matters is (iv) a second-order, not a first-order, concept. There is a promissory note here, of course, but we have turned an important corner in favoring the finding:
(5.15) where rationalization is epistemically pertinent, the domain explained is not and cannot be "naturalized."
I understand by naturalized, then, any explanatory account that (i) is causal in structure or governed by considerations of causal explanation alone, or is otherwise confined to first-order (usually physical) resources (whether by obviating or by "reducing" legitimative matters), and (ii) specifically precludes rationalizations or explanations that are not causal in structure or are not governed by the requirements of causal explanation. I do not say that rationalization is, or ought to be, or may be permitted to be incompatible with relevant causal explanations. I say only that the logical structure of a rationalization is essentially different from that of causal explanation. (More will need to be said, particularly regarding legitimation as a species of rationalization.) I should add, for compendiousness, that (iii) naturalisms are, characteristically or strictly, externalist doctrines. (In this sense, Kant's philosophy cannot be a form of naturalism, although, since Kant opposes symbiosis and is an apriorist, his doctrine, thus interpreted, is as close as a "transcendental idealism" can come to externalism.)
But it needs to be said here (Davidson makes the point compellingly) that to "explain by reasons " (to "rationalize") is to invoke the internal relata of some "holist" schema in terms of which human thought and behavior may be normatively appraised (as "rational"). Thus, one cannot speak of what an agent intends to do, except when properly coupled (relationally) with what that same agent may be said to believe: belief, intention, and action are predicable, therefore, only on the assumption of a ("holist") model of "rationality" (in a new sense of holism). The obvious difficulty in this (for "naturalizing" strategies but not for ordinary inquiries) is that the explanatory model conforms (so far forth) with the principle of internal relations—and in explicitly epistemic contexts—whereas causality is thought to behave extensionally (Davidson).
It is not enough that rationalization be normative in structure. It is normative, but, more than that, its normative structure is second-order . (I return to the issue in chapter 11.) I must add that "holist," in the epistemic context of rationalization, signifies that (i) the explanantia are treated as relata, but (ii) are employed as such by cognizing subjects already suitably individuated and functioning in our (textualized) world. There is no paradox there, but we may mark the difference by subscripting "holism" thus: holistr (where "r" stands for rationalization). Naturalizing fails, therefore, both because it will not admit explanations other than causal, and because it treats all discourse as suited to extensionalist regimentation.
We have reached a benign stalemate. I shall regard it as the promise of a victory over naturalizing philosophies. You cannot fail to see that if the argument holds for epistemology, it is bound to hold for every inquiry that depends on legitimative matters (notably, as we shall see, in moral matters). I conclude directly, therefore, that
(5.16) science cannot be naturalized;
and, of course,
(5.17) philosophy cannot be naturalized.
I have perhaps led us out on a limb, but the argument is not in the least unpromising.
I end this chapter with two brief remarks. First of all, there may well be causes of (our) beliefs, and there may (therefore) be a causal explanation for our having the beliefs we have. Nevertheless, there can be no defensible sense in which the causal explanation of our beliefs counts as an explanation of our knowledge (if what we believe, we also know), unless the very analysis of knowledge can be shown to be of an entirely causal sort. If, for instance, there is a conceptual connection between knowledge and belief —one in which knowledge is no more than a normative ranking or grading of belief—then (I should concede) knowledge might be causally explained. But if that normative connection is such only if legitimated in terms of what it would be rational to believe (so that the explanation of why that counts as knowledge is itself a form of [second-order] rationalization: holistr ),f then (I should say) the causal model was not the right one. It is a fact that many analytic philosophers (Quine of course, Pears, Goldman) have offered strenuous defenses of
the causal theory of knowledge. I say they are off the mark if they cannot show (I don't see how they can show) that the explanation of what knowledge is is not merely normative but legitimative as well. That is,
(5.18) the norms of knowledge apply only to what is antecedently rational(ized).
By normative, let me say at once, I mean whatever (whether first-order or second-order) concerns "ranking" or "grading" with respect to some norm or excellence. From what has already been said, it should be clear that legitimation is a form of explanation that is (i) second-order, (ii) normative, as regarding what it is rational to believe, (iii) with respect to philosophical questions (about knowledge, about the world, about conduct, or the like).
But if that is so, then
(5.19) no "proper parts" of knowledge can, as such, be causally explained,
which, of course, is hardly to deny the relevance of causal considerations. By a proper part of knowledge, I mean anything that either (i) is wholly specified as what is known or knowable (facts, preeminently, as distinct from states of affairs ), or (ii) is what may be specified only in ways that presuppose or entail knowledge or cognitive success (reference, for instance, or any supposed successor to reference, as with Russell and Quine), or (iii) is ingredient to knowledge in the way of functioning criterially as to whether knowledge obtains or does not obtain (the ingredients of JTB, for instance). Bear in mind that, admitting intransparency,
(5.20) knowledge of what exists or is real cannot be causally explained.
For, of course,
(5.21) knowledge is a form of rational belief adequated to truth.
Also, of course, some believe—I do—that one can refer to fictional entities (that is, entities that do not exist). If so, then, in accord with what is usually intended,
(5.22) the causal theory of reference fails,
that is, the theory that successful reference (perhaps as an instance of JTB) can be explained solely in causal terms. (Theorists like Kripke and Putnam, then, are either mistaken or, perhaps more accurately, never subscribed to the causal theory of reference as such.)[11] The more important finding is that the philosophical (and legitimative) question about what to count as successful reference cannot itself be ensured in merely causal terms.
Sometimes, the issues are more indirect. For example, on the JTB (justified-true-belief theory of knowledge), paradoxes are known to arise—famously, the "Gettier problem."[12] What is important about the Gettier problem (there are puzzles of other kinds as well), is that a question properly arises as to whether, in counting this or that as knowledge, certain inferential rules are admissible: e.g., that if it were reasonable to believe that "p, " then it would be reasonable to believe that "p v q "; and then, if it turned out that "p " was false but "q " was true, that instances of that formula could ever count as knowledge. One counter-argument holds that, in epistemic contexts, the inference from "p " to "p v q " is invalid, unless "q " is suitably introduced, antecedently, as an independent belief of the rational agent in question. Similarly, the inference, in epistemic contexts, from "Peter is at home" to "Someone is at home" is invalid; perhaps the correct inference is: "Someone [who is Peter] is at home." That is, from a singular proposition one can only infer (in context) the conjunction of that singular and the existential generalization that (in standard logic) would be admitted. (Gettier-like paradoxes are thereby obviated.) Examples of this sort confirm that the applicability of the would-be rules of logic is a function of the domain to which they are said to apply. So the objection to Gettiera counterexamples (which trade on the rules just mentioned or similar rules) raises the legitimative issue. Hence, what appears to be a purely formal question is no such thing, and therefore may be brought to bear against the supposed adequacy of the causal model. (Of course, the argument counts against any exceptionless adherence to bivalence as well.) All matters affecting the truth of beliefs, constative acts, and reference and predication inevitably implicate the holist model of rationality. You cannot naturalize cognition if you cannot naturalize the epistemic powers of cognizing agents. On the argument, you cannot do that, unless you can naturalize legitimation; for truth is not merely normative, it is also legitimative. Simply put:
(5.23) naturalizing knowledge is an ellipsis for naturalizing the cognitional and rational powers of human agents.
In short:
(5.24) naturalism (also, its rejection) implicates an ontology of persons or selves.[13]
By these devices, and following strategies invoked earlier, analytic models of knowledge are shown to be deficient, once cognitive privilege is denied, without yet invoking the puzzles of history and cultural preformation. In this way, the critique of standard epistemological questions is managed on grounds internal to analytic practice. The need for ampler conceptual resources is therefore drawn from its deficiency, and we begin to sense the advantage of going beyond the usual accounts of Anglo-American philosophy. The puzzles of part I of this primer lead ineluctably to the ampler vision of part II.
Chapter VI
Existence and Reality
There is a potential conflict between the spare taste of modern conceptions of logic and the ancient philosophical concern to exclude nothing from our ken that has the least claim to be said "to be."[1] Between Parmenides and Quine the tradition raises a remarkably constant question about "what there is." Quine and Parmenides answer in the slimmest of ways. Parmenides says: what is, is; and what is not, is not; and that is necessarily so. Of what is not, nothing (more) can be said. Quine says: what is is "the value of a bound variable"; we discern, in what one says, what one says "is."
Parmenides's view has been nearly universally construed as a disaster, since it appears to exclude from "what is" (or "what there is") whatever may be said to change (for, on the argument, what changes "is not"). If so, then all the sciences of nature would be swept out as meaningless or incoherent. There is the sticking point. The entire Greek world, including the Presocratics who were familiar with Parmenides's thesis, and Plato and Aristotle, were obsessed with reconciling the world of change with what appeared to be the ineluctable truth of Parmenides's dictum. The supposed elegance of that dictum obliges us, it was held, to admit that to affirm that "what is not, is" (which the atomists tried to defend) is contradictory; hence, that what "is not" is also impossible to treat as a suitable object for the constative discourse by which we make truth-claims about what genuinely "is."
But there are grave difficulties with Parmenides's teaching. It forces a philosophical concession from us that appears to collide, although in
a very different way, with the intended economy of Quine's formula. Quine's resolution is similarly open to challenge. One is inclined to ask of both Parmenides and Quine: When you speak of "what is," would you please explain what you have in mind? "What is what? " You don't understand, we may suppose each to reply: I am speaking only of "what is "!
The truth is that Quine is quite clear that when he speaks of "what is" he is speaking only of the individuatable things (things capable of individuation) about which we utter and confirm certain (favored kinds of) truth-claims in certain suitably formal paraphrases of what we ordinarily say. In the current jargon linking philosophy and logic, it is said that Quine uses "is" in its "existential" sense: the sense governed by the canonical use of the "existential quantifier." The implication is—and it seems quite correct—that Parmenides failed to grasp the difference between that existential sense of "is" (or something like it) and its predicative sense—that is, the sense in which the question I put a moment ago is pertinent, namely, "What is what? "—the sense of "is" in predicative position ("being red," "being round," and the like).
Parmenides seems to have misled himself into thinking (i) that "what is" ("existentially") cannot support any form of ("predicative") discourse about "it" that affirms change ("what is not"), and (ii) that "what is not" ("existentially") cannot be spoken of at all (either "existentially" or "predicatively"). I say Parmenides is wrong on both counts. Not everyone agrees. The philosophical tradition has struggled since that day to distinguish between at least the existential and the predicative senses of "is." Some would say—I certainly would—that Sherlock Holmes was ("is") a fictional character, "something" that does not and never did exist, but that, nevertheless, I can (counterfactually imagining Sherlock Holmes to exist) talk about him, mention him, and even say what is true of him (that he never married, for instance). If so, then, (i) it is entirely possible to speak (in the existential sense) of what does not exist, and (ii) it is also entirely possible to affirm (in the predicative sense) what is true of "it": without confusing what is actual and what is not actual and what changes and what does not change—and without producing paradox as a result. If so, then Parmenides's stern dictum is either false (in supposing that one cannot speak of what "is not" [existentially] or say anything of "it" [predicatively]), or false (in supposing that one cannot say of what "is" [existentially] that "it" changes [predicatively]), or preposterous (in supposing that one cannot ever say what is false—either existentially or predicatively).
Part of what is needed may be supplied by distinguishing between what "exists de re " and what "exists de dicto ." (These are meant as terms of art with respect to the existential sense of "exists.") Let us say that
(6.1) the existential sense of "exists" is a purely formal sense, indifferent to the distinction between "de re existence" and "de dicto existence."
(The "existential sense" of "exists" is a misnomer.) I suggest we treat (6.1) as no more than a terminological recommendation. The point of offering it, however, is this:
(6.2) formal or syntactic analyses of sentences entail no de re consequences; de re import is a function only of "interpreted" sentences,
where, by interpreted, I mean only: that feature of sentences (apart from their syntax) in virtue of which they are construed, when uttered, as conveying information about the actual, or some imagined, world.
On that reading, "exists de dicto " signifies, trivially, that what "exists" is no more than the grammatical referent of a given discourse ((6.1)), suited for predication—in the abstract space of that discourse —whereas "exists de re " (usually) signifies that what "exists" (de dicto ) may also, in some further (ontic and epistemic) sense, be discernible in the actual world . Clearly, there is no logical compulsion to concede that what "exists" de dicto also "exists" (because of that) de re —as if to say (as Searle seems to say)[2] that fictitious "entities" are a part (de re: if only a peculiar part) of the actual world. Nor is there any logical compulsion to say that whatever "exists" de re also "exists" de dicto . That might suggest the preposterous view that there is nothing that is actual about which we do not already know enough to have made determinate referential provision for, or that what is actual is made such by being spoken of "existentially."
But if these corrections are conceded, then they fall afoul of Quine's thesis. For, if I can speak of Sherlock Holmes without supposing that Holmes exists de re (is real: "is," in the sense Quine seem to favor), then, either (i) Quine is wrong to suppose that the "existential" sense of "is" (that he favors) must (for reasons beyond his own logical tastes) concern what actually exists (or is thereby taken to exist) de re, or (ii) in speaking of "any thing" in the de dicto sense intended, we are necessarily com-
mitted to the entailed claim that "it is" (also) de re (that it actually exists). That is a plain non sequitur . The existential sense of "exists " (or "is") is a purely grammatical sense, signifying "that about which" one is speaking—alternatively, the intentional sense of our discourse —is sometimes captured (grammatically) by the denotata of the sentences in use or the referents of our speech acts.
Quine never explains what the sense is, relative to (ii), to which in making predications of what we are speaking "about" (de dicto ) we necessarily, for purely syntactic reasons, commit ourselves, de re, as well. Is it a recommendation about how to view the canonical (syntactic) analysis of so-called "existential generalizations"? For instance: if I say "Some cats are fierce"—and if I agree with Quine about the standard formal representation of what is thus said, so that I am, in effect, committed to the truth of "($x ) (Cat x & Fierce x )"—is it also true that I am (must be) thereby committed (de re ) to there "being" at least one actual cat? Am I committed "existentially" in the de dicto sense alone (grammatically)? Or am I also committed to the "de re existence" of cats (their actual existence in the world)? And if I am committed in either way, on what grounds am I thus obligated? Is it contradictory or incoherent to say (in ordinary English), "Some unicorns are fierce" and believe (in so saying), or (even) actually say, that there are no unicorns (that no unicorns actually exist)? It certainly seems possible. Or, am I committed to unicorns in the de dicto sense (alone) when I predicate fierceness of "some" of "them "? The latter seems a ponderously worded but utterly vacuous consequence; the former seems arbitrarily strong—and false. (Clearly, Quine is bent on obviating any idiom in which we may be said to "intend" to be speaking "about" this or that—which might then claim to be able to serve as the value of a bound variable, although not because of the natural-language grammar by which the logical clarification may be standardly endorsed.)
It clearly won't help to say, "Some unicorns (although imaginary as far as our world is concerned) are fierce," because "they" exist in some "possible world" (other than our own), in virtue of which (somehow) we refer to them in our world. This looks very much like a Rube Goldberg device for referring to what we should otherwise say, tout court, was imaginary. (And what of "possible worlds"? Do they exist or are they real, or are they merely imagined to be real? Which, thus uttered, duplicates the original puzzle.)
My own recommendation is this. When introduced referringly, possible worlds[3] only "existde dicto, " that is, trivially: they "exist" in the
"existential sense" of "exists" ((6.1)) that I just assigned Quine's account (against his own interpretation). Thus I recommend that
(6.3) there are no "possible worlds," other than the actual world, that exist de re .
Some philosophers (Goodman, notably) insist that there are many actual worlds (though Goodman neglects to tell us how to individuate and identify them). Others (Kuhn, for instance) worry that, within the (one) actual world, there are many "different worlds" that may be demarcated (according to Kuhn: Priestley's "world" and Lavoisier's "world" at least, which Kuhn has no intention of construing in Gadamer's way). Still others (Lewis, for instance) hold that "possible worlds" are in some sense "real" (or at least not imaginary). As far as I can see, what is possible signifies what, (i) for predicative purposes, (ii) is (on independent grounds) actual, exists, or is real, or (iii), relative to what is (thus) actual, might exist or be real, or (iv) is (in accord with some idiom projected from the conceptual resources of constative discourse) conceptually compossible, that is, coherent and free of internal contradiction, paradox, or the like (in effect: relative to whatever is admitted to "exist de dicto ").
I cannot see how to prove, in any nontrivial or nonarbitrary way, that (6.3) is true or false: it seems to me to be no more than a prudent philosophical recommendation, an economy. I cannot see how, otherwise, to avoid the endless paradoxes that "possible-worlds" discourse generates. I regard such discourse as a conceptually dependent idiom, therefore. That is, I cannot see how "possible-worlds" talk can be detached, in any interpretively robust sense, from talk of what is "possible" relative to the actual world. I take this to be implied in imagining "possible worlds" counterfactually close to the actual world or closer than others—for instance, in adhering to the causal laws of our world but not to certain admitted facts about it (thinking of how, say, the moon might have been produced).
There seems to be no discernible syntactic difference between "Some cats are fierce" and "Some unicorns are fierce." It cannot (then) be said straightforwardly that (for "purely logical reasons") the first is true and the second false. Because, it may well be that "Some unicorns are fierce" is true, either (i) because (surprisingly) James Thurber knew something about the actual world the rest of us do not, or (ii) because, in spite of there not being any actual unicorns, "they" (the ones being referred to)
are fierce (de dicto ). It seems reasonable to say that we should not want the logical form of "Some unicorns are fierce" to change, magically, as a result of contingently discovering unicorns in the garden. It also seems reasonable to say that we can talk about imaginary unicorns, if we are careful.
This is obviously an overly busy way of speaking. Two findings are clear. For one:
(6.4) there is no incoherence or paradox in predications made of referents we suppose do not actually exist,
as in a deliberate fiction or in imagining something to exist that we believe does not.
We may recommend against notationally "entrenching" (in some supposedly "canonical" way), in the logical structure of sentences like "Some unicorns are fierce" but not like "Some cats are fierce," the familiar natural-language assumption (or intention) that, in speaking of fictitious things, we do not ever mean to commit ourselves to their actually existing. But our policy about formal notation need not affect or alter our ordinary linguistic habits—or the actual world, for that matter. If we suppose (with Quine) that the logical notation I've loosely employed for "Some unicorns are fierce" (involving "quantification," as Quine would say) should (or must ) be construed as entailing an "ontic commitment" (as he also says) to unicorns—one that expressly signifies that at least one unicorn actually exists (de re ) because (and, here, the weight of that "because" is quite uncertain) we have agreed to use a particular term (designating a member of a would-be set of things called unicorns) to designate a "value" (or a variable to which a set of "values" correspond) "bound" (as we say, with Quine) by the existential quantifier ("$ "), now taken to range over our representation of the sense of the ordinary English sentence "Some unicorns are fierce"—then (and here's the second finding):
(6.5) the syntactic representation of "ontic commitment" cannot be more than conventional.
This fits very nicely with what I have already shown, in chapter 2, regarding (2.4) and (2.5), and in chapter 3, regarding (3.10), which, taken together, argue that, although alethic and ontic considerations are inseparable, prioritizing the one over the other (epistemically) amounts
to advocating a form of cognitive privilege. (Quine might then be accused of an excessive zeal—of just the sort I claim to have found in Dummett.) We see, therefore, the potential danger of reading Quine's analysis of ontic commitment in too sanguine a way. I say it is a purely verbal and formal device, utterly lacking in ontic import by itself, or simply obscure regarding the import it acquires as a result of the larger philosophy in which it is embedded. It may be true that we cannot resist admitting (in general) that "something (or other) exists" (actually exists in the real world), but that is hardly because of a choice of logical notation; also, that hardly commits us to some unicorn existing (as opposed to something existing ). Quine has moved too quickly. The utterance "Something exists" strikes me as belonging to that set of informationally degenerate utterances , of which "Cogito ergo sum " is the most famous: those, that is, whose content never exceeds the information entailed by their being uttered.
I don't deny that we presuppose that something or other exists, but I cannot see that we are necessarily committed to the entailment (that some unicorns exist) as a consequence of employing (as in my illustration) the "existential" sense of "exists" ((6.5)). Furthermore, there is every reason to believe that the supposed assertion "Something (or other actually) exists" cannot but be logically derivative from the more robust (epistemically and ontically freighted) assertions and commitments of actual life. Singular assertions dominate market life, you see. (That was just the point of worrying Gettier-like counterexamples in the last chapter.)
This helps to confirm that
(6.6) we cannot determine what is actual, or what we take to be actual, solely from the syntax of what we utter;
and, as a consequence,
(6.7) ontic commitment, effected by the use of any canonical notation—for example, by uttering the existential quantifier ("$ ")—is equivocal: as between (i) conforming with the mere de dicto use of that notation and (ii) intending, by its use, to represent accurately, de re, some real-world belief.
On the fairest reading of Quine's notion, ontic commitment is (still contrary to Quine's own proposal) properly given by (ii); in (6.7) it does
not encumber the actual world or our actual commitment, but even that signifies only an intention to convey what one supposes to be actual. I can put this more provocatively:
(6.8) "ontic commitment," in sense (ii) of (6.7), is "intentional" and cannot be captured by the resources of sense (i), which is "informationally degenerate,"
in the sense of "intentional" in which one deliberately acts to satisfy sense (ii). (Effectively, this is the import of (3.10).) I take it that Quine's usage is itself "intentionally" motivated in this sense (although, of course, Quine's avowed intention—in his program for formulating a canonical notation—was to retire all the intentional constructions [in a different sense of "intentional"] of natural-language discourse of the sort Brentano thought he had captured).[4]
Logic, abstracted from the context of the normal use of natural language, in which language first acquires its "ontic commitment," has no ontic import of its own. This accords completely with the sense of (2.4), linking language and world indissolubly, but it also confirms (6.2). Hence,
(6.9) there is no uninterpreted syntactic relationship between natural languages and logic that determines ontic commitment.
The negation of (6.9) contradicts (6.8) and is incompatible with (3.1), which affirms that language is not an autonomous domain, as well as with (4.21), which makes ontic commitment—in fact, all truth-claims—interpretive.
Hence,
(6.10) no logical canon said to represent the "ontic commitments" of a natural language is more than an abstraction from natural-language use, just as the analysis of a natural language is no more than an abstraction from larger lingual practices.
This—theorem (6.10) or something close to it—is what I took Dummett to have arbitrarily opposed. Theorem (6.10) is, I believe, the master theme of Wittgenstein's conception of "forms of life" (Lebensformen ) of the post-Tractarian manuscripts. (It has taken a bit of labor to isolate that fact.) It is a specimen of the most profound modern analogue of
Aristotle's refusal to separate the forms of argument from the practices of natural-language rhetoric, except that Aristotle, as we have seen, opposes (2.1), the denial that reality necessarily has an invariant structure. Wittgenstein's thesis is: (i) the minimal doctrine consistent with (2.1) and (6.6); and (ii) a doctrine specifically in accord with the claims of symbiosis and holism and intransparency that are central to my own argument. Wittgenstein's notion is also the natural replacement for Kant's "constructed" picture of phenomena, under the condition just mentioned (the "post-Kantian" orientation), so long as we still lack an account of historicity. (Wittgenstein has almost nothing to say about history and, truth to tell, little to say about Lebensformen .)
These considerations bear, of course, on the resolution of the Gettier problem that surfaced in the previous chapter: there is no transparent, purely syntactic rendering of any of our(or commitments) about existence or reality or truth or knowledge. The Gettier problem is a genuinely profound problem if it is taken to embody and legitimate a philosophical policy, for
(6.11) logic, construed as an autonomous and uninterpreted discipline, sets no prior constraints on—indeed, has no particular relevance for—the analysis of lebensformlich discourse.
We see, therefore, that we have been proceeding by quite small steps toward the following straightforward finding:
(6.12) attributions of existence and reality are not merely alethic, logical, syntactic, or formal in any sense at all; they are inseparable from our constative powers intransparently "entrenched" in our Lebensformen .
By entrenched, I mean (adding to an earlier recommendation): (i) tacit, incompletely fathomable, within the practices of our Lebensformen ; (ii) made manifest by exercising our linguistic and lingual powers; (iii) judged in a "holist" sense to contribute to our society's continuing survival; (iv) hospitable, therefore, to determinate realist claims within that same holist space; and (v) rendered determinate only by reflexive interpretation (naturans: by "parsing"). (I seize the occasion to insist on the following usage as well: I say [entirely without argument] that any philosophy that is (i) an-archic, (ii) committed to symbiosis and intransparency, and (iii) disposed to construe our epistemic powers as "en-
trenched" in the sense just given, is [in my usage] an expression of [philosophical] pragmatism. Dewey is the exemplar.)[5]
I should perhaps take an additional moment to make a few further distinctions clear that bear on what has already been said. The distinction between de re and de dicto discourse (that is, discourse about reality and language) makes instant sense for anyone who thinks along Aristotle's lines or for anyone who accepts an externalism or a strong form of the correspondence theory of truth. For a theorist like Quine, there is really no way of distinguishing in principle between the two: any attempt to do so would violate Quine's treatment of the analytic/synthetic dogma (although there are always contingent uses that may be needed). Furthermore, the de re/de dicto issue is not the same as that regarding the distinction between the "material mode " and the "formal mode " of discourse, which both Quine and Carnap entertain.[6] The second is merely a difference in a "style" of discourse, a difference in two different ways of making the same assertions (regarding which both suppose the "formal mode" to be the more perspicuous). On the second view, to speak as if we were speaking of the parts of language or the parts of the real world ("semantics" and "metaphysics," say) is a façon de parler that does not affect the substance of what we say. When, therefore, Dummett prioritizes semantics over metaphysics, it looks as if he favors the first distinction (de re/de dicto ) rather than the second. Furthermore, the de re/de dicto distinction is usually invoked in speaking of necessity, or natures or essences, or existence, or in some related way. Quine clearly subordinates the first distinction to his preference for the "formal mode" regarding the second: hence, he speaks only of "ontic commitment." Theorists like (the later) Goodman and Lewis cannot concur, since the first admits many actual worlds and the second, many (robust) "possible worlds" of which the actual world is but one. I see no way of denying that "is" or "exists" is used, constatively, equivocally, in a way natural-language does not capture syntactically—which, as I've tried to show, Quine obscures unnecessarily. I think this signifies that there is a point to holding to the de re/de dicto distinction—in intentional terms. When, however, we speak of "natures" or "essences" in accord with that distinction, we usually mean to speak, respectively, of real essences and nominal essences in a sense akin to what, in their very different ways, Locke and Leibniz intended. There, too, the distinction has a use.
Let me now return to the original issue raised about the existential sense of "to be": Parmenides's confusion is now easier to resolve. For,
if Parmenides did not merely confuse external and internal negation (negation accommodating, respectively, the existential sense of "exists" and negation in the predicative sense: as, respectively, in the sentences, "it is not true that there are fierce cats" and "it is true that some cats are not fierce"), then Parmenides's doctrine must be entirely arbitrary, since it denies the very possibility of coherent discourse about change; or, if he had indeed confused the two, concluding (somehow) that "what is not" (does not exist) is tantamount to "what is not this or that or thus and so" (for instance, is not round, red, just, beloved—for any range of changeable predicables), then he has unnecessarily impoverished discourse in the deepest possible way. We owe it to Aristotle to have cleared up Parmenides's mistake: being-not-fierce (the "complement" of being-fierce ), for instance, is not the same as not existing at all (the denial that there is [there exists] a certain "a "); the difference is not a mere difference of predicates. It is entirely possible "to be" and "to be non-fierce"—without risking contradiction or incoherence. Parmenides was confused.
As Aristotle puts it, there's a world of difference between "being not this or that" (round, red, just, beloved) and "not being anything at all," that is, "not existing." There must be a third way between Parmenides's "two ways" (Truth and Opinion). That is what Aristotle means by hyle (matter), that in virtue of which what exists undergoes change, without ceasing to exist and without ceasing to be real while undergoing change. (Hyle, however, is a confused notion: it tries to treat an epistemic or grammatical matter as if it were entirely an ontic one.)
Aristotle has his own complex metaphysics, of course (which I have already sought to undermine by advancing (2.1), the denial of invariance). But if, with Aristotle, we concede that "primary substances" (ousiai, particular things) manifest "being" insofar as they have an invariant structure (are, say, instances of fixed natural kinds or species), and if such things manifest "nonbeing" (as well ), since they may be subject to change (kinesis or energeia )—for instance, in the way of generation and birth—then existing things may be real with regard both to "being" and "nonbeing"; for, as Aristotle says, "nonbeing" is not tantamount to "not being at all." Parmenides, he supposes, had thought they were the same.
We may treat Aristotle's recovery of this strategic distinction as an ancestral version of the distinction we require. But I warn you as well that we are not obliged to follow Aristotle (and could not now justify doing so) if we tried to construe our sense of "exists" and "is real" in
tandem with his notion of potentiality and actuality. The latter notion is clearly tied to the doctrine of the necessarily invariant structure of what is real (the denial of (2.1) and the affirmation of the "teleologism" of the natural).
Here, I must intrude a strenuous aside. By teleologism, I mean that theory of change according to which either (i) a "final cause, " the telos or end-state of a finite process, brings about that end-state because it is "actual" and real and effective as such through the entire potentiated process, or (ii) there is a continuum of novel and evolving "necessitations" serially effectuated through each and every (fully) actual phase of that (real) process, proceeding from any stage (of it) to the temporally next neighboring stage, without there being any final cause assignable (prospectively) to the entire continuum or to its aggregated stages.
The formula offered in (i), as by Aristotle (in Metaphysics and Physics ), presupposes a changeless "actual" nature by which the potencies of individual things (sharing that nature) are normally actualized. The formula offered in (ii) applies sui generis to the process of history, to (all the "parts" of) reality insofar as it (they) forms (form) an inclusive (actual) history. This is as close as I can come to fathoming what Hegel may have intended in the Phenomenology . The necessity it posits is historicized, ontically emergent, not lawlike (in the usual inductive sense), not analytic (in the logical sense), but such that, retrospectively and only retrospectively, viewed holistically from the vantage of each evolving stage (hence, "actual" or "actualized") and only thus, the continuous process is, in reality, one of self-realization—as of rationality or freedom—but not by way of prior potencies (in Aristotle's sense). Hegel's account is not "constative," I think, but "mythic" in a sense I introduce later.
I take the trouble to isolate the formula of (ii) because it is the master theme of nearly all twentieth-century teleologized forms of rational optimism and progress. (I find it for instance in Peirce, Popper, Habermas, and Putnam. I suggest that, without confirming the necessity posited, such optimisms are no more than arbitrary. That's to say, they are arbitrary ((2.1)).) Their rejection lays the ground for a robust form of relativism. Heidegger believes that Hegel's notion is an enormous sham, an ahistorical pretense to the effect that he (Hegel) subscribes to genuine historicity.[7] I think Heidegger is mistaken in this. But I admit I cannot find a plausible sense in which Hegel's thesis could be shown to be true or could be tested. Nevertheless, I am inclined to think Hegel may have offered what he took to be a conceptual model irresistible to human
reason reflecting on the phenomenal world under the form of an inclusive and real history. (I take Hegel's model to be heuristically intended. The would-be potentialities are the actualities of a new stage of history read back, teleologically, into the past.) History, for Hegel, is a certain contingent temporal change narratized under the guise of necessity and mythically assigned as the evolving reflection of an all-encompassing "subject" (Geist ). If Hegel means to speak literally of "historical necessity," I confess I cannot see his argument.
In any case, a better account of existence and reality (than Aristotle's) is offered by Peirce, although, for my present purpose, I must (in the same spirit) detach it from the details of Peirce's complex philosophy (which is also teleologized). (I am attracted to part of it, and I am strongly opposed to part of it; but that is neither here nor there.) What Peirce isolates in an extraordinarily clear way is the sense the notion of existence conveys of the brute "resistance" or "opposition" of the things of (and in) the actual world: their causal, affective, effective, reactive, resistant powers vis-à-vis us or other things. "Resistance" is not a criterion of what exists, but it fixes the primitive idea underlying every fruitful theory or criterion of what does exist ("what there is"—the de re "existential" sense I was worrying in Quine's company a moment ago). Quine's view seems tepid alongside Peirce's, although Peirce's does not quite settle the deeper question of precisely what, distributively, does exist. That is, on Peirce's view, existence is "attributed" to the "brute" world, but what is "brute" is not discerned in any privileged way, it is interpretively encumbered.
On Peirce's view, it's clear that numbers and universals don't exist. (Nor do "possible worlds.") They don't "resist" us in the unyielding way a wall does, say. That hardly settles the question of whether numbers or universals are real . They may be real (according to one theory or another), but it would take a version of what Peirce calls "Platonic nominalism" to support the strenuous claim that numbers and universals are not only real but existent as well. The expression "as well" signifies a possible equivocation on the meaning of "is" ("what is," "what there is").
Quine does not equivocate in this regard. He answers the question "what there is" solely in the "existential" sense, the sense of the existential quantifier, although he is not clear (he is downright equivocal, even vague) about whether the "existential" should be construed de re or de dicto —or what that means. His analytic strategy is always de dicto, but his philosophical intent is clearly de re . (That is the point of
his speaking of ontic "commitment" rather than of "what there is" simpliciter .)
Peirce's idea, if I understand it rightly, is that, granted we exist, whatever else exists in the world resists us (or other things), or would resist us (or other things), in the "brute" sense sketched. The intuition is that we would not be able to ignore what exists, as we moved through the world, but that sense may be easily extended. (Think of the micro-theoretical world.) Peirce calls this Secondness: a brute feature of things, taken dyadically, that cannot be reduced to anything else. Existence, for Peirce, is, therefore, the manifesting of "Secondness," ascribed, epistemically, on our awareness of "resisting" (or being "resisted" by) other things (bumping into them, moving them). Cannily, Peirce isolates the distinction of Secondness from "Thirdness, " which signifies primarily the interpreted, "constituted" nature of the intelligible world, but he never fails to acknowledge that Secondness is itself discerned only within the space of Thirdness. I regard this as a much-enriched analogue of Berkeley's robust sense that there is a "brute" aspect of the world we perceive when we open our eyes, in spite of the fact that what is "externally" real is, in some sense, an idea in God's mind.[8]
I like the idea. I find it compelling, sensible, uncluttered, hospitable to indefinitely many alternative ontologies of "what there is." Peirce clarifies what we should mean by saying that something exists, without thereby settling the question of what actually exists. Hence, I recommend we combine the main force of Aristotle's and Peirce's separate views (while avoiding commitment to their special philosophies: invariance, in Aristotle's case; cosmic mind, in Peirce's).
Causality and agency surely count among our best intuitions here. There is no need for greater precision, simply because we are not looking for an invariant or indubitable criterion of existence, merely a clue regarding the plausibility of competing views. Let me put some reasonable theorems before you, therefore. For one,
(6.13) ascriptions of existence are validated and legitimated on the strength of interpretations of what "resists" us and other things in the world.
Peirce speaks here of Thirdness; Davidson and Rorty, derogatorily, of tertia . Putnam, as a Peircean of sorts, defends tertia .
Paradigmatically, existence is a dyadic notion, as Peirce remarks. Nothing is said to "exist" tout court, except, derivatively, in the context
of some interpreted "resistant" space. To say that God exists, for instance, is (apparently) to say that he created the world in which "resistant" things obtain, or that Lucifer "resisted" him, or some such thing; to deny God's existence is to deny all such putative facts. Hence, it is reasonable to hold:
(6.14) causality and effective agency obtain only among existent things;
and:
(6.15) being legitimative, terms like "exists," "is real," "is actual," "is imaginary," "is fictional," "is," and the like cannot be first-order predicates, but they can and do function as predicates in first-order discourse .[9]
Theorem (6.15) will require further scrutiny. Keep it in mind. I hasten to add that I have not, of course, defined "causality" or "agency." Here, what is important to emphasize is that causality and agency "implicate" existent (real) things, are dyadic notions in the sense in which they involve brute resistance (Secondness, in Peirce's idiom). Nevertheless, dyadic relations, Secondness, causal relations, and effective agency are all, like the executive notion of the "external" (of which they are instances), ontically located in some more complex ("internal") symbiotized space (4.11). (I shall return to Peirce's notion of Thirdness in a moment.)
In the sense being developed, we are easily led to concede that
(6.16) individual things exist: ourselves preeminently, that is, individuatable, resistant "things."
The sense of (6.16) plainly favors what Aristotle calls "primary substances" (ousiai ), without having to adopt Aristotle's theories about ousiai: that they possess invariant structures or essences, for instance, that the processes of change they exhibit accord with the logic of "potentiality" and "actuality," that they are inherently directed teleologically to actualizing their essential natures. It also favors Peirce's notion of Secondness, without obliging us to adopt Peirce's full-scale ontology of Firstness, Secondness, and Thirdness, or the brooding notion of cosmic "thinking" apart from human agency. One is certainly tempted to say that it follows from (6.16) that
(6.17) existent things = material things.
Theorem (6.17) is tautological, if we understand material to signify (as I suggest we should) no more than "capable of resistance." Even Aristotle's Unmoved Mover is "material" in this sense. So is the biblical God. (I shall contrast the "material" and the "physical" in a moment.)
Resistance, then, is the putative brute power of existent things contingently encountered as such in the world: derivatively, it is simply our sense of what is undeniably in the same world we inhabit (along the lines, as I say, of Berkeley's sense of opening our eyes and being unable to construe as imaginary what strikes us as palpable and present and cannot be ignored).[10] It is a primitive notion, certainly not criterial, basically pretheoretical, yet theoretically posited; it explains the sense in which the world, although "artifactual" (symbiotized), is also the gathering of those things that we find ("indicatively") the existence of impossible to deny—the things that cannot be swept out by mere inventive interpretation. Resistance is brute, (i) in the epistemic ("phenomenal" or even "phenomenological") sense of being palpably encountered, resistant to the possibility of being denied existence, being explained away (although that, of course, is not a suitable criterion of what does exist), and (ii) in the ontic sense that existence itself (as dyadic) cannot be entirely subsumed as a mere posit within some interpretive scheme of the world ("triadic," as Peirce would say: a mere form of Thirdness). (But I see this as exposing the extraordinarily limp sense in which Quine speaks of existence.) In fact, these distinctions catch up the dialectical connections between naturalism and phenomenology and realism and idealism explored in the previous chapter. (It is for this reason that causality need not be "nomological," lawful or lawlike; causality implicates what exists, whereas nomologicality implicates the interpretation of what is real.) The "brute," I must emphasize, has no evidentiary force, no power to confirm or disconfirm. (It certainly has no "originary" role.)
I am prepared to consider strengthening or extending (6.16), but no ground has been given for doing so. For instance, I see no reason for supposing that an ontology of events, replacing an ontology of material things, something like ousiai, must be ruled out a priori .[11] Such an ontology has its inconveniences, but some philosophers have explored the possibility. Theorem (6.17) certainly does not mean that what exists possesses only physical properties. Physicalism —at least the strongest version of what, in the current philosophical literature, is meant by "physicalism" (that which exists are only physical things possessing only
physical properties)—does not follow from (6.17). Characterized thus, physicalism remains an open question. Correspondingly, "material" (or, "material thing") signifies only that existent things are formed, insofar as they are complex or composed of parts, in a way that involves "matter" (in a sense akin to that of Aristotle's hyle and Peirce's Secondness, except for the caveats noted).
This, as we now understand matters, usually means that existent things do have physical properties, although not necessarily only physical properties. There's the straightforward reason for opposing any idiom that declares that numbers and universals exist (except in accord with our term of art—"exists de dicto, " which is to say, do not, as such, exist de re at all). Lacking "matter," they cannot be located spatiotemporally, they cannot enter into causal relations, they cannot "behave" in the resistant way (in the same world in which we exist). The same may (perhaps) be said of God, in the biblical sense. This points to the clever reason the biblical God is "supernatural." (He makes himself felt through his creation.) So, too, are Frege's numbers! In any case, it is not conceptually impossible to extend the claim of existence to numbers, classes, kinds, universals, or God, but it is (I think) uneconomical and philosophically unnecessary. I see no reason to press the point further.
Conceding this much, we may go on to consider reality . First of all, surely,
(6.18) what exists is real.
"Reality" must be a more inclusive category than "existence." It would be anomalous to deny that what exists is real. But it is also clear that the real cannot be confined to what exists—if, for instance, we oppose saying that the properties of existent things also exist. Propertied things, not properties (whatever that may now mean), manifest resistance and are individuatable. Part of Parmenides's difficulty seems to have depended on failing to keep these distinctions clear. For reasons akin to (6.18), we should then not deny that
(6.19) the "properties" of existent things are real.
Of course, just as the definition of "exists" says nothing about what exists, so, too, the definition of "real" says nothing about the "nature" of real properties. Nevertheless, theorems (6.18) and (6.19) specify what is "real."
That question returns us to the "problem of universals." In Peirce's hands, it obliges us to consider (i) whether there are "real generals," whether the fact that general predicates may in principle apply to plural things (existent things, even fictional things, depending on how we treat "ontic commitment") signifies that the real predicables they designate are, in reality, general, or whether they may or must be construed in an altogether different way, and (ii) whether, if "generality" is real, it is also true that "real generals" exist (that is, whether universals exist de re ).
Peirce says that real generals exhibit Thirdness, which relates predicates to his attractive theory of signs (the exemplars of "Thirdness"); but we may, more neutrally, see in Peirce's Thirdness an acceptance of something very much like the doctrine of symbiosis (or the inherent interpretability of what is symbiotized) introduced earlier, for instance at (4.8)–(4.9): that is, the thesis that language and world, and cognizing subjects and cognized objects, are indissolubly united to form the space within which constative discourse alone obtains and has point ((4.11)). That is, the ubiquity of Thirdness (in Peirce's sense) corresponds (at least in part) to what I have been calling "symbiosis." (I adopt it as a term of art, in that sense.) But I hasten to add that Peirce construes Thirdness as operative in nature apart from human interpretation (or interpretive "constitution")—hence, as somehow the work of a very shadowy cosmic mind. In this sense, Peirce is not entirely a "Kantian" and not entirely wedded to what I am calling symbiosis.
Peirce's extraordinarily clever point (which he offers against Hegel) is that the distinction of Secondness cannot be merely absorbed within the terms of Thirdness (as Peirce thinks—I believe wrongly—Hegel believed),[12] but that Secondness, which accommodates the requisite "external" relationship between subject and object (hence, the individuation of existent things) also constrains Thirdness. "Real" Thirds manifest Secondness, Peirce thinks, and "existent" Seconds, Thirdness. Extraordinary economy! But what does all that mean for us? I think it means at least that symbiosis must yield, somewhere, a working distinction between Secondness and Thirdness. To fail in that is to lose our grip on "objectivity." But so saying is merely to admit the ineliminability of the "brute" world: it says nothing about "it" in any determinate ontic or epistemic sense.
The history of philosophy amply confirms that the principal objection to admitting "real generals" has been the following worry: if they are admitted to be real, must they also (or something on which they depend)
be conceded to exist? Peirce shows quite elegantly that that consequence need not follow. The trick is this. We must be prepared to affirm:
(6.20) what is real, but does not exist, cannot be real apart from what does exist.
That leaves but one option:
(6.21) except for what exists, what is real are the properties, relations, states, and the like of existent things: only what is predicable of existent things can be real.
One imagines it is merely a prejudice of contemporary nominalists that they resist admitting (as existents) "abstract particulars"—numbers, for instance. But there are excellent reasons for insisting on such a policy, without addressing predication, namely: (i) what exists must be "material" ((6.17)), must "resist" other things in the world; (ii) what is real cannot be real apart from what exists ((6.20)); (iii) numbers (and other would-be abstract particulars) would have to be admitted to exist if we acknowledged that they were real (in their separate world, as Frege and Popper concede); and (iv) no systematic advantage thought to be gained by admitting numbers to be real (as well as other abstract "entities") is lost by insisting that their "reality" is conceptually bound to the existence of something else —the material world (or the achievements of science). This, I believe, is the undeveloped motivation of Mill's and Quine's "empiricist" treatment of mathematics and logic. It leads to the powerful finding:
(6.22) logic and mathematics ultimately address a radically abstracted subset of the real predicables ascribed to what exist.
Theorem (6.22), like (6.21), conveys no epistemic import at all: it serves only to distinguish (in the ontic sense) the difference between the use of "exists" and the use of "real." (Notice, by the way, that the mention of "a subset" of [mathematical] predicables no longer conveys, as it would for Quine, more than a façon de parler; for "ontic commitment" now means "commitment" in the benign de dicto sense.) Furthermore, there is an oddity among the dominant natural languages that easily misleads us: we nominalize what we predicate of existent things—as properties, attributes, qualities, predicables. But if we see matters thus, we need not
suppose that properties are more than the (grammatically) nominalized denotata of whatever we predicate of what exists (or of other suitably specified referents). Thus, we say that red is a property (perhaps redness, then, is a "universal"), when we merely (mean to) affirm that Peter's balloon is red. Here you have another reason for retaining the de re/de dicto distinction.
It should be clear that the denial of the existence of numbers in no way settles the question of whether there are "real generals" (whether generality itself is real )—a fortiori, whether numerical properties are real. The nominalist has missed the import of his own intervention.[13] He confuses the matter of the existence of numbers with that of the reality of general attributes —including (of course) attributes of the numerical kind. Strange to say: the nominalist's thesis is a muddle that cannot fail to remind us of Parmenides's muddle.
The contemporary nominalist (Goodman, preeminently), holding that numbers "are not" (do not exist), somehow supposes that that settles the question of whether numerical properties also "are not" (are not real)! Obviously, there is a gap in the argument. Contemporary nominalists (like Goodman and the early Quine) worry, it is true, primarily about whether abstract entities (numbers, classes, kinds) exist . (Quine worries much less than Goodman.) But there are also the ancient nominalists (Ockham, preeminently), who worry rather about whether "generals" (general predicables) are either real or exist . The first kind of nominalist does not address the questions of the second, but the reverse is not true.
If you now recall what was offered in chapter 3 regarding predication—in particular (3.12)–(3.14), which affirm that predication depends upon, is embedded or entrenched in, and manifests, the consensual practices of linguistically apt societies—then one sees that it is not necessary and not reasonable to deny that, in some sense, there are "real generals." There must be : given the spontaneous flow of successful linguistic exchange that extends general predicates beyond any conceivable initial paradigms, nominalism cannot but be bankrupt. (That was a part of chapter 3's argument.) Put very simply: you cannot explain linguistic behavior without explaining predication; you cannot explain predication without explaining the apt use of general predicates; you cannot explain general predicates without admitting real generality; you cannot account for real generals without admitting what we (now) mean by "exists" and "real"; and you cannot account for such linguistic behavior without attention to the enabling Lebensformen, by conforming to which referential and predicative success is vouchsafed.
"Real generals" (the phrase is Peirce's) means, I suggest, that generality is real. By that, I mean no more than that predicates, used in valid predications, regarding what exists, designate what is real. But (real) generality is itself nothing but the joint or multiple predicability of what a predicate rightly predicates (in constative discourse) of plural (existent) things. It is not, as such, explanatory in any epistemically relevant sense. Various and many things, we say, are red, but there is nothing (existing apart, denumerable) that particular things somehow "share," "participate in" (as the usual account of Plato's Forms has it), when it is true that this and that are red.
The linguistic clue—namely, that "is red" is no more than a "part" of a sentence, something that is not even such except in its functional role within a well-formed sentence—confirms that what predicates "designate" (properties ) are no more than the nominalizations of what, in the real world, answers to true predications. The best we seem to be able to say is this: it is existing things' being F that true predications signify: as in, "Those balloons are red, but these are not." There is no F -ness, but things (really) are and can be F . For the same reason, things cannot have "being" or "reality," though they can of course be real—in the predicative sense. Being is not a first-order predicable, as I have said; even in being a predicable, there is nothing that exists or is real, apart from a predicable's predicable function, which we invoke in predicative discourse and thought. (This answers, I believe, both Strawson's subtle questions about subjects and predicates and the medieval puzzles about universals.)[14]
The modern nominalist is wrong on two counts: first, in supposing that the admission of real generals commits us to their existence; second, in supposing that there is a viable way of denying that predicative generality is real. If, then, we introduce the (old) term universals as a term of art, as signifying "abstract particulars" or "nonindividual existents"—those would-be entities nominalized from what otherwise may be predicated of what exists, but are now themselves candidates for existence—we may concede (as a direct consequence of the gathering argument) that
(6.23) universals do not, and cannot, exist.
Between the admission of (3.12)–(3.14) and the grounds for advancing (6.23), all the classical parties to the dispute about universals can be defeated: nominalists and conceptualists, because they suppose either
that there are no real generals since universals don't exist (the thesis of the "nominalists") or that, although there are universals, they exist only in the mind (the thesis of the "conceptualists"), and so-called realists as well, who simply hold that universals do exist ante rem (the "Platonists"), in re (the "Aristotelians"), or post rem (the "conceptualists" again). (Conceptualism is a somewhat labile category, as a comparison of Abelard and Duns Scotus confirms.) Peirce, I should say, failed to grasp satisfactorily that the defeat of nominalism and conceptualism (on his reading) had nothing to do with denying that a proper answer to the question of real generals could (and would) concede an essential role to the way the human mind functions—not some cosmic mind functioning as a deus ex machina invoked to ensure the independent reality of "generals." (Peirce clearly confuses the epistemic and ontic aspects of "generals.") Also, of course, conceptualism is incompatible with symbiosis.
Finally, along related lines, we may take note of the fact that, in the philosophical jargon from Parmenides to the present, one hears it said that what "exists," or is "real," "has being ." It was in fact Parmenides's haste in theorizing about "being" that misled him about the distinction between "exists" and "is real"—which, of course, Aristotle corrected by distinguishing (in effect) between nonbeing (the state of being subject, as such, to change, or the continuing condition of whatever continues to exist as it changes) and not being (the mere absence of existent things). It now turns out (if we insist on the term, inasmuch as its use is widespread) that:
(6.24) what exists is a being —one among a plurality of denumerable beings.
This means that "being" (or "beings") is, as a term of art, used only distributively.
It is worth remarking that "nonbeing" also bears on the ancient distinction between complementary properties (whatever, inclusively, is, as a property, merely other than whatever a determinate predicate designates: "nonred" as opposed to "red," say) and contrary properties (whatever, within some theoretically specified set of alternative properties—in respect of color, say—is other than a given determinate property but is itself similarly determinate: "blue," say, within the range of the "nonred," as opposed to "red"). Greek philosophy seems to have favored contraries where, as with Plato and the Presocratics, it introduces schemes of "opposites" in order to understand change ("Hot"
and "Cold," for instance, or "Moist" and "Dry") as articulations of nonbeing. The Buddhists, by contrast, if I understand them rightly, introduce (not altogether unlike the Taoists) some predicatively undifferentiated source (sunyatta:not a void in the sense of "not being" and not expressible either in complementary or contrary terms) within the holist (undifferentiated) space of which complementaries and contraries play their transient role.[15]
It is also just barely possible, of course, to conjecture about what might be meant by "being," taking the term without any qualification whatsoever—"Being" (as the typographical convention has it), as Heidegger's enormous effort in Being and Time claims to explore. The answer (against Heidegger) is this: whatever we say about "Being" will be derivative from whatever we say about plural "beings." At best, Being (Sein )—lacking number, lacking attributes, lacking a determinate "nature"—cannot, except metaphorically or by abstraction, be said (as Heidegger does say) to be "the ground" of "beings" (Seiende ). Remember: on the argument given just above, what is real is real only relative to (what may be rightly predicated of) what exists. Otherwise, the assignment is patently a form of privilege. Actually, Heidegger's program does involve prioritizing (what he calls) "ontological" inquiry over "ontic" inquiry, that is, discerning the "internal relation" between "Being" and "beings" as distinct from making "objective" truth-claims among ("externally related") beings. There is a curious sense (which the Kyoto Buddhists have pursued) in which Heidegger's Sein has been assimilated to sunyatta, or vice versa. (I cannot pursue the matter further. But I turn, in chapter 8, to a larger issue that bears on Heidegger's extravagance.) In any case, it is conceptually incoherent to treat the "relation" between "Being" and "beings" as epistemically prior to distinctions regarding plural "beings"—"ontic" and "ontological" distinctions, for instance, as Heidegger has it. The result is that, contrary to what Heidegger clearly intends, there can be no principled distinction between the usual discursive properties attributed to individual things ("ontic") and those properties attributed to the same things "in virtue of" their first manifesting (in some sense) the more primordial "relation" of Sein and Dasein ("ontological").
One final question may be ventured. (I have touched on it briefly.) It is nearly canonical to say, with Kant (as in his discussion of the proofs for the existence of God), that "existence" is not a predicate.[16] Kant's charge is certainly justified in a way, but in a way it is not. For instance, it is entirely natural to say (usually in a way meant to be emphatic):
"There are coelacanth. They were thought to be extinct, but live specimens were discovered in 1938. They exist ." Now, there is no way to treat the sentence "There are coelacanth" as ill-formed, but then it does seem reasonable to construe the "are" in that sentence as at least a predicative use of "exists": viz. "($x ) (Coelacanth x & Existent x )." This shows again how unsatisfactory Quine's formal treatment of "ontic commitment" is. Notice, please, that if we elect to construe Quine's "existential" use of quantifiers as having only de dicto import, then we are pretty well obliged to concede that "exists"—in the de re sense, in natural-language contexts—is fairly construed as predicative. (Refusing that possibility, I suggest, leads to worse extravagances.)
A moment's reflection should convince the reader that the quantificational paraphrase of "exists" (Quine's option) obscures rather than informs the criterial question. Certainly, Quine nowhere explains in a sustained way just how alternative schemes of "ontic commitment" are to be appraised. Quine actually obscures the fact that prioritizing "holophrastic" sentences over "parsed" terms violates the doctrine of the "indeterminacy of translation." The truth is, Quine is "ontically committed" by that disjunction, though he denies it.
Construing "exists" predicatively confirms that the pertinent judgment is essentially "interpretive" and that the would-be raw data are "intransparent." But it is also true that the question of "real generals" (as with first-order perceptual discrimination) does not directly arise with the use of any of the "ontic" predicates. That is, mere "ontic" predicates are not descriptive. Oddly, then, these terms appear to fit both the quantificational and the predicative reading. Something is being missed.
Our formal paraphrase clearly raises the paradox Kant had in mind in asking us to distinguish between an imaginary hundred Thaler and a real hundred Thaler . It can't be the case that there "are" two numerically distinct things—both, hundred dollar bills—indistinguishable in all general respects except that one happens "to be" real and the other happens "to be" imaginary! That would oblige us to think of real and imaginary hundred dollar bills as distinct species within a common genus. That won't do at all. (It would mean that an "imaginary" hundred dollar bill did exist—but only as imaginary!)
"Real," in the sense of "exists" ((6.18)), cannot serve as a (first-order) predicate truly ascribed to things in addition to the usual first-order (descriptive) predicates ("round," "red," "just," "beloved"). Although "Coelacanth exist" is syntactically well-formed, it is not predicatively
informative about existent things (descriptive of them) by way of the mere use of "exists." It implies, in a fully predicative sense, only that there are as yet unspecified (real) properties that (existent) coelacanth possess. Hence,
(6.25) predicatively construed, "exists" and "real" and similar "ontic" expressions (or ontic predicates ) are descriptively vacuous (not predicatively informative in any first-order sense);
and
(6.26) the descriptive meaning of no first-order predicate is affected or altered in any way by the attribution (to the same referent) of any ontic predicate ("imaginary" or "real," for instance).
Alternatively put,
(6.27) predicatively construed, "exists" and "real" and similar "ontic" expressions cannot but be second-order predicates, even if they are uttered in first-order discourse.
(By descriptive, I mean no more than that feature of predicates in virtue of which they are confined to first-order discourse: are perceivable or suitably linked to what are perceivable, are subject to change for instance, are of a physical nature, and the like. The distinction will have another hearing in chapter 11, where I consider values as predicates.) For present purposes, it is enough to say that descriptive predicates = first-order predicates. Notice, by the way, that second-order predicates, which by definition are not descriptive—ontic predicates, for instance—may be coherently used in first-order discourse. The distinction is critical for the discussions of the final chapter. I suggest, for instance, that, as a second-order predicate, "exists" is evidentially supported by reference to first-order grounds regarding Secondness. The account is rather trim as a result.
Theorem (6.27) is an interesting finding, because it shows why questions of existence and reality and truth cannot be detached from the second-order legitimative concerns of philosophy . Hence, the arguments of the postmodernists are instantly rendered incoherent. Finally:
(6.28) inherently, natural-language discourse is philosophically encumbered.
We cannot, therefore, escape the philosophical import of our discourse ((1.5)). We cannot treat philosophy as "therapy," in Wittgenstein's sense. We cannot be coherent postmodernists. To speak of what exists and is real is already to implicate the lesson of (6.28). Observe as well that the foregoing argument is intended to show only that a heterodox reading of "exists" is both coherent and philosophically instructive and definitely not ruled out by the force of any would-be canon. On the contrary, it confirms the reasonableness of entertaining alternative philosophical strategies. For instance,
(6.29) the rigor of ontology is not weakened in any way by construing its truth-claims in accord with a relativistic logic.
A further question suggests itself, but I shall only mention it here. (I shall make a brief allusion to it later.) To say that particular things exist, in the sense I have favored, is to say that they are determinate (or determinable) with respect to number (read predicatively). But that, it should be said, does not ensure or entail or require (for the sake of coherence) that what (in first-order discourse) is predicable of them is also, in all cases, determinate . By number, read as a predicate, I understand (i) that predicate (or predicable) (ii) admissible in first-order discourse, (iii) that corresponds to the individuation of particular things—which, you remember, was not a matter that could be decided predicatively at all. That is to say:
(6.30) number, read as a predicate, signifies the determinate individuation of particular things; it sets no constraints on the determinateness or indeterminacy of (other) first-order predicables (for instance, "natures").
Theorem (6.30) will, as we shall see, have an important inning in speaking of cultural entities . It has a corresponding importance in Peirce's system, since, for Peirce, real predicables involve Thirdness—in effect: interpretability. It comes as a surprise, then, that
(6.31) the determinateness of individuation does not entail the determinateness of the "natures" of individuated things—nor, a fortiori, the determinateness of any of their properties.
I shall want to make considerable use of (6.31). For the moment, let me simply say that by determinateness I mean (i) the effective resolution of questions of individuation or of predication with respect to what is individuated, (ii) in accord with the principle of excluded middle. Indeterminacy, then, is similarly defined, except that, by (ii'), excluded middle is abandoned or restricted for cause. Hence, it is quite possible that the individuation of a thing be determinate, although the descriptive attributes predicated of what is determinately individuated may remain (relatively) indeterminate. Peirce effectively acknowledges a similar possibility by treating "real generals" as "triadic": hence, as made (increasingly) determinate only (as Peirce says) by interpretation; or (as I prefer to say) as open to being interpreted and continually reinterpreted because, in addition to being indeterminate, they signify what is culturally significant (Hamlet, for instance).
Peirce's idea is that what is interpreted becomes determinate, without thereby precluding residual or further indeterminacy. I agree, but I depart from Peirce here. For what is interpreted may (I say) be interpreted in an open-ended way, in alternative and emergent ways, and in "in-congruent" ways, and particular interpretations may themselves be reinterpreted in the same sense. Peirce also usefully admits that determinate things may not be determinate everywhere with respect to the application of the principle of noncontradiction; that is, determinate things may be objectively vague (e.g., as in being "bald"). It is easy to see that such a view is most plausible on the assumption of a symbiotized world in which interpretation (Thirdness, in Peirce's idiom) effectively forms ("constitutes") the real things of the world in ways that are of increased determinateness. The fruitfulness of these possibilities will be seen to best advantage among cultural entities. But Peirce also construes these processes (with respect to indeterminacy and vagueness) as irreversible and quite robust; that is, he was firmly committed to a kind of cosmic evolutionism (or teleologism) and did not distinguish in principle between physical and cultural worlds. We need not follow him in these regards. (But the matter bears decisively on Aristotle's archism.) In any case, a powerful consequence follows, namely,
(6.32) the reality of "generals" lies in the (symbiotized) fluency of natural-language predication, not separately, in the mind or in the world; that is, its reality is lebensformlich .
Hence,
(6.33) predicative generality is inherently informal, consensual, not criterial; alternatively put, predicative criteria themselves presuppose the fluency of our lebensformlich practice.
The ancient quarrel about universals is a great confusion; we need no more than "real generals" to secure objectivity. But then, "real generals" have no criterial function either; they are no more than a (nominalized) shadow thrown by objective discourse. That is, if we admit objective truth-claims, then predication must have a realist function. In that sense (alone), there are "real generals." But there are none that can be antecedently discerned, in virtue of which objectivity may be conferred. Given the ubiquity of predication, there is no way to deny "real generals." They have, however, no separable epistemic role. They are implicated in the lebensformlich viability of natural-language discourse.[17] Put another way:
(6.34) real generality is entailed by the symbiotized condition of objective truth-claims; that is, it can be affirmed holistically or, if distributively, then only trivially and parasitically.
I hasten to assure you, in closing this particular chapter, that the analysis of "what is" has produced the greatest philosophical extravagances. Some confusing specimens have surfaced in my account—for instance, regarding Parmenides, Aristotle, the medievals, Hegel, Peirce, Frege, Heidegger, Quine, Goodman, and Lewis. I could easily have mentioned others. I recommend the sparest possible policy regarding the theorems concerned with the meaning of "exists" and "real."
Chapter VII
Identity and Individuation
There is a radical strain in what I have collected up to this point, although what we have is still on the mild side. Certainly, the discussion of the preceding chapter, which moves toward its close by conceding that "existence" may, without paradox, be treated predicatively and that ordinary language is already philosophically freighted and cannot be relieved of that burden, goes contrary to familiar lines of theorizing. But I intend to press the argument along lines that are still more strenuous, as I hinted at the very end of the last chapter. Were the new turn allowed, a fundamental revision in certain well-entrenched ways of viewing philosophy would follow, and that, I suggest, would prepare the ground for even more radical possibilities.
You will have noticed that, in speaking against Heidegger's way of legitimating inquiry regarding Sein —not against the potential fruitfulness of reflecting on "the question of Being"—I said that since Being lacked "number," it lacked "(a) nature." There you have the key to what we must now explore, and what may serve to introduce a bolder line of speculation.
It's an intriguing fact that much of medieval theorizing about God was occupied with this baffling question: How could God's "being" be confined by number and nature without diminishing God's grandeur, or without God's being limited by anything in the order of nature itself? The use of any predicate, it was thought, generated a query regarding the congruity between that attribution and the "nature" of what any predicate could be applied to. Deny a would-be "being" number and
nature, and (it appears) you deny that it exists ((6.16)); admit that it exists, and (it appears) you admit that it is limited by the nature it has, as opposed to the natures other things possess (which it specifically lacks, and generally will lack). Furthermore, particular (material) things are limited by other things having the properties they have—even if they share the same properties ((6.17)). For instance, the volume one person's body fills in an elevator precludes another person's body from occupying that same space. The trick, in Christian philosophy, is to understand the sense in which "attributes" may be ascribed to God without God's having a nature or being one of a number of "beings" that share the same generic attribute (in precisely the same sense in which, in Quine's pretty phrasing, predicates exhibit "divided reference"). For, of course,
(7.1) natures are predicables .
That is,
(7.2) plural beings are intelligible only if, among other beings, they can be assigned "number" and "nature"—privatively; only if they can be individuated and identified; only if they can be referred to and can support predications.[1]
The question of how God may be treated predicatively (without limiting His "being" or powers) is certainly an engaging one; we owe the medievals a word about their athletic feat in reconciling discourse about the ordinary world and what they suppose distinguishes the Creator. But even the medievals recognized that the question has a certain directionality: we must make sense of the "natures" and predicables of individuatable (existing) things and then consider how that idiom may be adjusted to make sense of discourse about God. We cannot proceed in the opposite direction. We must begin, as St. Thomas observes, from what is most familiar. In any case, predicates are human artifacts even if we suppose there are predicables that apply exclusively to God.
In a way, this is the same puzzle that bedevils Parmenides and Plato. Applied "opportunistically," the theologically inspired speculation begins to presume, philosophically, that it can reach to certain discoveries about divinity. That is, from the confinement of the phenomenal world, philosophers like Aquinas supposed that they could discern the de re necessities on which the entire created order itself was originally made intelligible. But it is hard to see how such conjectures could possibly
escape the charge of cognitive privilege. If archism fails (2.3), claims of necessary existence cannot be compelling.
In an eccentric way—which I believe implicitly makes the same point about God and the created order but deliberately chooses not to dwell on it—Heidegger says: "Being is not a being."[2] Similarly, Parmenides treats Being as "one," which is not to attribute number to Being, since the "One" signifies the inadmissibility of plural number within the space of Being. The medievals deny that God has a "nature" in the sense in which plural beings have natures and are individuated with respect to those natures. God is "one," they hold, in the sense that there is no pertinent respect in which any distinction may be introduced, bearing on God's "attributes," that first applies to the created order. God is not, it is said, ontically limited by his own creation. (There may be a need, some supposed, for distinctions regarding God's being triune, but that, it was said, does not bear on attributing a "nature" to God.)
God, as Aquinas says, is self-subsistent (ipsum esse subsistans ),[3] which apparently entails that no predication made of God (or God's being—his "aseity") is conceptually delimited (in the way ordinary predicates are): affirming that something is blue, for instance, precludes the possibility of coherently affirming that it is red ("not blue") under the same circumstances—where the conceptual relation between the predicates mirrors the ontic relation between their designated predicables. God is not rightly thought to be thus affected. (Of course, phenomenal or physical predicates in the ordinary sense do not apply as such to God—"red" and "round," for instance—but what of "potent" and "benign" and "knowing"?) Hence, the predicables ascribed to God must be ontically different in a fundamental way from those that apply to ("created") things, things said to have "natures" or to be particular "beings" (among a multiplicity of "beings").
One medieval solution purports that to ascribe to God whatever "attributes" are admissible as determinations of being—the so-called transcendentals, which also constrain the "natures" of plural "beings" by way of a derived predication (unity, goodness, truth, as Thomas usually puts it)—is to make ascriptions of God, as the exclusive uncreated source of all "there is," without thereby assigning God a determinate nature among plural beings. Inevitably there will be a puzzle regarding attributing properties to God, since any properties (or predicables) in some sense subtended by the transcendentals (but not the transcendentals themselves) will be subject to gradation and a certain involvement with "matter" (hyle, in the sense extracted from Aristotle's account ((6.17)), in accord with our having rejected its intended archism).
Properties rightly ascribed to created things are, in principle, present, it was claimed, only in some admixture of that property and its "opposite ." Natural properties are present oppositionally , both as contraries and as complementaries, present in different degrees, present in such a way that, normally, the capacity of any (natural) thing to possess a property is ipso facto the capacity to lose that property and acquire another—but there can be no such admixture in the transcendentals . For that reason, the transcendentals are (as such) not predicated of any numbered (or natural) beings. Beings that exist and have determinate natures are subject to change, are created and may be destroyed: the continuous career of any created thing must make allowance (it is supposed) for the acquisition and loss, over time, of any property linked to its specific nature. Hence, I understand transcendentals to be defined as predicables such that their "opposites" (in the created order) are not (as a matter of necessity) predicable of that of which they are predicable (God, as in St. Augustine's account). On the contrary, they are necessarily inapplicable, or such that they have no "opposites," although the natural properties they subtend do. I designate predicables that are rightly attributed to things that change as privative, meaning by that only that, where they are rightly ascribed, their "opposites" are also necessarily predicable. (I draw no conclusions about whether there is a God or whether, if there is, God is such that only the transcendentals are rightly ascribed to him. I note only that, since we have invented the idea, we are entitled to judge whether it is coherent and self-consistent.)
This is very close to Plato's view of self-predication with respect to the Forms: the Form of Justice is said, by Plato, in Republic, to be "just," but not in any way that would admit an admixture of the "just" and the "unjust" (which applies only and essentially to the changing, or created, world). On this reasoning, although changing or created things may be said to "resemble" the Forms, the Forms do not "resemble" the things of the changing world. The Form of Justice is (itself) "just" sans phrase, but no human polis can similarly be said to be "just" without its being true as well that, in being just (in its way), it is also rightly characterizable (in some measure and for that reason) as "unjust"—necessarily, not unmixedly just. (Much of this is not my ultimate concern, of course, but it pays to have a sense of the history and complexities of the question being addressed.)
Furthermore, predication or attribution (or, some reasonably strong analogue of predication) does apply to God, insofar as rational discourse about God is possible. (The medievals, you will remember, were divided
in their mind about whether to say that God was discursively ineffable or whether some form of rational discourse could, or even must, be admitted. Shestov, perhaps, has made the most of this, among modern writers.)[4]
Hence, on the argument that attributions may be made of God—although God has no "nature" in the determinate sense in which natural things do—it must be possible to make attributions of what cannot be assigned a determinate and "privative" nature, and there must be predicables (the "transcendentals") that are not, and necessarily are not, "oppositional." The high medieval theory is meant to defeat, or supersede, the ancient Presocratic theory of the oppositional nature of the predicables; of course, it also trades on it. (There's a curious gain here that we shall draw on shortly.)
Leaving God aside, what may we say about the "natures" of things? Certainly,
(7.3) "natures" form a subset of what is predicable of what exists.
That is, if things have "natures," then they may be ascribed attributes congruent with those natures, that are not, as such, ascribed because they are part of the natures of the things in question. Alternatively,
(7.4) things that have "natures" have attributes compatible or congruent with their "natures."[5]
On Aristotle's view, since the "natures" of ("material") things are invariant, although those things are subject to change, ousiai will have accidental attributes as well, attributes that change relative to their changeless natures but that can be lost without loss of "essential" nature. For example, one may at one time remember and then forget this or that, relative to possessing a suitable psyche (or "nature"), but the loss of that "nature" (Aristotle would hold) signifies the loss (the failing to continue to exist) of that ousia itself. Leibniz, on the other hand, treats every attribute of particular things as essential to their being the particular things they are (Socrates's being snubnosed, for instance, is as necessary to Socrates's identity as is his being rational). Theorem (7.4) may be read in a variety of ways, therefore: in Aristotle's way, in Leibniz's, and in a way (in accord with the doctrine of the flux) that (says that) there are no de re necessities at all ((2.1), (2.6)).
What needs to be borne in mind is that Aristotle's and Leibniz's mo-
tivations were entirely different, but, in the present context, equally pointless. Leibniz's concerns individuation and numerical identity; Aristotle's, invariance and change relative to the reality of existing things. Aristotle cannot suppose that the "divided reference" of general predicables ("natures" included) could account for individuation or numerical identity, and Leibniz makes reference and reidentification humanly unmanageable by insisting that the seemingly accidental features of particular things are (actually) essential to their numerical identity. Both strategies may be coherently displaced by conceptually slimmer options. (Still, Leibniz's strategy does bear, as I have suggested against Quine, on the inherent informality of reference.)
Our present topic, remember, is that of individuation and identity. On that matter, I must say that Aristotle strengthens the notion of the "nature" of a thing, making it invariant, even though its identity and reidentifiability are not facilitated in the least by any theory or criterion regarding its general nature only. And Leibniz makes all the attributes a thing instantiates (including the relations it enters into) necessary and essential to its identity as the particular thing it is, in spite of the fact that that incapacitates any independent mortal effort to fix the numerical identity of particular things. (Reference, as I have said, has point only if we can succeed.) I have already shown (i) that Aristotle's modal claim about invariance is false ((2.1), (2.3)), and (ii) that (relative to Quine's claim, which bears on Leibniz's), it is impossible for human inquirers to fix the referents of—a fortiori, the numerical identity of the particular things referred to in—natural-language discourse by means of general predicables alone ((3.2), (3.5), (3.8), (3.9)).
These developments tell us a little more about individuation and identity, but not much. If "natures" were invariant, as Aristotle supposed—if beings (ousiai ) belonged to "natural kinds" (in having invariant natures), in virtue of which they were individuated as the beings they are—the distinction between the "natures" (or "essences" or "definitions") of things and any other predicables would have an entirely legible purpose. Aristotle's science (episteme) presupposes such a doctrine. But Aristotle actually claims, in supporting his view of science, that what is real is necessarily invariant qua real, and that —his modal claim—is simply false. It is true, by (7.2) and (7.3), that anything that supports predication "has a nature" (in some sense—minimally, being subject to predication), but, by (2.1), it does not follow that such a "nature" must be invariant. What, then, could that qualification mean? Aristotle thought, you remember, that to deny things a fixed nature (as Protagoras
is supposed to have done) is to court self-contradiction and incoherence at some point in our discourse. I have shown that that is not a bona fide difficulty.
We may consider the matter in another way. Let us say, reformulating the sense of (7.3) and (7.4), that
(7.5) whatever is truly predicable of what exists must be compatible or congruent with its rightly assigned "nature."
But now, on the strength of (7.3) itself, this means only that, in making predicative claims of anything, true predications must be logically compatible with one another . That is hardly news. Of course, what (7.3)–(7.5) suggest, in the context of (2.1)—the denial of invariance—is that:
(7.6) it is not incoherent, or paradoxical, to assign particular beings inconstant natures.[6]
Now, that is a radical option. (By inconstant, I simply mean "not invariant as to essence.") I freely admit that (7.6) cannot hold if its implementation violates the requirements of reference or predication. But why should it? Still, some forms of predication compatible with (7.6) may not be compatible with bivalent truth-values—for instance, in interpretive contexts. I shall come to that later.
My principal line of argument in chapter 3 had been to construe reference and predication as inherently informal, logically, and as effectively "entrenched" in our Lebensformen, consistently with (2.1). If all that were admitted, then (7.6) would follow at once. We may of course still question referential and predicative practice, but (7.6) belongs to the same theory as the theorems of chapter 3.
The argument is now very trim. We cannot resist theorem (7.6), but we need to understand what it entails. Consider this:
(7.7) existent things ("beings") are effectively individuated insofar as they are identified and reidentified, over time and under changing circumstances, as one and the same individual things;
and
(7.8) no existent things are effectively individuated or assigned number by predicative means alone.
Theorems (7.7)–(7.8) are simply the counterparts of (3.2), which affirms that reference cannot be replaced by any predicative resource. In the ontic sense, therefore, the haecceity of any thing (its "thisness"—its being this rather than that particular thing)[7] cannot be the same as, and cannot be determined by, its quiddity (its "whatness"—its having this "nature" rather than that). In a nutshell, to specify the "nature" of any existent thing is to ignore what specifically individuates it . For, insofar as natures are sets of predicables ((7.1)), "natures" logically admit of "divided reference." You cannot say which horse you see, as between two actual horses, merely in virtue of grasping that what you see is (predicatively) a horse, nor can you say which horse you see by narrowing the scope of its "nature" (or by adding further general qualifications, as "accidents"), unless you know, by other means, that, in the context given, those accidents uniquely mark this horse or that ("a chestnut mare," for instance). Haecceity cannot be merely a narrowing of quiddity.
Many have supposed (quite wrongly) that, in Aristotle's thought, hyle (matter) accounts for the individuation of things with respect to the common natures particular things share with their fellows. But, first of all, since hyle relates to the process of the characteristic development or change of (or in or with respect to) things (that have particular natures), hyle has more to do with predications made of particulars than with actually individuating particulars, or first fixing their identity in any ontic way . Second, even if it were true that every particular thing possessed this bit of "matter" but not that, if we were to bring hyle to bear on the issue of individuation or numerical identity, that itself would require that we individuate hyle . In regard to this, Aristotle has no suggestion at all, beyond the counterproductive admission that hyle has no formal structure. Hyle, it seems, has no "nature"! Hyle is nothing but the abstract capacity of particular things to possess changeable attributes (and to acquire and lose whatever, in Aristotle's sense, is their nature). It was in fact an error on Duns Scotus's part to suppose that haecceitas mediated in some way between the individuality of individual things and their generic natures by virtue of their instantiating some putatively narrower set of predicables (sorted in the mind), said to belong to this rather than to that particular (within the larger scope of their common nature). But—either haecceity is not determined predicatively at all, or, if it is, it requires something akin to Quine's solution of Leibniz's question.
Thus, (i) if reference is inherently informal, if (ii) individuation and
numerical identity correspond, ontically, to what effective reference captures, if (iii) nothing is individuated by adverting to its nature, and if (iv) natures need not be invariant, then
(7.9) to individuate existing things is to specify their persisting careers, not their natures .
I distinguish career from nature (as terms of art) in that "nature" (but not "career") is purely predicative, whereas "career" functions not only predicatively but individuatively as well (that is, in terms of numerical identity). Hence,
(7.10) the careers of things are and must be compatible or congruent with their natures;
and
(7.11) careers are not the temporal orderings of the instantiations of general predicables (whether "natures" or not), but, rather, the temporal orderings of such instantiations in, or as constituting, particulars.
Furthermore, if the argument holds thus far, then, admitting (2.1)—which denies that nature (or reality) is, necessarily, invariantly structured—it follows, trivially, that
(7.12) existing things may constitute or manifest singular careers in spite of possessing (through the length of those careers) "inconstant" natures.
Of course, (7.12) entails (7.6).
I fully and freely concede that a horse (say) cannot remain a horse—a fortiori, the particular horse it is—if it loses its "nature" or if it ceases to be a horse. Hence, again trivially,
(7.13) individuated ("individuating") careers preserve through their "phases" the natures (constant or changing) of the particular things they are.
I say only that affirming (7.13) is not incompatible with affirming (2.1), for
(7.14) the "natures" of existent things need not be invariant, either de re or de dicto .
I am obviously bound to affirm (7.14) as a matter of sheer consistency, but I back it with conviction. It is plainly central to the an-archic undertaking. Beyond that, I wish to affirm a more radical and more strategic claim for which, quite frankly, I have not yet laid a proper ground, namely,
(7.15) certain "kinds" of things are nothing but kinds of careers: particular things of such "kinds" have only careers; or, if they have or (are said to have) "natures," their natures are nothing but careers.
More perspicuously (and more provocatively),
(7.16) particulars of certain "kinds" are nothing but careers; such things may be said to lack "natures."
You see, therefore, the reason for distinguishing with care between the logic of "careers" and the logic of "natures": natures are no more than predicables; whereas, although they may be treated predicatively wherever anything is assigned a nature,
(7.17) careers are referents, or surrogates of referents when viewed in terms of the unicity of their instantiated "phases" over time and change.
From this, it is readily concluded that
(7.18) phases are the temporal parts of individual careers (individual things, individual existents).
To say that "phases" are the temporal parts of individual things (or "beings") may be admitted, but it fails (as Hume appreciated in recognizing the need to go beyond his official pronouncement) to say something about what counts, criterially, in deciding whether we have before us—through time and change—one thing or not, and why .[8]
Famously, Strawson introduced the idiom of sortals (the "kinds" [or better, the terms designating the "kinds"] to which individual things belong), in virtue of employing which, things may be differentiated or counted as the individually distinct things (they are) of the "kinds" they
are. Strawson also introduced phase sortals (the temporally segmented subaltern "kinds" of the sortal kinds to which particulars belong), in virtue of employing which, the particular things first individuated in accord with sortals continue to be implicated as such (that is, more narrowly, by reference to their phase sortals). Thus, an acorn is a phase of an oak tree (taken as a particular thing), but (predicatively) an acorn is, also, an oak tree (perhaps one should [with suitable caveats—against Aristotle] say "potentially," in the sense of being "of the kind," oak tree ).
It is true that Strawson admits "events" into his ontology. But he does not bring the matter to bear on such theorems as (7.15) or (7.16), and he has grave doubts as to whether "events" could be (what he calls) "basic particulars" or whether they are ever more than marginal to a metaphysics (lightning flashes, for instance)—or whether they are any more than the result of some convenient nominalization of what is predicated of rightly denominated individual things ("the stabbing of Ceasar," for instance, or the "stabbing of Caesar at 2 P.M. ," and so on, offered by Davidson for the sake of supporting some would-be canonical notation). Kinds, I should add, are predicables either (i) of the sortal or phase-sortal sort—hence, including "natures" or "phasic" predicables subaltern to given "natures," or (ii) those that do not normally facilitate individuation, or (iii) those applicable to anything that can be individuated. "This yearling" (functioning as a term) denotes a sorted phase of a particular referent's career, whereas "yearling" designates a phasic predicable subaltern to the predicable "horse" (that, as a term, functions as a well-entrenched sortal). However, the term "part of my property" may be read as designating a kind of "thing" pertinent to the law, that is, with respect to something antecedently individuated, without itself functioning as a sortal. "Things" may be individuated without reference to sortal natures, just as they may be individuated, even in accord with sortals, although their "natures" be "inconstant." Certainly, if events may be individuated, then it is reasonable to suppose that individuation need not always involve sortals, although (trivially) it involves kinds. Sortals, we may say, are entrenched kinds: essential or lebensformlich .
To admit the foregoing is to admit that events may be counted as individuatable "things," in the sense that they are merely individual temporal continua . Whether it is always useful to count events, or whether events may, in some sense, be made "basic" to a systematic metaphysics is not a particularly important question. For, on the gathering evidence of this primer (for instance, favoring symbiosis, intransparency, flux,
historicity, the absense of first principles, the defeat of archism), there can be no basic particulars (in Strawson's familiar sense, in his book, Individuals ): particulars on which the existence of all other particulars depends but that do not themselves depend on the existence of other particulars.
I see no reason to suppose that particular "things" that (apart from essentialism, teleologism, and the like) more or less correspond to Aristotle's ousiai need be more reliable, referentially and predicatively, than events (or careers or histories). Certainly, if (7.6) be admitted, the objection cannot be convincing. I find it entirely reasonable, therefore, to concede that
(7.19) careers (or "things") that conform with (7.15) and (7.16) may be construed as individuatable events.
But I do not say that all discourse about careers is discourse only or primarily about events. For there are also "things" (as we shall see) that have natures and have careers, or have natures that are no more than careers, in virtue of which it is entirely possible to view the referents in question as, alternatively, individuated "things" or individuated "careers," depending on which aspect of their individuation is being featured: possessing bodies (things), for instance, or being integrated histories (careers). I see no reason to deny that
(7.20) for certain particulars at least, one and the same individual may, without contradiction, be construed either as a "thing" or as a career (or event).
In the sense that ranges indifferently over referents like horses and lightning strikes, "thing " may be said to signify whatever is merely individuatable. But in that equally familiar sense that means to distinguish horses as "things" from lightning strikes as events, "things" signifies (i) whatever has a sortal nature, whether constant or inconstant, such that (ii) it is sortally individuated as a kind of body or physical object or is sortally individuated in some more complex way that specifically entails sortally individuatable bodies or physical objects. (The full import of this last distinction will be clear shortly. I am running ahead of my story.) If, furthermore, we remind ourselves that "existent things" are "material things" ((6.17)), then, for convenience, we may now construe
the second sense (of "thing") as meaning to distinguish between material things and material events.
On that basis, it is still true that events exist as do things, although it is more usual to say that events occur, where what we mean is that they exist as events . This is clearly the case with those particulars that may, alternatively, be individuated as careers and as material things ((7.20)). For we could hardly suppose that such particulars cannot rightly be said to exist, when viewed under the second usage, if they may be said to exist when viewed under the first. We see, therefore, that
(7.21) the distinction between material things and material events may, at times, be no more than a façon de parler .
It may be doubted, if (7.21) is true, that the same option holds for every "thing" and every "event." In any case, we have secured an additional measure of flexibility. (I shall shortly bring these distinctions to bear on the analysis of cultural entities.)
Clearly, anything that can be "sorted" (or counted) can be (extensionally) collected as the members of a set (or class ) of things. Hence, sets may be heterogeneous with regard to the "natures" of what they collect. If they are sorted with respect to sortals (or kinds), then the resultant sets have assignable "natures," and the "members" of such "sets" are then also instances of (the) kinds marked by their common "natures." (In chapter 10, I introduce a third classificatory distinction, the token/type distinction, the use of which I restrict to "cultural entities.")
Strawson nowhere makes sortals and phase sortals do the work of individuating things. They cannot do such work, on the argument supplied by (7.8). The individuality of a thing is the "being-brought-together-in -the-one-thing-that-is-the-one-referent-of-all-the-instantiations-of-all-the-predicables-that-truly-belong-to-that-one-thing." Leibniz seems to have confused a thing's "individuality"—its being "one" (thing) with what determines its "numerical identity," what "individuates" it, or what is merely true of it; Strawson grasps the complexity of a thing's being individual over time and through change, but he goes on to set constraints on the coherence of ascriptions of numerical identity, not on what makes an individual thing individual. (This accounts for the weakness of Strawson's theory of persons.)
I have in mind, however, making provision for an entire world of entities that Strawson does not discuss, that accommodates what he says
about phases, but I construe his distinction in a way that depends on taking theorems (7.15)–(7.17) more seriously than he does. I can barely introduce them here. Their proper admission depends on much more than I have yet supplied. But they are so important to any comprehensive metaphysics that to mention them here and now is to give some evidence of the deeper plan of argument that governs this primer. Also, of course, the promissory features of what I mean to introduce will not (should not) be permitted to affect the developing argument itself: for the time being, it signifies only a further objective.
In particular, I shall want to establish that
(7.22) cultural entities have, or are, only careers; they lack natures or have natures that are nothing but careers.
I understand cultural entities to be persons, artworks, texts, (certain familiar) actions (for instance, the writing of a check or the signaling of a turn in an automobile or the performance of a dance), words and sentences, histories, and the like. The things of the world of human culture, I hold, are individuated only or primarily as careers. I also argue that
(7.23) cultural entities are (referentially), or have (predicatively), histories.[9]
I must leave (7.23) unexplored for the time being. I think it is fair to say, however, that, on the argument I have in mind, histories are a kind of career: just what kind cannot yet be profitably said. Nevertheless, I assure you that what is still to be offered in analyzing "cultural entities" and "histories" is meant to be the capstone of this primer. Its mention may suggest at least the point of the caution, in chapter 3, of distinguishing between the syntactic informality of Lebensformen and their inherent historicity. That is, I shall need to make room for conceding that there is an important sense in which things that have "careers" may not yet have "histories" (acorns and horses, say), whereas there are "things" that clearly have "histories," if anything can be said to have a history (the American Revolution, the Cubist movement, Henry VIII's reign, a person or a person's life) and that (in some as yet unspecified regard) have properties vastly different from those of the others. (Wittgenstein, you must remember, hardly touches on history.)
Let me now intrude a strange specimen for heuristic purposes. It is
possible, say, that even where we insist that existing things have "natures," they may be said to lose one "nature" and be ascribed another, without jeopardizing numerical identity! What is arresting in this specimen case is (i) that it may, without difficulty, be coherently described, and (ii) that it involves the peculiar (and widely ignored) distinction between "natural kinds" and "kinds" that bears in a distinctive way on those things that (as I say) have or are "histories," things that are not readily construed as (belonging to) "natural kinds" themselves. The essential point, of course, is this: on the best of the analytic accounts—on Strawson's, for instance—numerical identity is made to depend on the fixity of the sortals (essentially or by lebensformlich entrenchment) in terms of which things are individuated. But the case I have in mind preserves individuality (hence, numerical identity), while abandoning the fixed sortals (or "natures") on which individuality is supposed to depend .[10]
This comes as a surprise: it makes a very important conceptual option clear. Consider the story of Lucy Westenra in Bram Stoker's Dracula . Lucy Westenra was a woman (that is, a member of Homo sapiens sapiens ). She was bitten by the vampire and became a vampire (presumably another "natural" kind, albeit a most "unnatural" one). In the story, she loses one "nature" and acquires "another," but she remains one and the same person . Qua person, she either has no nature (only a career or history), or she has no nature other than her career, or the nature she has (first) and (then) loses need not deprive "her" in any logical or conceptual regard of the unicity of "her" career—hence, the unity of "her" (inconstant) nature. (I risk using the terms "unicity" and "unity" here: they are intuitively clear. But I return to provide them with a more formal sense in chapter 9.) For the time being, I ask you only to concede that these conceptions are coherent and self-consistent—and (I urge) promising as well. For, consider that you must in any case explain what the conceptual relationship is between "natural-kind" terms (like "man" or "human being") and "kind" terms like "person," both of which function as sortals (in Strawson's sense), although not necessarily in tandem. I suggest that we should have no trouble at all, in a real-world setting, in individuating and reidentifying someone like Lucy Westenra.
It is certainly possible to hold, as many do (notably Williams), that "persons = bodies" (that is, that persons are identical with individuated specimens of Homo sapiens ).[11] But that cannot be right if, as we are clearly prepared to admit, there may be persons in the universe that are not human (Martians or artificial persons, possibly chimpanzees incip-
iently, not to mention vampires or angels or gods). Certainly, when formulated abstractly (rather as with the principle of noncontradiction—on which it ultimately depends), "everything," as Bishop Butler says, "is itself and not another thing."
I am entirely prepared to concede that
(7.24) necessarily, whatever exists is identical with itself (is self-identical ).
You must realize, however, that what I have already said about "careers" ((7.9)–(7.14)) confirms the vacuity of (7.24) in a sense analogous to what I said about contradiction when uninterpreted . There is no known criterial reading of (7.24) that, on an altered interpretation of the nature and career of any particular thing, might not confirm that an apparent criterial violation could be reasonably obviated. The claim, for instance, that "Lucy Westenra" cannot possibly be the name of one and the same person, because it cannot, on the hypothesis of the story, be the name of one and the same woman (or one and the same human being ), instantly founders if one concedes (coherently enough and consistently with what has gone before) that
(7.25) attributions of numerical identity are assigned on one (ontic) interpretation of a career or another.
I do not need the vampire case to make my point: I introduce it only to draw attention to the plain difference between our treatment of natural-kind terms (a subset of sortals) and terms like "persons" (sortals, but not natural-kind terms). Nevertheless, the example does help to fix some additional terminological distinctions. Let me say that by identity (or numerical identity ) I mean (i) that (universal ) predicate (or attribute) (ii) that is "descriptively" vacuous but (iii) may be constatively uttered in first-order discourse as a second-order predicate ((6.27)); (iv) that presupposes "number" ((6.30)); (v) that, in virtue of (iv), veridically extends to everything that exists; and (vi) that, in virtue of (v), trivially entails, with respect to that same extension, the descriptively vacuous second-order predicate (or reflexive relation) of self-identity . Here, by a universal predicate (or attribute), I understand (i) a predicate ascribed to individuated things, (ii) solely in virtue of their being individuated, (iii) vacuous in the "descriptive" (or first-order) sense, and (iv) ascribed
trivially, without exception ("necessarily"), and without being "interpreted." Number, of course, is similarly a universal predicate of existent things.
By individuation I mean, paradigmatically, (i) a discursive (or constative) act, (ii) that, in "parsing" the world, (iii) first ascribes "number" to existing things (that is, in the idiom of natura naturans, parses things at once as existing and as having number), (iv) thereby (trivially) preparing them for subsequent ascriptions (natura naturata ) of "numerical identity" and "self-identity," (v) by way of instantiative rather than mere predicative criteria. By instantiative, I mean, paradigmatically, (i) criteria in virtue of which predications are instantiated de re or de dicto, (ii) in persisting careers or things that have careers. The essential point is this: predication presupposes individuated things; instantiation entails individuation itself. In this sense, Frege's famous puzzle regarding the Morning Star and the Evening Star is as much about individuation as about numerical identity.[12] The point is nearly lost. Theoretical identities, whether "same level" (Morning Star/Evening Star) or not (bolt of lightning/suitably ionized patch of atmosphere), are first confirmed in accord with (7.25) and then applied in accord with (7.24). Of course, the paradoxes of Bell's theorems (for classical mechanics) lead us to speculate about what quanta might be and how they might be related to the things of the classical-mechanical world.[13] But that is a question, first, of individuation—not, as is usually supposed, of numerical identity. The paradoxes call for a replacement, at least at certain points, of our ontology, not a resolution of local infelicities relative to an ontology in place.
I should perhaps add that by entails I mean only "deducible from," strictly or logically, as in saying that "this figure is rectangular" is deducible from "this figure is square." By presupposes (a much more complicated notion), I mean minimally that logically relational feature of a proposition that (i) is not actually uttered and not entailed by what is uttered, but (ii) the truth of which logically bears on, or affects, the truth of what is, in context, uttered or entailed and (iii) would, on a pertinent theory or interpretation, be affirmed, conformably with (ii), by whomever competently uttered the speech act in question, or would be recognized (by such a competent agent) as falling within the scope of (ii); or, alternatively, that feature of a supposed state of affairs such that (iv) some proposition or other falling within the scope of (ii)–(iii) would be elicited from a competent speaker conformably with (ii)–(iii).
The exemplary case, of course, is due to Strawson's analysis of Russell's specimen sentence, "The present king of France is bald." On Straw-
son's analysis, that there exists a present king of France is neither asserted nor entailed by what is asserted in uttering Russell's sentence, but is only presupposed in considering straightway whether what is uttered is true or false. On one reading of Strawson's case, the truth or falsity of what is uttered presupposes the existence of the present King of France; on another, that "The present King of France is bald" is rightly taken to be true or false presupposes the king's existence, for, otherwise, what is uttered cannot rightly be assigned either value. Failure of reference here, it is said, produces truth-value gaps or requires the admission of a third truth-value, viz. "indeterminate." Clearly, the matter of presupposition cannot be restricted to the puzzles of referentiality. (It is also, as I shall later say, "Intentional.")
I need to pause here to provide as well some further distinctions—in effect, interpretations of what may be meant by the expression "one." This should gather together some remarks scattered through the early part of this chapter (as well as allusions to earlier chapters) that may strengthen the sense of the order of this entire inquiry. For, of course, to speak of numerical identity is to speak of what is "one and the same"; to speak of individuation is to speak of "this one" and "that one" among a set of distinct "ones." A career, I should say, is "one and the same this one " tracked through time and change, whereas a (mere) individual thing is "one and the same of this kind, " regardless of the "instantiative" episodes that compose its career (and regardless of whether it is a "thing" or an "event," in the disjunctive sense supplied a short while ago). Determining numerical identity, I say, presupposes individuation, whereas deciding individuation entails determining numerical identity.
I have alluded to other senses of "one," some of which are predicative and some, at least not merely, referential. The matter is quite strenuous: I offer the following only as a compendium of convenience; I hardly dare claim that it includes all interesting senses.
For what it's worth, then, by one, we sometimes mean: (i) numbered, hence denumerable as one or another of a set of countable things, in accord (therefore) with individuation; (ii) self-identical, hence one and the same (with itself), in accord with the notion of numerical identity; (iii) lacking number butinclusive in a discursive sense, hence said of "the (one) world" or "the (one) universe of discourse" (in Husserl's sense), which includes (or cannot be made to exclude) whatever, when mentioned, is mentioned; (iv) unconditionally self-subsistent (as in Aquinas's account of God), hence without assignable or "privitive" nature, though still "subject" to rational discourse; (v) not subject to any predication
at all, hence ineffable, undifferentiated, said of the ontic source of whatever otherwise exists or is differentially real (Heidegger's Sein perhaps, the Buddhist sunyatta, the Chinese Tao ), normally marked by negating what may be truly predicated ("oppositionally") of what exists or is real; (vi) necessarily and unconditionally real, hence unable (in the same sense) to support any further attribution, thus, the One,not subject to partition or differentiation of any sort (as in Parmenides's "One"); and (vii) common, "qualitatively the same," "similar," said of predicables (without prejudice to their ontic analysis, as in Quine's notion of "divided reference").
Bringing these distinctions together facilitates further findings. For example, item (v) shows at a stroke Heidegger's conceptually very strange invention (after the Kehre and evidenced somewhat before the Kehre ),[14] that construes Sein as determining the fateful projects of imminent history, despite the fact that Sein has no intelligible structure by which to do so. I have already remarked on the cognitive privilege buried in Aquinas's conception of the Creator ((iv)); for surely, as with Aristotle's metaphysics, Aquinas's schema cannot claim to be a necessary one and will not permit itself to be viewed as a mere conjecture. But there are more instructive payoffs still.
The intuition that things have "natures"—must have "natures"—is probably linked to the idea that their changes must be explained in terms of those same "natures." That is Aristotle's thesis (and, I think, Strawson's as well—although Strawson's is thinner). Aristotle believes, of course, that any departure from the necessary invariance of the "natures" of real things leads, somewhere, to incoherence or self-contradiction. The "natures" he assigns particular things (ousiai ) are precisely what he needs to ensure the explanatory powers of his science: everything that changes (well, nearly everything), Aristotle thinks, changes in ways that can be explained by reference to a thing's "nature." Modern theorists similarly insist on the necessary invariance of the universal laws of nature.[15]
Instructively, Popper argues that we are wrong to treat our nomological conjectures as more than "verisimilitudinous": wrong to suppose that our nomological guesses are ever fully correct in an invariant and realist sense; on his own option, however, Popper insists on the ultimate invariance of the laws of nature. He needs the doctrine to sustain "verisimilitude." Hence, although he excoriates Aristotle for his "essentialism," he is lenient enough to admit that his own doctrine is a form of "modified essentialism." What this means is that exceptionless laws
(however captured—whether inductively or by other means) are thought to range over natural kinds (in fact, to define them as favorably related to the explanatory powers of the laws).
My present purpose is to suggest the conceptual link between archism and modern theories of science—among the Positivists, among those committed to the unity of science program, and among the Popperians. It is true that, more recently, the explanatory laws of nature have been denied a purely realist standing, causality and nomologicality have been disjoined, and both phenomenological and explanatory laws have been treated as constructivist (as no more than artifacts of symbiosis or idealizing interpretation). Yield in that direction, and you cannot discount the an-archic reading of "natures."
I press the point to locate conceptual connections that are unexpected and often ignored and to draw attention to the tolerance of variable "natures"—in opposition to both Aristotle's model and the model of the natural sciences that has dominated Western thought through more than the first half of the twentieth century. But my argument regarding the conditions of individuation and identity does not depend on those vagaries. I mean of course to cast doubt on an archic reading of the natural sciences. On the archic reading (as in the unity of science conception), granting that reference may be replaced predicatively, the descriptive and explanatory language of a bona fide science is, in principle, homonomic, that is, the same —under real covering laws, when the world is successfully "reduced," because, ultimately, what is real is (it is claimed) what falls under exceptionless, invariant, nomological universals, laws of nature —necessary de re, hence such as to range over (fundamental) natural kinds . (Archism, of course, is, ineluctably, externalist.)
I take it to be symptomatic, for example, that theorists as diverse as Hempel, Popper, Quine, Davidson, Strawson, Putnam, and Kuhn never address the logical possibilities raised by theorems like (7.6), even though they generally oppose the usual forms of essentialism. I suggest, in fact, that the fundamental logical difference between deciding individuation and determining identity has not been rightly perceived because (among other things) the archic thesis (in some form or other—witness Strawson's account of sortals) has been thought to serve individuation. But that is clearly false, both because of (2.1) and because, for instance, although "person" is a sortal term, it is not a natural-kind term .
I put it to you that
(7.26) there is no evidence that the conceptual resources of anarchism are not adequate for any first-order inquiry, or are less adequate than those of any archic alternative.
That is, not only is (2.1) true, but scientific and philosophical strategies in accord with it compare favorably with any archic alternative. I say that a stalemate here is tantamount to victory. As a consequence,
(7.27) the "natures" assigned individuated things—even "natural kinds"—are no more than salient regularities, sufficiently well demarcated for local classificatory and recognitional purposes and/or the purposes of scientific explanation and prediction.
It's clear that (7.27) is simply an elaboration of (7.6); also,
(7.28) attributions of lawlike behavior are essentially a predicative matter—
subject, therefore, to the vagaries collected under (3.13)–(3.15), which affirm the inherent informality of predication itself. It may be remarked again that, currently, among philosophers of science, there is a strong inclination to cast doubt on claims that there have been discovered, or that there must be, nomological invariances in nature or that they must be realist in some unsymbiotized sense.
The truth is, once (2.1) is conceded and the vagaries of reference and predication made clear, insistence on the "natures" of things, on things having "natures"—a fortiori, on things having fixed natures (whether in essentialist terms or in terms of real nomological invariance)—is a matter of the most profound indifference . Notice that the question of the logic of reference and predication does not depend in the least on a choice of archic or an-archic options. On the contrary, the independent treatment of the question distinctly tips the scales (empirically) in favor of the an-archic alternative. That has been the leit-motif of my entire strategy.
This leads us to some version of the following (which makes explicit that which may be plain enough):
(7.29) "natures" are no more than provisional, salient, revisable, capable of being represented by jointly nonconverging sets of predicates relative to which, within our Lebensformen, we individuate, identify,
and reidentify the existing things we share, in sharing a contingent culture.
Theorem (7.29) confirms (i) that "natures" are relativized to the different Lebensformen (or the temporal phases of the Lebensformen ) of different societies, (ii) that they are adequate as far as "reality" is concerned if they service, coherently and adequately, according to our lights, our referential and predicative needs, our needs regarding identity and individuation, (iii) that there is nothing anomalous about our using plural, even opposed, ways of designating the "natures" of things, and (iv) that, since reality itself need not be invariant, we may be obliged to regard each conceptual strategy as vindicated only "for the most part" (as Aristotle himself sometimes affirms). (Leibniz, for instance, does not require "nominal essences" to be mutually compatible.) But if you grant that much, you see at once that
(7.30) where we individuate referents only or primarily as careers, we may also (if we wish) construe "natures" as no more than a façon de parler, without risk of incoherence;
that is, as no more than instrumental for referential and related purposes, or as heuristic in terms of lebensformlich compliance. That is simply to conform with the finding—within the terms of the flux—that individuation is not primarily concerned with "natures" but with "careers" (7.9).
Bear in mind that, as in debates regarding the systematic advantages of competing theories of the unobserved genetic (inherited) transmission of traits, we may, by a principle of charity (but not otherwise), claim to have kept the theoretical referents of such debate constant, even though the theories in which they are completely embedded change in the very process of debate;[16] also, that the evolving theories of the physical "elements" (gold, for instance) regularly change the assignment of the "natures" of those same elements, without supposing that their identity (the numerical identity of particular pieces of gold) is put at irremediable risk; also, that, as Kuhn remarks (troubled though he is by having to do so), Priestley and Lavoisier "lived in different worlds." But to say all this is to confirm (7.6) and (7.14). (These considerations raise the "incommensurability" issue, but this is not the place to make sense of it. It remains unresolved in Kuhn.)
Once we have this much in place, the triviality of (7.13), in regard to
the alleged need of the "careers" of things to accommodate the "natures" of those same things, dawns on us:
(7.31) to specify the "careers" of existing things is no more than to join together the serial instantiations of distinct predicables so as to form (through time and change) unitary (individuated) careers that signify the numerical identity of the particulars they are .
To persist, therefore, is to exist or be real over time, to possess a career. Hence, from (7.31), it follows that
(7.32) the numerical identity of "persisting" things is entailed in their individuated careers, but the structure of a thing's career is only presupposed by affirming its numerical identity.
Furthermore, if symbiosis is conceded ((4.11)), then it is reasonable to claim, in the light of (7.31) and related theorems, that
(7.33) everything that exists and is real is socially constructed.
It remains to say a last word about discerning the careers of things. Certainly, for one thing,
(7.34) natures and careers are specified only within the context of our Lebensformen .
This signifies that individuation and identity function under the same constraints as reference and predication: everything thus far presented in chapter 7—for instance, (7.27) and (7.29)—bears this out. The argument also shows why Leibniz's law does not help matters. The difficult question is how to determine when you have a single "career" before you. It is identifying "this one," rather than reidentifying "the same one," that is the decisive matter. The reason is obvious: as Hume already realized, it is the singular thread of the instantiation of plural predicables that constitutes a single career, not the indiscernibility (in different contexts) of the different instantiative sites of given predicables . Leibniz's law never addresses what to count as individuals, in any epistemically operative way, it merely presupposes that that has already been decided. (Leibniz, of course, permanently baffles its solution.) The resolution of the puzzle (with respect to reference and individuation) is the analogue of the resolution (with respect to predication) of the puzzle of "real
generals." (Discrepancies, remember, can always be deflected and assigned, in principle, to the different "phases" of any one—or any many—s.) I take it to be important to affirm, therefore, that
(7.35) in general, among things that change, the conditions or criteria of individuation cannot be derived from their conditions or criteria of identity.
I favor Hume here, against Leibniz and Frege (and, implicitly, Quine). If I am not mistaken, the philosophical force of (7.35) is nearly neglected among the principal analytic philosophers.
Hence, by way of summary, I now say:
(7.36) the realism of "natures" and "careers" is affirmed only in accord with the consensual practices of our Lebensformen .
Theorems (7.34)–(7.36) I take to be tantamount to:
(7.37) "natures" and "careers" are "social constructions."
This, of course, is already entailed in (7.33). But to put matters this way is to begin to explain the meaning of the doctrine of social constructivism (or, constructionism ), namely: (i) that what, under symbiosis, is constructed implicates "brute" existence, although without evidentiary standing; (ii) that what is real is constructed under symbiosis; and (iii) that what obtains under (i) and (ii) is lebensformlich . Conditions (i)–(iii) signify that what is real qua construction may nevertheless exhibit Secondness (in Peirce's sense). This means that, although it is true that, conventionally, "idealism" signifies that what is real is in some sense humanly "constructed," idealism (in the sense here featured) is always, also, symbiotized. The analytic opponent of idealism is ordinarily an externalist, whereas, on my account, constructivists (like Kant) are not far from being externalists as well.[17] The analysts do not admit the more strenuous possibility; they do not see that realism and idealism cannot, on pain of cognitive privilege, be disjoined in principle.
Alternatively put:
(7.38) the real world = the intelligible world.
where, by intelligible, I mean no more than discursible, accessible, under symbiosis, to constative discourse, in referential and predicative terms.
Of course, I do not mean by this that there is no part of the real world that is unknown, or that what is known to exist in the real world does not exist "apart" from its being intelligible. Nevertheless,
(7.39) the independence of the world is an intelligible feature of the world.
(By independence, as already remarked in chapter 2, I mean, "determinately structured, apart from human inquiry and understanding," said of what exists or is real.) Hence,
(7.40) that the real world is independent is compatible with its being intelligible and with its being socially constructed ((7.33)–(7.37)).
For, of course, the independence of the world signifies that Secondness rightly applies to what may be found in the world, but what may be found in the world is objectively found there under the condition of symbiosis ((4.11)–(4.13))—hence, in the constructivist sense. There is no insuperable paradox there: symbiosis never functions criterially, only holistically, and both skepticism and cognitive privilege about the independent structure of the world are obviated by admitting the constructivist constraint. Hence,
(7.41) that the world is independent = that the world is legitimatively posited as independent.
Chapter VIII
Legitimation
The convergent benefit of the arguments of the last few chapters lies in bringing the resources of constative discourse—reference, predication, individuation, identity, reidentification, ascription of truth and falsity and of existence and reality—into line with the general thesis: (i) that none of these discursive competences functions (or need function) algorithmically or in any way that is governed by universal syntactic rules, or in any other archic way; (ii) that the rigor with which any of these is exercised depends on its being conjointly "entrenched" in our Lebensformen; and (iii) that all are legitimated, without privilege, under the conditions of symbiosis.
This is a remarkably spare finding. If we suppose that everything that exists or is real obtains in "one" universe (or, "one" universe of discourse )[1] —in the sense in which Husserl correctly observes that it makes no sense to apply "singular" and "plural" to the inclusive context of whatever we affirm or deny or to whatever we suppose language refers—then, on the foregoing arguments:
(8.1) the conceptual boundaries of the universe, "the universe of discourse"—the context of all contexts —cannot be determinately specified.
Those "boundaries" need not be specified, of course. What is important is that it makes no sense to suppose that we can "violate" the limits of our "universe." There are no limits or boundaries to be reckoned with.
To speak of the inclusive "universe" is merely to register the fact that we can bring anything we wish to say to bear on anything else we wish to say. The universe is "one" in that sense only. "It" sets no constraints on relevance or coherence or consistency or commensurability or even success with respect to whatever, distributively, is individuated, identified, or otherwise specified.
I conclude:
(8.2) the universe is trivially implicated in any and every discourse;
and
(8.3) constative discourse is contexted,
that is, "implicates" some more inclusive (however inchoate) context of discourse—ultimately, the universe—within which all its truth-claims obtain.[2] From this it follows:
(8.4) the universe of discourse cannot be "totalized";
and, emphasizing (8.3):
(8.5) the universe of discourse cannot be a proper referent in constative discourse; alternatively put, "universe of discourse" is no more than a nominalization ranging over all contexted reference.
To say, therefore, that the universe of discourse is implicated in all discourse is to express, by a figurative use of the constative idiom, that which, more in accord with the strategies of negative theology, is expressed by (8.1). That is,
(8.6) constative discourse implicates some further nonconstative function of discourse.
I acknowledge that I have extended here the use of "implicates" beyond what I explored in chapter 4, for instance at (4.10). There I used the term to signify certain conceptual relations among the symbiotized parts of constative discourse alone; here "implicates" is made to link the constative and (what I shall soon characterize as) the "mythic" use of language. But the extension is straightforward and no paradox need result. In both cases, the implication links first- and second-order considera-
tions and confirms the advantage of allowing de re /de dicto distinctions.
I mean, by totalized: (i) that supposed feature of any conceptual scheme in virtue of which the inclusive set of all possible predicates apt for constative discourse may in principle be specified or otherwise systematically delimited; or (ii) that feature in virtue of which we can be assured that the concepts, categories, rules of discourse, or the like apt for some sector of the universe are not subject to change or replacement as a result of the apt analysis of any other sector of the universe; or (iii) that feature in virtue of which the boundaries of the universe are known to yield criteria for the admissibility of all substantive concepts apt for constative discourse in any sector of the universe. (By apt I mean adequate for formulating all the principal truths about any sector of the universe required for a valid explanation of its phenomena.) The extravagance is clear, and the philosophical corrective works best by exposing the limitations of would-be candidate theories. Hence, the would-be modal necessities that Aristotle speaks of (in Metaphysics Gamma) and that Wittgenstein speaks of (in Tractatus ) are, implicitly, paradigms of "totalizing." So too is Carnap's early systematic physicalism and extensionalism.[3]
This is, also, the same charge that I have brought against Dummett's prioritizing semantics over metaphysics. The Francophone structuralists (Lévi-Strauss, preeminently) suppose—more grandly—that, in effect, (8.4) is false. It is, of course, Derrida's principal objective—in his notorious deconstructions —to insinuate, or show, just how, for any would-be totalized scheme (or "system"), some unanticipatable conceptual supplément is always possible, can always be counted on to subvert the claim. This single theme—"radical alterity" (in Derrida's phrase), the excluded but not yet specified différend (Lyotard's neologism), the inchoate l'autre or l'autrui[4] (of Levinas and French feminism), and the like—is, I judge, the essential theme of contemporary poststructuralism, which opposes "totalizing" in every form and assimilates all instances of cognitive privilege to that philosophical "offense." L'autre (or l'autrui ) is what, admitting (8.4), is always omitted, neglected, marginalized, disadvantaged, or the like. Poststructuralism monitors the fact and its implications. Implicitly, every archic thesis is a form of totalizing, since, on the argument, no conceptual discovery can dislodge what it discloses.
There is, actually, an entire family of philosophical strategies by which, implicitly or explicitly, totalizing claims are serviced. They are not all of the same gauge or presumptive power. The archic thesis itself, claims of modal necessity and invariance imposed de re or de dicto, are,
ultimately, forms of totalizing. So are all the forms of apriorism, prioritizing our cognitive resources, drawing noumenal truths from phenomenal data, specific claims in favor of the "originary" and the "apodictic," the rejection of symbiosis, presumptions of transparency, the denial of interpretive tertia, the appeal to first principles, insistence on the neutrality of particular philosophical options, reference to the exceptionless laws of thought or the laws of reality, indifference to the relevance of context and history. I gather them all now as obvious forms of totalizing. That is,
(8.7) totalizing signifies that the universe is, ontically, a conceptually closed system, the parts of which are, in some respect or other, not contexted.[5]
It need not be closed in any temporal or epistemic or causal sense (Peirce). It may be said to involve relata only (Lévi-Strauss) or discrete elements (the unity of science program). It may be "rationalist" (Fodor's conceptual nativism) or "empiricist" (inductivism in the sciences). It may be explicit (Aristotle's archism) or relatively inexplicit (Davidson's various doctrines: the universalizing of Tarski's semantic conception of truth, the "supervenience" thesis). It makes no difference. All versions hold that, at some decisive point at least, a totalizing constraint is known to obtain. In that sense, the "universe of discourse" is said or thought to be such a "system."
I claim therefore:
(8.8) within the ("one") universe, constative discourse cannot overcome the conceptual inaccessibility of the totalized.
It was in this sense that I affirmed theorems (4.14)–(4.17), that is, that the world is a "text"—hence, always and everywhere "interpreted" in accord with our tacit conceptual schemes. We see at once, therefore, the global importance of bringing the informality of our Lebensformen (which affects the logic of reference and predication) into accord with symbiosis. It affects in the most profound way the logic of legitimation. Hence, to acknowledge that discourse is contexted ((8.3)) is, effectively, to admit that
(8.9) no natural-language domain, no world, can form a closed system.
It may, of course, always be arbitrarily closed.
I find it helpful to speak of a world when I mean to speak of some relatively delimited but inclusive conceptual "space" comprised of a set of denumerable "things" said to exist or to be real (or, by analogy, some "compossible" world). (By space, or a space, I mean a "world" apt for constative discourse, a world "parsed" for reference and predication, for individuation and identity, and the like.) By contrast, I speak of the universe when I mean to speak of the inclusive "space" of all (individuatable) "worlds," regardless of compatibility or commensurability. Of course, the "universe" is never interpretively totalized and never individuated. If the distinction be allowed, then I say:
(8.10) "worlds" are contexted within the universe;
and
(8.11) the universe, construed as "the context of all contexts," cannot, on pain of contradiction, be constatively addressed.
For this reason, reclaiming (8.6), I say:
(8.12) discourse about the universe is mythic rather than constative.[6]
I don't deny that we "talk about" the universe in some sense. We even say things that we judge to "fit the universe," felicitously—for instance, I am bound to say, and willingly do say:
(8.13) truth-claims are contexted within "the" universe.
But the fit itself (if we allow the expression) is "mythic" only. We cannot affirm such a fit constatively. (In "constituting" a determinate universe, we constitute no more than a world ((8.10)).)
A myth, in the sense I have in mind, is a (numbered) conceptual "picture" intended as a representation of the universe of discourse, constructed to accord with this (denumerable) "world" or that. Mythic, as a term of art, designates: (i) a kind of dependent speech act, (ii) not itself constative but one that makes a logically degenerate use of the constative, (iii) obtaining in referential acts and all speech acts insofar as they entail or imply reference, (iv) by which we are said to "speak about" the universe. But discourse about the universe is logically inert, in the sense that the universe is implicated, without relevant differentiation, in
every constative act. We may assess (in the "mythic" sense) the relative fit of competing representations (of the numberless universe) only in terms of the would-be truths about the worlds they "subtend." This may also explain my attraction to Eastern notions like the Tao and sunyatta . If I understand them rightly, they postulate de dicto an inherent limitation of constative discourse relative to what, de re, is implicated in constative discourse itself—under the condition of symbiosis. On that reading, all the annoying paradoxes of such idioms are meant as reminders of the possible confusion of the mythic and the constative. But so long as the modal presumption about referential and predicative discourse is avoided, nothing is lost with respect to the precision or validity of discourse itself. On that reading, it is false to suppose that the world is delusive (which is self-defeating in any case). If either Buddhism or Taoism means to treat the world as no more than delusive, then I must admit I can make no sense of them.
It is possible, therefore, to judge which myths, relative to which given "worlds," are apt or relatively more apt for generating important truths; conversely, it is possible to judge which "interpreted" worlds best fit this or that (mythic) picture of the universe. Clearly, myths are denumerable representations of the universe, but they are thus intended (stipulated). By denumerable, I simply mean "countable" by isomorphic matching with the natural numbers. Myths, then, are denumerable, but what is "mythically predicated" is "predicated" of what is not denumerable.
As far as truth is concerned, it goes without saying that such discourse conforms with (8.12). In assessing "fit" (not "correspondence"), the body of truths that any world subtends is already assumed to be in place: granting (8.11), it could hardly be otherwise. (I offer this, also, in the spirit of interpreting Wittgenstein's final puzzle, in the Tractatus, regarding the "mystical." Nevertheless, I oppose Wittgenstein's presumption that it makes sense to introduce a demarcation between what is "within" the world and what is not.)
Beyond this, we are led to admit:
(8.14) philosophy concerns the second-order legitimation of conceptual schemes "linking" worlds "and" universe—
in both directions: facilitating the discovery of important truths about acknowledged "worlds," maximizing the coherence and scope of any "one" such world relative to others. Theorem (8.14) leads to the very interesting conclusion that
(8.15) philosophy cannot be confined to the constative;
on the strength of (8.14), then,
(8.16) philosophy functions "mythically"—in functioning constatively.
Hence, also, recalling (1.5):
(8.17) first-order constative inquiries "implicate" the mythic function of philosophy.
I take (8.17) to be tantamount to (8.4).
It was in this sense, I believe, that Kuhn remarked that Priestley and Lavoisier lived in "different worlds." (Trivially, Kuhn meant that their "worlds" obtained in the [same] "universe.") Recall, however, theorem (8.1) and what follows from it. It will then be seen that
(8.18) no "world" can meaningfully be said to approximate the "universe."
Still, on a given theory or criterion,
(8.19) "worlds" may be said to be more or less inclusive than other worlds .
Theorem (8.19) is the normal presupposition of the goals of scientific "reason." It also introduces a bona fide legitimative question. But (8.19) must also be duly reconciled with (8.4), (8.8), (8.18), and the like. In that sense, legitimation "implicates" mythic considerations. It is in fact (8.18) that, in their different ways, Peirce and Popper and Habermas (and Hegel, on a doubtful reading) violate. Progressivism, as we may term the denial of (8.18) or any analogue of it, is, therefore, another form of totalizing. (But that is not true of first-order appraisals of progress, which might be said to implicate competing "Kuhnian" worlds.)
I suggest, therefore:
(8.20) legitimation is the exemplar of philosophical reason, in spite of the fact that reason and its norms are artifacts constructed within "constructed" worlds.
Theorem (8.20) poses (and resolves) the essential puzzle shared by Wittgenstein (in Investigations ) and Foucault (in The Order of Things ), except that Foucault (but not Wittgenstein) historicizes that puzzle radically. On my own reading, Foucault has isolated the single most important enigma of the next century's philosophy and science. The pity is, he did not stay to analyze it with sufficient care. Reason, then, in any philosophically pertinent sense, is: (i) a second-order competence, (ii) not merely normative (although it is normative), but (iii) normative in virtue of legitimated interpretations of our symbiotized world, hence (iv) compatible with an-archism and (v) itself an artifact of a constructed world; also, (vi) subject, reflexively, to historical reinterpretation. If this gives a fair reading of "reason," then (in the context of (8.16)–(8.17)):
(8.21) philosophy, science, legitimation, reason itself, cannot be construed in the naturalizing way.[7]
That is, they cannot be construed in any way reducible to, or replaceable by, or functionally equivalent to, any (first-order) psychological competence itself construed as free of philosophical (second-order or legitimative) "contamination." In short,
(8.22) naturalizing is, at least implicitly, a form of totalizing.
It is quite true, as Quine insists, that epistemic reflections cannot fail to involve our (first-order) psychological powers; it is also true, as Millikan (among others) has shown, that operative norms may be construed in the "naturalizing" way—for instance, in theorizing about the "functions" of species-specific organic life. But no one has shown that legitimative, second-order norms are or can be convincingly construed in naturalizing terms . I take theorems (8.21)–(8.22) to be extraordinarily important, therefore, in exposing a decisive lacuna among the "naturalizing" philosophies of our day (those particularly that have followed the lead of Quine's well-known paper, "Epistemology Naturalized"). I see no way in which they could resolve the difficulty posed by (8.21) without falling back to one form or another of archism or cognitive privilege or totalizing. For,
(8.23) cognitive privilege entails, by its effective exercise, the principled irrelevance of the context or contextedness of constative discourse, which is tantamount to totalizing.
In this sense, admitting (8.21)–(8.23) avoids the anti-psychologizing tendencies of the Fregeans and yields to the contingent psychologizing tendencies of the Quineans, but, in doing that, the admission does not presuppose the necessity of psychologizing epistemology or the validity (if psychologizing is avoided) of a privileged account of epistemology. As far as I am aware, there are no analytic strategies of (naturalized) epistemology that cannot be shown to be inadequately defended—on the clue collected at (8.21)–(8.23).
I had introduced in chapter 5, I may say, the "naturalizing" option Quine has brought to such high prominence. But I linked it there, for dialectical reasons, with Husserl's opposed and derogatory use of the epithet "naturalistic." I trust, therefore, that it will be useful to say now that, by naturalizing, I mean any philosophical policy or strategy that: (i) denies the ineliminability of legitimative (or normative) reason, along the lines just sketched; or (ii) claims to be able to analyze (reduce ) such reason and its operative norms to first-order psychological or causal or extensional terms; or (iii) precludes the need for legitimative reason (for instance, by defeating the doctrine of symbiosis—a fortiori, the need to admit second-order interpretations); or (iv) repudiates the epistemic relevance of countextedness, except in local terms keyed to ignorance—said to be surmountable in principle; or (v) pretends that a given natural-language inquiry ranges over a "closed system" or supersedes the "mythic" limits of constative discourse. (A closed system, therefore, is a world masquerading as the universe. The idea is not far from Wittgenstein's notion of the "mystical.")
I claim only that there is no compelling argument of the naturalizing kind, nor any that is promising about how to establish its case, nor any that satisfies the objections I have posed. I don't claim that the naturalizing option is incoherent or contradictory; I do say that stalemate here is as good as victory, for stalemate yields a sense of the barriers that must be and are not yet surmounted. In this sense, frankly, this primer is committed to the conviction that the "naturalizing" strategy cannot succeed.
I summarize all this as follows:
(8.24) the contextedness of truth-claims = the exclusion of cognitive privilege = the inseparability of first- and second-order discourse = the regulative function of legitimation = the symbiosis of world and language = the impossibility of totalizing = the implicated link between the constative and the mythic.
Count theorem (8.24) a rational "bet," not a self-defeating claim that insists that it is itself necessarily true. That, I say, is the principal benefit of the doctrine of the flux.
We have now reached something of a plateau. For, taken alone, (8.18) does not —cannot —disallow the presumption of de dicto or de re necessities; or the prioritizing of alethic, epistemic, or ontic considerations vis-à-vis one other; or foundationalist conjectures; or a priori confidence about the ultimate universal adequacy of favored philosophical strategies. No, it admits, initially, all coherent and promising strategies. Still, theorem (8.17) sets a condition on them all: it "relativizes" their supposed privilege by "entrenching" them in the Lebensformen of particular societies. It does not reject them outright: it diminishes their would-be privilege by acknowledging others that advance incompatible but comparable claims. That's all! It makes a historical artifact of modal necessity .[8]
I take this policy to be the gist of Foucault's extraordinarily perceptive (even good-humored) account of the "historical a priori " (the lesson of his "archaeology"), which, in a sense, is also a deepened version of the lesson of Wittgenstein's Lebensformen . For what Foucault suggests (and clearly believes) is that once we recognize that our science, our politics, our philosophy are entrenched in our society's epistemes (or, according to Wittgenstein, our Lebensformen, although now historicized), we cannot avoid admitting some version of the following:
(8.25) thinking is "prejudiced," tacitly constituted, impenetrably preformed, endogenously limited, "interested," perspectived, artifactually "rational," plastic, variable, culturally "constructed," alterable merely by being exercised, at once cognitively restrictive and enabling;
and, as a consequence of thinking's being thus formed,
(8.26) conceptual schemes appear from time to time to yield, as artifacts of our enabling epistemes, de re and de dicto necessities, apodictic and related totalizing powers.
Theorems (8.25) and (8.26) are meant to be blunderbuss sketches of what I make of Foucault's notion of the "historical a priori ": conceptual necessities (i) relatively successful in organizing inquiry, (ii) reasonably convincing as a consequence of the formative power of our epistemes,
(iii) alterable and replaceable under the force of changing history; but (iv) endogenously persistent. The theorems mentioned, therefore, are meant as stepping stones to what I (now) take to be Foucault's and Wittgenstein's principal discovery, namely:
(8.27) the "salience" of apparent cognitive privilege, modal necessity, and other totalizing powers is itself an artifact of our epistemes or Lebensformen .
Foucault perceptively observes that our sense of privilege and necessity is not so much mistaken as misinterpreted, for it is the touchstone of our own age regarding what to feature as the exemplars of "rational" argument. Hence, I take (8.25) and (8.26) to define the sense of Foucault's archaeological method (notably, in The Order of Things ).
The very idea of differentially weighting philosophical and scientific arguments—as open to rational ordering —betrays an (ineluctable) sense of epistemic salience, which, if it is not genuinely privileged, will seem completely arbitrary. By epistemic salience, then, I understand: (i) a first-order conviction about what is "rational" in the way of second-order legitimation, (ii) engendered by our Lebensformen, (iii) open to "rational" revision under the same constraint, and (iv) applied to what was salient in the way of legitimation in earlier phases of our continuing inquiries. (These distinctions lead, ultimately, to the topics of chapter 11.) Progressivism, we may now say, is the systematic (second-order) presumption that some sustained first-order inquiry, work, or commitment can, over time, be shown to approximate some valid invariant (archic or privileged) second-order norm.[9] Peirce, Popper, Hempel, Kuhn, and Habermas are all progressivists. So is Putnam, clinging to the "regulative" function of truth. Progress, by contrast, is a relatively modest first-order assessment (or the corresponding feature of actual inquiry thus assessed) of an approximation to any contingently posited norm. We may fairly claim to make progress in designing automobiles for fuel efficiency, but what does it mean to claim that we are making progress in designing a rational society? Assessments of progress are inevitably contexted; progressivist assessments are ultimately context-free, totalized.
Foucault resolves the puzzle posed by (8.27) by insisting on the symbiotized link between (i) the formation, in us ("upon our bodies"), of certain "epistemic saliencies" and (ii) our judging (representing to ourselves), by virtue of that enabling power, what is best in the epistemic
way. That is what I take Foucault to mean by the "historical a priori " produced (as he says) by that curious "empirico-transcendental doublet," the human self (ourselves—the selves of the modern world) thus constituted by its own formative history. But, we may well ask: From where does its legitimation come? The stunning point is surely this:
(8.28) the regulative function of legitimative reason entails its functioning reflexively as well.
Hence,
(8.29) the denial of (8.28) is tantamount to a claim of privilege.
I read Foucault's argument as a breakthrough regarding the puzzle of (8.28). If I understand it rightly, it says in effect (and if so, I agree) that
(8.30) the exemplary standing of legitimation is itself artifactually projected from its native epistemes, which set a tacit limit on what it thus projects.
There is no way to understand (8.30) except to say that
(8.31) whether first- or second-order, thinking is historically formed, historicized, and historically limited and empowered.
As I say, I accept something close to (8.31) as the postulate of this primer. Now, however, it begins to surface as a theorem. What is essential to my argument is this:
(8.32) no legitimative rationale can be disjoined from a companion theory of the causal and cultural formation of the self and its enabling epistemic powers.
Hence,
(8.33) there is no way to escape the dilemma of favoring privilege or favoring sheer arbitrariness (on legitimative matters) except by construing the regulative function of reason as itself projected, at some present time t, from whatever, at some past time t-k, will have served a similar regulative function.
In short,
(8.34) the legitimation of our legitimative practices cannot fail to be privileged or arbitrary if it does not acknowledge that it is itself endogenously dependent on saliencies that cannot be privileged.
Postmodernism, therefore, was quite right to voice its suspicions of legitimation, but entirely wrong to suppose that there could be no third way between privilege and arbitrariness (or between privilege and a non-legitimatable loyalty to local practices). (I take (8.30)–(8.34) to fix the essential theme of Foucault's genealogical interpretation of the "archaeological.")
I draw the following inescapable conclusion:
(8.35) all thinking, including legitimation, is "horizonal."
I mean, by horizonal, that aspect of thinking, both collective and individual, in virtue of which its reflexive regulative function—what is meant by "reason" or "rationality," whether first-order or second-order—is: (i) symbiotized; (ii) intransparent; (iii) socially constructed; (iv) intrinsically lebensformlich; (v) historically preformed and continually transformed in merely being exercised; (vi) tacitly confined to one "world" rather than another; and (vii) contingently so perceived within the terms of its own evolving history.[10] ("Horizonal," in the sense intended, accords with that offered by Husserl and Gadamer, but now as a term of art.) Thinking is horizonal, but there are (I say) no (denumerable) horizons to which thinking belongs. Furthermore, by tacit, I mean no more than that "horizonal" aspect of our cognizing powers, which we cannot fully fathom, and which—itself made transient through our own activity—transiently affects our sense of what is rational, relevant, valid, legitimated, normatively responsible, and the like.[11]
If this be granted, then,
(8.36) legitimation cannot escape the dilemma of privilege or arbitrariness, except by historicizing its own regulative function.
In effect, in abandoning all archic, apodictic, modal, originary, and totalizing presumptions, reason cannot fail to regard itself as horizonal. I have therefore reclaimed in part the promissory note of chapter 1, theorem (1.4), to the effect that first- and second-order (legitimative)
discourse are linked in a historicized way. For, by historicized, I mean (with respect to cognition) horizonal.If thinking were not historicized, then, reflexively, either legitimation would be privileged, would come to an end in some self-evident first principle (Aristotle), or it would be forever unable to rationalize its own legitimacy (Rorty).
I take (8.36) to confirm as well the sense in which Kant illicitly recovers the "noumenal" (with regard to cognizing subjects) from the contingencies of the "phenomenal" world (despite the demurrers of the preface to the second edition of the first Critique ). Foucault may have had this maneuver of Kant's in mind; although, in "What Is Enlightenment?" he very cleverly makes Kant the "genealogist" of the "historical a priori ."
In accord with these developments, I recommend that we concede that:
(8.37) thinking is Intentional .
I draw your attention to the use of the capital "I," by which I mean (later) to distinguish between its use and that of the more familiar terms, "intentional" and "intensional." I add—definitionally:
(8.38) the Intentional = the cultural;
or, now, more completely:
(8.39) the Intentional = the cultural = the horizonal = the contexted = the constructed.
Theorem (8.39) is the orienting theme of the rest of this primer—the nerve of the an-archic strategy. It brings together all the standard questions of philosophy under this executive postulate: thinking is a history . My argument is simply this: legitimation produces insoluble paradox (the dilemma of privilege and arbitrariness) if it is not construed in historicized terms. Why so? Because legitimation is reflexive and therefore subject to a regress. Only a solution like that implied in Wittgenstein (in On Certainty ) and that of Foucault (in The Order of Things ) could possibly resolve the dilemma without abandoning legitimation itself, because such a solution entrenches reason's regulative function in its own enabling Lebensformen but not on privileged grounds . It makes a philosophical virtue out of the insuperable threat of its own regress. It en-
trenches the regress in the historicized way. Doing that, it escapes both privilege and arbitrariness. The pretty point is this:
(8.40) legitimation = the legitimation of legitimation; that is, legitimation is reflexive.
I take theorem (8.40) to follow from (8.36) and to provide an "improved" analysis of Foucault's notion of the genealogical; that is to say, legitimation under historicized conditions.
I have already shown, in chapter 3, how reference and predication are similarly lebensformlich . Since first- and second-order discourse are conceptually linked ((1.2), (1.4)–(1.5)), to have shown why reference and predication function satisfactorily without cognitive privilege is to have addressed, however obliquely, the same question that now confronts us. The solution cannot fail to be the same. The only novelty I insist on is that, now, we see, more as a theorem than as a postulate, why thinking is inherently historied.
The solution to the riddle (call it an antinomy —there will be others) rests with this: (i) the puzzle and the solution are both inherently lebensformlich and (ii) whatever is lebensformlich is horizonal. Wittgenstein addresses (i) but neglects the historical dimension of societal life; Foucault addresses both (i) and (ii) but prefers to put the lesson in the form of an antinomy (the antinomy of history, let us say). Only by "genealogizing" something like Foucault's "archaeology" (which is what Foucault himself favors in historicizing the "a priori, " the "historical a priori, " the work of the "empirico-transcendental doublet") can we bring the riddle to resolution.
Theorem (8.39) is already entailed by what has gone before. I have made it explicit in order to cast a certain light on the developing argument. What I have been trying to do (against certain fears) is demonstrate that historicizing what is best in the analytic tradition need not produce insuperable paradox. That comes as a surprise.
There is said to be a deeper (a further) issue—incommensurability . But to grasp its curious history, you must first concede that the gathering argument comes to this:
(8.41) contextedness and horizonality are the synchronic and historicized analogues of one another.
Theorems like (8.41) are suspected (for instance, by Davidson) of harboring pernicious forms of "conceptual incommensurability."[12] The
odd thing is that they serve rather to defuse such fears. For, if incommensurability appears in benign ways in referential and predicative contexts, if the pernicious forms are marginal at best, then incommensurability cannot be very troublesome.
I grant the line of argument I favor appears to make the threat of incommensurability seem particularly dire. But if (2.1) holds, then it cannot be altogether avoided. More than that, there is no single form of incommensurability: most of the familiar forms prove benign, and those that are not are dispensable.
Consider only this: if reference and predication are, as I have argued, successful in a lebensformlich way, then, if there are plural, horizonally limited "forms of life," there cannot fail to be some (minimal) forms of conceptual incommensurability; also, those cannot fail to be benign. The phenomenon appears everywhere. Foucault actually considers epistemes that begin quite locally by environing a small change in practice (say, in the prison or the court or the asylum) and then spread out (like a virus or a colonizing strike) against other local practices that make a similar effort. And Wittgenstein remarks that even if we suppose we understand an alien language in some abstract way (perhaps by deciphering a written script, while never living among the native speakers, such as the Chinese, say), we may not understand what those native speakers mean by what they say, simply because we have no understanding of their actual practice. (These themes in Foucault and Wittgenstein are not very distant from Kuhn's—or from Marx's, for that matter.) The decisive consideration is this: the denial of incommensurability, after admitting the diversity of cultures, rests with the confusion of "worlds" and "universe." For, if we admit a plurality of cultures, how can we deny that conceptual incommensurabilities will arise? Remember: incommensurabilities need not be indiscernible or unimaginable or unintelligible or even untranslatable. Admitting incommensurables is really a local affair: they fall within a common universe, often within a common world. (That is of course the burden of Kuhn's "paradigms.")
What I have said about reference and predication must be understood in this spirit. The minimal form of incommensurability probably takes this form: skilled native speakers will notice that, moving between entrenched (but distinct) idiolects, dialects, slangs and codes, professionally discrete patterns of speech, diverse theoretical paradigms local to different groups of practicing scientists, different theorizing habits in situ, different languages (accessible to apt bilinguals), there will be stretches of fluent linguistic practice regarding each of any such pairs—
both (in some fair sense) intelligible to the same speakers—which cannot be easily detached (or detached at all) from their original entrenching practices, or inserted piecemeal, smoothly, anywhere in the counterpart practices . That's all that incommensurability entails. In short,
(8.42) the detection of conceptual incommensurability is a mark of cognitive and communicative competence, not evidence of its local failure or self-defeat.
Davidson has misjudged the threat. For consider that the linear measurement of the sides of a right triangle by integers is "incommensurable" with the measurement of its hypotenuse—even though the same agent understands the fact and can reconcile both measurements, conceptually. (This is the original case.) Complicate this sort of specimen—say, along Kuhn's or Feyerabend's or Hacking's or Foucault's or Wittgenstein's lines—and you will surely hit on any number of cases of plausible and genuine incommensurability. Nothing of importance hangs on that, "except" (i) the fact that conceptual incommensurabilityis a bona fide phenomenon that (ii) takes relatively shallow and profound forms, (iii) signifying that it is unlikely that any natural-language practice forms a single legible "system" (that excludes incommensurability). Incommensurability will also obtain of course wherever (iv) symbiosis and cognitive intransparency are acknowledged, or (v) totalizing is rejected.
Finally, at the opposite end of the continuum being sketched, grant (for the sake of the argument) that incommensurability = unintelligibility (or, = untranslatability ). As far as I know, Davidson is the only major American philosopher to have construed incommensurability along these strong lines. Accordingly, he treats it as an incoherent or self-defeating "idea": there cannot then be plural "conceptual schemes," he says; there cannot be a viable form of incommensurability.
He's right, of course—on his own grounds. But the argument fails nevertheless. First of all, there are no theorists (certainly no important theorists) who champion the thesis Davidson attacks: certainly not Whorf, not Kuhn, not Feyerabend. Second, if the distinctions I've introduced between "world" and "universe" and between "constative" and "mythic" hold, then Davidson cannot sustain his claim unless there is (or could be formulated) at least one reliable "system" encompassing the "world" and capable of accommodating (without incommensurability—neutral, as Davidson clearly affirms) any new disclosures in the "universe." But we cannot possibly know anything of the sort ((8.5)).
Third and finally, if the account I have given of reference, predication, individuation, numerical identity, ascription of truth-values, interpretation, and legitimation holds, if our discursive acts are entrenched in the way I have sketched, then Davidson cannot deny that some minimal or intermediary forms of incommensurability occur—benignly.
Davidson does not actually address these matters, though his reading of Tarski's theory of truth and his opposition to interpretive tertia do help to explain his own motivation. Beyond that, it needs to be said that "translatability " (or untranslatability)—Davidson's criterion—is nowhere defined in the philosophical literature.[13] (It bears of course on the disaster that befell Positivism.) If, then, one insists on translatability (beyond moderately successful piecemeal translation ), much more than incommensurability will suddenly be at risk. I conclude, therefore, that
(8.43) incommensurability
unintelligibilityuntranslatability.
Winch's reflections on the Azande, for instance, show how natural it is to suppose that conceptual incommensurabilities will arise ubiquitiously in field anthropology. Winch was simply misread: he never opposed theorems like (8.42), but he never ignored incommensurability either. By contrast, Quine seems to disallow incommensurability because, as he says, paraphrase and translation and understanding are all and only internal to our own language . But Quine does not acknowledge bilingualism in any pertinent way; more than that, having admitted there is no principled difference between intra- and interlinguistic communication, he offers no grounds for demarcating that linguistic community relative to which translation and paraphrase are entirely "internal ." (Quine precludes the question of intra linguistic incommensurabilities.) Furthermore, Quine's doctrine of the "indeterminacy of translation" is, effectively, the admission of a benign form of intra linguistic incommensurability. (Instructively, the same oversight appears in Rorty's insistence on "solidarity.") In short, the resolution of incommensurabilities (of any sort) does not presuppose the accessibility of a neutral language: one in which (i) all incommensurabilities can be translated; one in which (ii) no incommensurabilities arise; and one regarding which (iii) we have compelling legitimative reasons for believing that what independently obtains in the real world is formulable in its terms. I rest my case.
What, despite their differences, is missed by Quine and Davidson and Rorty and those Wittgensteinians who deny that it makes sense to think of plural "forms of life" is simply that notions like "our society" and
"our Lebensform " are holistic notions—notions that never function criterially . Hence, if one speaks of managing paraphrases of the alien utterances of the speakers of "another" society from the vantage of "ours" alone, one intrudes a sense of the boundaries of our collective practices that we cannot, in principle, confirm. For, these same "interlinguistic" puzzles arise "intralinguistically" as well. There is, in principle, no difference between the two. If you treat "our" society as a universe, then the resolution will be arbitrary; if you treat it as a "world," then you will have acknowledged plural "forms of life" in spite of your intention.
The point of these remarks is to demonstrate that a seemingly decisive puzzle (incommensurability)—in effect, a decisive objection against claims like those adumbrated in (8.41)—arises and is easily resolved in synchronic terms; if you grant that, however, there will be no reason to suppose that incommensurabilities generated by historical processes need be any more difficult in principle.
I therefore content myself with listing a series of provocative theorems that follow from all this:
(8.44) legitimative claims need not preclude, ignore, or override incommensurabilities;
(8.45) epistemic incommensurabilities may take a coherent form;
(8.46) epistemic incommensurabilities entail some form of "robust relativism"—although relativism need not entail incommensurabilities of any kind;
(8.47) truth-claims and legitimative claims are inherently horizonal;
hence,
(8.48) epistemic incommensurabilities entail the ineliminability of relativism and historicism within the strong sciences.
I take these findings to be the effect of Kuhn's work (at least in recent American philosophy), although Kuhn has to some extent retreated from them.
I remind you (in the spirit of chapter 4) that I am not relativizing truth or "relationalizing" the (alethic) reading of truth-values. Hardly anyone, however, denies the "relativity" of epistemic questions. Most suppose (even Putnam, who energetically opposes relativism) that our conceptual and epistemic resources do betray a certain "cultural relativ-
ity." My own argument is that, once "epistemic incommensurabilities" are admitted, some form of robust relativism cannot be precluded. By epistemic incommensurabilities, I mean no more than incommensurabilities affecting the use of epistemically pertinent distinctions—even regarding legitimative reason. I see no way of admitting incommensurabilities and denying that they pertinently affect our assessment of what we know. I should perhaps add, to clinch matters, that
(8.49) horizonality is, intrinsically, a source of incommensurabilities—preeminently, epistemic ones.
This is the common lesson to be drawn from Kuhn and Foucault and Gadamer, and, read in accord with (8.41), it confirms the ease with which Wittgenstein's remarks on Lebensformen may be made to yield findings of a similar sort.
History, however, can claim no privilege in effecting conceptual change. On the contrary, history remains forever blind: it is the mere medium of thinking. That is the key to the resolution of the antinomy of history (mentioned earlier), viz.: that, on the one hand, we judge what to regard as true in our historicized world; that, on the other, the grounds on which we so judge are themselves historicized.[14] If we accede to this and simply continue in our inquiries, we are merely arbitrary in our claims, and if we advance legitimative criteria by which to escape the historicizing process, we merely return to the presumption of privilege (transcendentalism, for instance, in Husserl). The dilemma is a real one. The resolution then is this:
(8.50) there is no principled epistemic resolution of the antinomy of history, but there is no need for one: epistemic practices that are neither arbitrary nor privileged are lebensformlich—although not criterially .
I am convinced that (8.50) is the best (and only) possible solution, but I admit its meaning may not be sufficiently clear. I have a reason for not coming to that too quickly: I want to emphasize first that the solution of the epistemological question will be the same solution that will serve a comparable question in moral philosophy. We shall go through the horns of the dilemma in both inquiries and, in doing that, confirm at one stroke the futility of all "naturalizing" strategies. For, "naturalizing" invokes the same dilemma everywhere .
The key, then, is this:
(8.51) epistemic competence is lebensformlich, but lebensformlich practices are merely viable.
What I mean is: what is lebensformlich are no more than the collective practices of our society within which our cognitive powers take form—and change through actual use . They are not criterial in any invariant sense.[15]
You may find the answer disappointing; I say rather that it is incomplete. The trick is this: one solves the antinomy by retiring it. That is what I mean by historicizing thinking . I do not historicize truth or knowledge, only assumptions. Nevertheless,
(8.52) although it affects its own history, legitimation cannot, epistemically, anticipate its own historical revision.
Hence,
(8.53) history is epistemically blind.
I shall risk one further remark. Admit theorem (8.51) and what it collects from (8.44)–(8.50) and (8.52)–(8.53), and you cannot fail to see that
(8.54) ascriptions of knowledge cannot be made distributively, unless derivatively on lebensformlich grounds.
Being holistic, the lebensformlich cannot be privileged—because it cannot be criterial. Knowledge is primarily an artifactual competence, not a determinate psychological state (for instance, by (5.1), (5.3), (5.10), (5.18), (8.35), (8.37)): to know that snow is white, for instance—or that " 'Snow is white' is true"—is to be able to satisfy whatever pertinent criteria may be drawn from the collectively defined instantiated competence our Lebensformen may support. Just as the meaning of isolated sentences cannot be disjoined from their deeper lingual context ((3.17)), so too seemingly isolated instances of knowledge cannot be disjoined from their instantiating a certain lebensformlich competence.
The objection may be put more tellingly.[16] It cannot fail to be question-begging to insist (with Goldman) that the conditions of knowledge may be determined "scientifically" or (with Kitcher) that the his-
tory of scientific success simply supplies the evolving methodological conditions of knowledge. Goldman tends to treat epistemologies as either naturalistic or aprioristic. But if one considers theories like Duhem's holism (which, I warn, is altogether different from Quine's, although Quine's is often conflated with Duhem's) or van Fraassen's positivism (which replaces considerations of truth by considerations of adequacy, in regard to the relation between explanatory theory and observational data), then we are forced to acknowledge a "third" sort of epistemology (neither naturalistic nor aprioristic) that resists reduction of the epistemic to the non-epistemic: to the psychological or the causal or the like; that remains reflexively critical without invoking privilege; and that is noticeably congruent with an an-archic stance. (I am not endorsing Duhem or van Fraassen, but I cannot see how their theories can be initially disallowed—or subsumed under the usual disjunctive options.) The irony remains that appeal to the cognitive achievements of science may be adversely judged to have favorably (but arbitrarily) interpreted the deeper indeterminacies of scientific practice itself. Naturalizing epistemologists typically ignore the question. Once admitted, it shows that naturalism cannot be defended on naturalistic grounds.
This is the strong reason for disallowing all causal, naturalizing, or merely first-order normative readings of cognitive ascriptions of the form, "S knows that p, " where "p " ranges over discrete propositions. Epistemology cannot be satisfactorily grounded solipsistically or in terms of individual psychology or in accord with the usual readings of JTB. For instance, it counts against the usual versions of the now-fashionable doctrine of reliabilism:[17] that is, the doctrine that holds that knowledge may be rightly ascribed an agent who, believing that p, does so in a way that either "optimizes" or "satisfices" (sufficiently) the reliability-producing conditions under which he or she so believes and that resists admitting that the "reliabilizing" conditions are themselves inseparable from the profoundly horizonal constraints under which such ascriptions alone make sense.
One may say, for instance, in accord with (8.54), that it is more important to know who putatively knows that p (given his or her deeper epistemic competence: a mathematician or a child, say, with respect to an arithmetic proposition) than to know what isolated proposition is putatively known. No naturalist has ever been able to give a convincing and ramified sketch of what is to count as the sufficient causal conditions for confirming that anyone actually knows that p . The prospects for defeating any pertinent claims of the causal sort (however ramified) are plainly beyond control.
What I claim is that the ascription can never be more than an informal abstraction from a horizonally shifting lebensformlich competence. This is why, for instance, it is such a conceptual embarrassment to try to state whether and why, by our lights, "primitive" peoples simply lack knowledge altogether! But even to succeed in sustaining such ascriptions is to admit conditions that the causal or naturalizing formula usually ignores. In part II of this primer, I consolidate the distinctions I am alluding to, in terms of what I shall characterize as the "Intentional." For the moment, however, let me simply say that
(8.55) ascriptions of knowledge are validated more in accord with the pertinent phases of an agent's career than with the causes of that agent's particular beliefs.
For, of course, if the naturalizing of epistemology fails (5.13), then
(8.56) either the valid ascription of knowledge is not a causal matter at all or the causes of knowledge cannot be specified in naturalized terms alone.
That is, the causal account of knowledge, of the naturalizing sort, is simply question-begging.
Ascriptions of truth, knowledge, rationality, and the like are, as I have argued, artifacts of our horizonal reflections. They are in this sense incapable of being totalized, precluded from progressivist optimism, and subject to lebensformlich incommensurabilities. But it is also part of my argument that, despite these important constraints, intra- and intersocietal communication is plainly effective. My only caveat about communicative success is an analogue of what I have already said about progress and progressivism. Discourse, viewed in terms of first-order communication, is decidedly successful—whether we pursue first- or second-order questions. But that alone cannot confirm that success implicates our being guided or governed by any universal norms of rationality, any invariant second-order rules of thought or understanding. That, I say, is a complete non sequitur . I cannot see that it is significantly different in the universalizing presumptions of Kant, Husserl, Peirce, Popper, Gadamer, Apel, or Habermas. All these theorists, it seems to me, are guilty (in the same sense) of construing first-order success in understanding or communication as incoherent or inexplicable unless some invariant second-order forms of reason are acknowledged to gov-
ern our pertinent first-order successes ((8.14)). Once you realize that the imputation of first-order success is itself no more than consensual and horizonal—in precisely the same sense in which (as I have argued) reference is—that particular game is utterly lost.
Briefly, then,
(8.57) progressivism is untenable.
Where the rational guidance of inquiry, thought, work, commitment, and the like are said to be, necessarily, changelessly embedded in empirical thinking (as in Kant's view or, in a way, already in Descartes's "natural light of reason"), progressivism may be displaced by the possibility of punctuated instantiations of the requisite norms; where the reclamation of ideally adequate norms of rationality is itself the gradual work of societal learning (as in Peirce and Habermas), progressivism is directly favored. In the first, reason is said to be constitutive in first-order thinking; in the second, it is regulative only. On the argument of the first, what is normatively constitutive is also normatively regulative. Both options fail to satisfy (8.57).