JAN OCT NOV Previous capture 5 Next capture 2009 2010 2011 4 captures 14 Jun 07 - 5 Oct 10
email@example.com || Home Place
"Terminal Philosophy" -- Revised & expanded, 3/10/00 -- Chapter 1 of Minimal Epistemology: Beyond Terminal Philosophy to Truth, a book MS in progress (as one says, ever hopeful)
�1. Introduction �2. Archeology of the Terminal �3. Circularity, Transitivity and the Terminal �4. Terminal Irrealists �5. Track-Record Arguments �6. Internalism, Deontology andTransitivity �7. Using Language to Get Outside Language �8. No Higher Viewpoint? References �1. Introduction
The terminal philosopher thinks there must always be beliefs or other posits that are terminal: they can neither be justified nor criticized by inference from anything further. In any context whatever, reason giving must at some point leave off - not for practical reasons, such as lack of time, energy or resources, but in principle. No further argumentative recourse is possible at this level of fundamentality. The terminal matters must therefore be justified or criticized non-inferentially - that is to say, immediately - or not at all. Most terminal philosophers conclude "Not at all," on the ground that there is no such thing as unmediated knowledge, justification or critique.
Terminal philosophy splinters into many groups, among them the dominant philosophies of our day. The most rigorously terminal of them all may be foundationalism. The foundationalist argues with great precision that there must be matters not justifiable by inference from anything further, and that nonetheless they can be rationally justified, often enough, by some sort of immediate intuition, perception, or other epistemically privileged access to their truth or at least to their being more likely true than not, or prima facie so. What makes the foundationalist terminal is not the claim that the terminal matters can somehow be justified thus non-inferentially, but the prior claim that there must always be a level of fundamentality at which further argumentative recourse is impossible in principle; rational reason-giving must terminate. It is this prior claim that other terminal philosophers share with the foundationalist, who may share little else. Indeed most terminal philosophers pride themselves in being thoroughly anti-foundationalist. There is some irony then, in discovering that they share foundationalism's insistence on the inevitability of grounds beyond which argumentative recourse is impossible.
Even contextualism can be terminal. "Linear contextualism," as we may call it, regards reason giving as always linear, no matter the context, so that in each context of reason giving there must be terminal beliefs or judgments not justifiable even in part by inference from anything further in the context.(1) What linear contextualists add to this terminal thesis, thereby distancing themselves from foundationalists, is that relative to a given context the terminal matters need no justification; they are "contextually basic," though in some other context they might be open to question and require justification. In this way, linear contextualists can hold that nothing is immune to revision (not that foundationalists cannot). But even when they hold that nothing is immune to revision, they count as terminal philosophers in virtue of thinking that whatever the context, there must be matters not justifiable by inference from anything further in the context.
By contrast, non-linear contextualists, who may include some coherentists, tend to be non-terminal, as do non-linear pragmatists.(2) Thus even if you hold that at any particular stage of inquiry there will be beliefs that we are currently leaving alone, it does not follow that you are a terminal philosopher. Much depends on your reasons for holding this. If you hold that there will be such beliefs because reason giving must always at some point come to an end not for practical reasons but in principle, then you are terminal. If you hold that it must end only because of limitations of time, energy and the like, then you are not.
Historically there have been two main ways philosophers have concluded that we must go terminal. One is a kind of regress argument, presented in more detail in the next section: reason giving must at some point terminate because the alternative can only be either vicious regress or equally vicious circularity. A variant on this theme is Agrippa's trilemma, so called by Michael Williams.(3) When any proposition, advanced as justifiably believed true, is challenged, there are only three ways of responding:
(1) Refuse to respond, i.e. make an undefended assumption. (2) Repeat a claim made earlier in the argument, i.e. reason in a circle.
(3) Keep trying to think of something new to say, i.e. embark on an infinite regress.
Since there is no fourth option, according to Agrippa (and Williams, whose contextualism favors (1)), it follows that on pain of vicious regress, as in case (3), or vicious circularity, as in case (2), reason giving must terminate as in case (1). Agrippa's trilemma amounts to an application of the regress argument.
The other main way of going terminal, closely related, is a kind of circularity argument: any practice of reason giving presupposes some principle or medium M, such as non-contradiction, induction, some categorial or conceptual scheme, language game, community of discourse, whatever. Consequently, if reason-giving were applied to M in order to justify it, or even to criticize, then the very attempt to do so would necessarily assume M, a fatal circularity. No further argumentative recourse is possible at this level of fundamentality; rational reason-giving must terminate. The terminal matters must be justified either immediately - say by some sort of intuition or perception or other unmediated grasp - or not at all.
Despite the long-entrenched appeal of this kind of circularity argument, we see in this chapter, and in more detail in the next two, that it proves to rest on a presupposition both unwitting and false. So do the regress argument and Agrippa's trilemma. The presupposition is that inferential justification is transitive: if x inferentially justifies y, and y inferentially justifies z, then x inferentially justifies z. Stripped of this presupposition and its relatives, the many terminal philosophies become untenable.
The point is not that all forms of vicious or question-begging circularity presuppose that inferential justification is transitive. A would-be justificatory inference from a premise P to a conclusion C, followed by a would-be justificatory inference from C back to P, if meant to yield justification for P, would generally be question-begging whether or not the form of inferential justification involved is transitive;(4) so too for a would-be justificatory inference from P back to P. Rather, the point is that, whether in the course of a regress argument or Agrippa's trilemma or a circularity argument aimed at some medium M, the particular kinds of circularity argument advanced by terminal philosophers presuppose transitivity of the relevant forms of inferential justification, in the sense of requiring it as a premise, as we'll be seeing in some detail in this chapter and the next. Henceforth, when terminal philosophy is said to presuppose transitivity, the point is not that a bare assertion of the terminal philosopher's position entails the transitivity, or that its being true-or-false does. Rather, transitivity is presupposed in the sense that such arguments as terminal philosophers give for their position - or perhaps can give - require as a premise the transitivity of all the relevant forms of inferential justification.
Nor is the point that all forms of vicious or question-begging circularity involve the transitivity presupposition. Rather, the circularity arguments advanced by terminal philosophers presuppose transitivity of the relevant forms of inferential justification. Their circularity arguments treat any closed loop of inferential justification - any case in which x is in its own inferential ancestry, as when x justifies something that justifies . . . something that justifies x - as no different from the case of a would-be justificatory inference from a premise P to the conclusion that P. But this belief that large loops collapse into small ones, hence are epistemically just as bad as the small ones, does not follow given only that some proposition or belief or other matter x is in its own inferential ancestry; the assumption does not follow given only that x justifies something that justifies something . . . that justifies x. One needs in addition to assume that the kind of inferential justification involved in this sequence is transitive. Unless there is some other way to show that vicious circularity must result whenever something is in its own inferential ancestry - which is doubtful, as we see in the next two chapters - the many terminal philosophies become untenable.
The present chapter concentrates on exploring the remarkable extent to which terminal philosophers have relied on the transitivity presupposition and its variants, unwittingly. We'll also explore why terminal philosophy has been so widespread and tenacious throughout both modern and post-modern philosophy alike, and why so much is at stake when the issue is whether to accept or reject the transitivity presupposition. The transitivity presupposition may seem a mere technicality, remote from the large questions of philosophy, but it enables, if it does not drive, most modern and post-modern approaches to them.
The terminal philosopher has long been a familiar figure, though by more tactful names. Richard Rorty counts as terminal when he says there must be "final vocabularies," the factual included, beyond which there is no noncircular argumentative recourse, "only helpless passivity or a resort to force."(5) But Descartes, too, counts as terminal, at least to the extent that his most basic principles and intuitions are not to be justified (or criticized) by way of inferring them (or their denial) from something further. Kant likewise qualifies as terminal, at least if one interprets his philosophy as implying that the categories of the understanding can be neither justified nor criticized, as regards whether they conform to the things in themselves, by inference from matters that are intelligible only through those selfsame categories, as are the things in themselves; you cannot use the categories to get outside the categories. Historicize the categories, as do many of Kant's descendants, relativize them to methods of verification or to forms of life or to "frameworks," or think of them as conceptual, or linguistic, or as Derridian infrastructural matters of différance, and you have the recipe for today's typical terminal philosopher, who is often a relativist of some stripe.
Even Wittgenstein is terminal, at least if one reads him as supposing that there can be no noncircular grounding of our factual language - and no critique of it either - as regards whether key aspects of our factual language somehow conform to the world as it is in itself, prior to or independent of language. David Pears is among those who read him this way, as holding that
we cannot possibly justify our factual language by appealing to facts which can only be stated in it. In order to justify it, we would need to stand outside it and assess its appropriateness to phenomena specified in some entirely different way. But an independent specification of the phenomena is precisely what we lack.(6) On this reading, Wittgenstein's argument, however tacit, is a kind of circularity argument: the very use of factual language, even if not itself a belief or belief-like, nonetheless is a practice presupposed by our most basic factual beliefs, and thus is a required part of any would-be justification of them. They in turn would be necessary, in conjunction with other matters, for justifying our more complex factual beliefs, including any about the relations between language and the world. Therefore, if we appealed to the more complex beliefs to justify the very use of factual language - say on the ground that key aspects of the language somehow conform to the world as it is in itself - it would follow that the very use of factual language is part of what justifies itself, a manifestly question-begging circularity.
Instead, Wittgenstein says, "Grammar is not indebted to reality. Grammatical rules first determine meaning (constitute it) and are therefore not responsible to any meaning and are to that extent arbitrary."(7) The grammar is not imposed on us by reality or by any superior necessity, but by our own conventional language games or forms of life. We cannot describe a pre-existing situation and show that it forces us to adopt a particular grammatical rule; the very attempt to do so would require us to use language to get outside language. Shades of Kant.
Granted, the circularity argument some read in Wittgenstein is not spelled out in any of his texts. But something very like it can be found in philosophers who influenced him, it was considered a convincing line of argument by contemporaries he admired (as well as many today), and I have merely offered what seems a fair reconstruction of it. Furthermore, if this line is not what ultimately lies behind Wittgenstein's rejection of the very idea of grounding factual language - or behind its rejection by those who extrapolate this line from Wittgenstein - we are owed an account of what does. The same goes for other philosophers who, with little by way of explicit argument, likewise treat language as terminal - or, if not language, then some conceptual or categorial scheme, framework of practices, form of life, or whatever. If something like the circularity argument sketched above is not what is supposed to warrant their terminal stance, they owe us an explicit account of what is (see further Ё1.6-1.7, below).
In any case, terminal philosophy is widespread, deep, and tenacious. We may begin to appreciate the extent to which this is so by drawing on Ian Hacking's perceptive book, Why Does Language Matter to Philosophy?
language matters to philosophy now ... for the reason that ideas mattered in seventeenth-century philosophy, because ideas then, and sentences now, serve as the interface between the knowing subject and what is known. The sentence matters even more if we begin to dispense with the fiction of a knowing subject, and regard discourse as autonomous. The topics of this or that school, of 'linguistic philosophy', 'structuralism', or whatever, will prove ephemeral and will appear as some of the brief recent episodes by which discourse itself has tried to recognize the historical situation in which it finds itself, no longer even the interface between the knower and the known, but as that which constitutes human knowledge.(8)
The philosophers Hacking has in mind hold further that whether we regard discourse as the veiling interface between the knowing subject and some would-be object of knowledge among the things in themselves, or as that which autonomously constitutes knowledge, discourse can no more be used to get outside itself than, say, a Kantian category. The very attempt to describe a pre-existing situation, in order to see whether our discourse conforms to it as it is in itself, presupposes what would be at issue, namely that the discourse succeeds in describing the situation as it is in itself. Here, if nowhere else, reason-giving must terminate.
During what Hacking calls the "heyday of ideas," philosophers were driven to a terminal stance about the ideas. After all, they thought, because our knowledge of the world is made possible only through our immediate experience - our sensory impressions or ideas -- our knowledge of the world is limited by what we can infer from properties of our ideas. We have direct access to our ideas or experience, but not to the world itself. Unless we can somehow pierce the veil of ideas, all talk of how the things are in themselves is ultimately doomed to failure. Today, in what Hacking calls the "heyday of sentences," many philosophers replace ideas in this scheme with language as the terminal interface between us and the world. In the process, many go further and reject the very notion of a world independent of our linguistic and other schemes of representation, the very notion of something veiled. Thus Wittgenstein: "Philosophy simply puts everything before us, and neither explains nor deduces anything - since everything lies open to view there is nothing to explain. For what is hidden, for example, is of no interest to us." Nothing is ultimately hidden from a sufficiently supple and diligent analysis or phenomenology. Or so we are told.
How have so many philosophers been seduced - or perhaps coerced - into going terminal? What strand in the history of philosophy, what argument or assumption, has been enabling if not driving the terminal stance? A bit of archeology seems in order, beginning, as so often, with Aristotle. �2. Archeology of the Terminal
Aristotle recognized that if we are to have knowledge of the conclusion of an argument on the basis of its premises, we must know the premises. But if knowledge of a premise always required knowledge of some further proposition, then in order to know the premise we would have to know each proposition in an infinite regress of propositions. Since this is impossible, the regress must terminate with some propositions that are known, but not known by demonstration; there must be basic, non-demonstrable knowledge which grounds the rest of our knowledge.(9)
This kind of terminal conclusion - that there must be knowledge not demonstrable from anything further - together with the regress argument for it, lies at the heart of Aristotle's theory of demonstration. And it was this theory that set the agenda and the form of Euclid's Elements, which is basically a systematic application of Aristotle's theory of demonstration to geometry. The axioms, as they came to be called, represent the terminal, non-demonstrable knowledge which is to ground the rest of our knowledge of geometry. It is a measure of Euclid's influence that until just over a century ago the Elements went into more printings in the West than even the Bible. For a couple of millennia the Elements provided the received model of the structure of reason-giving, an ideal to which serious thinkers in all fields should aspire. This helps explain the continuing appeal not only of terminal ideas but of regress arguments generally, together with what they presuppose.
Notoriously, regress arguments for there having to be terminal propositions give rise to a crucial problem. How are the termini to be known - or at least justifiably believed - if not by demonstration or other inference from something further? Are there criteria whose presence to the aspiring knower would non-inferentially certify the truth, or at least the probable truth, of any propositions that satisfy them? If so, the propositions they assure could serve as the bedrock on which to build the rest or our knowledge; we could start with what is present to us, hence better known, then build outward by means of inference from the certified terminus in us.
This problem of the criterion consumed epistemology in the 16th and 17th Centuries. Descartes thought he had found the needed criterion in clarity and distinctness; anything I perceive with the same clarity and distinctness with which I perceive my own existence must be true. He aimed in this way to defend foundationalism and to silence the skeptics by solving the problem they had done so much to elevate to crisis proportions.(10) But we too often forget that there were other, non-foundationalist responses to the Pyrrhonian crisis. Fideists - Protestant or Catholic, dogmatic or moderate - argued that ultimately everyone must take certain basic matters on faith. Neither Descartes's cogito nor any other rational criterion suffices as the ultimate ground of belief; even reason itself is subject to doubt if not backed by faith, as Kierkegaard would repeat two centuries later.(11) Faith is what supplies the needed criterion, where faith is some (divinely inspired) self-certifying act, state or faculty that yields non-rational yet warranted trust in the otherwise unjustified beliefs. Relativists, by contrast, agreed with the skeptics in seeing little to choose between such fideism and Cartesian or other foundationalism. Talk of faith, of clarity and distinctness - or of any other would-be immediate justification or intuition or given - merely obscures the lack of any justification properly so called. Our beliefs can only be justified relative to the presupposed practices of an individual or community.
Thus there were essentially four parties to these disputes: foundationalists, skeptics, fideists and relativists. In various guises the same four continue to dominate much of the landscape, in that many philosophers tend to be one or another (often relativists of some sort). But what the four parties share is far more significant than what divides them. They share the presupposition that the structure of epistemic justification is foundational, despite the fact that many of them reject foundationalism - the view that the most basic or terminal assumptions can be rationally warranted, just non-inferentially. After all, rational method is a matter of inferring propositions from further and better known propositions. Since this cannot go on to infinity, there must be some terminus, some "foundation," beyond which there is no noncircular inferential justification, no noncircular argumentative recourse. This is the characteristic view of the terminal philosopher, who in our day is often a relativist or skeptic.(12)
What divides terminal philosophers is what to say about the terminus. According to Descartes and other foundationalists, then and now, there exist criteria that enable us to start with beliefs whose truth, or at least whose probable truth, is justifiably believed non-inferentially, in light not of tradition or authority but some sort of rational intuition or considered judgment (both of which can be quite tentative and fallible, according to many of today's foundationalists). Not so, say the fideists; the terminus cannot be justified by so-called rational intuition, even prima facie, but only by a non-rational state of faith. A plague on both your houses, say skeptics and relativists; no terminus is known or even justified, whether by means of faith or rational intuition or anything else.
Why have the terminal parties all presupposed, from at least the 16th Century on, that there must inevitably be termini beyond which there can be no noncircular argumentative recourse? One reason, obviously, is the chronic influence of the regress argument. We need therefore to put the regress argument under a microscope.(13) When we do, we find that the regress argument is a deductive argument for, in the first place, the conclusion that at some point inferential justification or reason-giving must end, on pain of vicious regress or of vicious circularity. Like all deductive arguments, this one can be represented as a reduction to absurdity of conjoined assumptions, each of which seems plausible in itself:
(1) There are justified beliefs.
(2) Every justified belief is justified by inferring it from some justified belief or beliefs.
(3) No belief justifies itself.
(4) If a belief x justifies a belief y, and y justifies z, then x justifies z.
(5) There is no infinite sequence of beliefs each of which is justified by inferring it from its predecessor.
These five propositions jointly entail a contradiction. For (1)-(4) entail the contradictory of (5), namely
(6) There is an infinite sequence of beliefs each of which is justified by inferring it from its predecessor.(14)
To escape this absurdity, Aristotle rejects (2): Not every justified belief is justified by inferring it from justified beliefs; some are justified non-inferentially, immediately, directly, intuitively. Among these non-inferentially justified terminal beliefs are some Aristotle counts further as known; there is non-demonstrable knowledge. Cartesian and other foundationalists likewise reject (2), as do the fideists, who insist only that the termini are certified not by rational intuition but by faith. Skeptics and relativists argue to the contrary that the culprit is proposition (1) - there simply are no justified beliefs.
What is striking is that none of the four parties questions proposition (4) - not that they are aware of it - despite the fact that what (4) asserts, namely the transitivity of inferential justification, is necessary to derive a contradiction; without the transitivity there is no absurdity.(15) The regress argument not only for foundationalism, but for the necessary prior conclusion that at some point inferential justification or reason-giving must end, presupposes that all the relevant forms of inferential justification are transitive. So too, therefore, do regress arguments for fideism, skepticism and relativism, which differ from the regress argument for foundationalism only in their response to (1) and (2). Insofar as Agrippa's trilemma amounts to an application of the regress argument, it too presupposes the transitivity.
Note also that the regress argument contains a built-in circularity argument, as it must. For in order to exclude not only vicious regress, which is expressed by (6), but vicious circularity, which is unexpressed (except for the special case mentioned by (3)), there must be some argument to the effect that there cannot be "closed loops" of inferential justification, namely sequences of inferential justifications in which some proposition is in its own inferential ancestry. For example, consider a simple kind of sequence of inferential justifications in which Jxn1 (x inferentially justifies n1), Jn1n2, Jn2n3, ... Jnnx. In this sort of sequence, x justifies something that justifies . . . something that justifies x; hence x would be justified, since everything in its inferential ancestry (namely, every proposition in this sequence) would be inferentially justified, in line with (2). The built-in circularity argument against closed loops is that since J is transitive by (4), this and any other closed loop would entail that Jxx - that x inferentially justifies itself - which is intolerable and is duly excluded by (3).
That is, the built-in circularity argument tries to exclude closed loops of inferential justification by reducing them to the case in which a belief x is both the conclusion and the premise of a single, one-step justificatory argument meant to create the sole justification for x (the case mentioned by (3)). And this case is a paradigm of vicious circularity, whereas a belief's occurring somewhere or other in its own inferential ancestry is not; a number of philosophers deny that the latter always entails vicious circularity, whereas no one denies that a one-step justificatory argument whose conclusion is its sole premise is viciously circular. The more complex situation in which x is meant to justify something only in part - that is, only in conjunction with other premises - will be considered below and in the next chapter.
The question whether a justified belief may occur somewhere in its own inferential ancestry is one of the crucial issues between terminal philosophers and a number of their leading opponents (often coherence theorists of some sort, but by no means always). Terminal philosophers follow their foundationalist colleagues in rejecting all such closed loops,(16) whereas a number of non-terminal philosophers emphatically do not. Hence terminal philosophers are not entitled to help themselves to the intuition (or conventional wisdom) that no justified belief may occur anywhere in its own inferential ancestry. This explains why so many of them invoke the circularity argument built into the regress argument, which is meant to advance the debate beyond intuition fights by reducing all cases in which a belief x occurs in its own ancestry to the case in which x is both a premise and the conclusion of a single justificatory argument of the form P therefore P. The latter, as they see it, is a paradigm case of vicious circularity about which everyone agrees. The trouble is that this built-in circularity argument presupposes that the justification relations involved in the ancestry are transitive, as we'll see below and in detail in the next two chapters. As we'll also see, not only are there serious counter-examples to the would-be transitivity, there are other, more general arguments against it.
For the moment, however, it may help to get at least a preliminary idea of what one kind of argument against transitivity looks like. Consider the would-be transitivity of a certain specific form of inferential justification, namely inference to the best explanation. Suppose that (i) x inferentially justifies y by virtue of y's being the best explanation of x, and (ii) y inferentially justifies z by virtue of z's being the best explanation of y. If inference to the best explanation were transitive, it would follow that (iii) x inferentially justifies z by virtue of z's being the best explanation of x. But by hypothesis (i), y is the best explanation of x, so that z cannot be the best explanation of x. Therefore, contrary to (iii), x cannot inferentially justify z in virtue of z's being the best explanation of x; inference to the best explanation is not transitive. This argument may seem too true to be good., as I remarked on its first appearance in print.(17) We'll see about that in detail in the next chapter. And in Chapter 3 we'll see whether philosophers so inclined could somehow restore the lost circularity arguments without presupposing that all the relevant forms of inferential justification are transitive.
Meanwhile, let us explore further the extent to which the circularity arguments used by terminal philosophers for a terminal stance toward some particular medium - categories of the understanding, language, final vocabulary, conceptual scheme, framework, form of life, practice - share a crucial presupposition with regress arguments for the inevitability of some terminus or other, namely the presupposition that all the relevant forms of inferential justification are transitive. The presupposition persists like the Cheshire Cat's grin, even among many philosophers who reject regress arguments. In this respect, the received ways of going terminal - a regress argument, Agrippa's trilemma, or a circularity argument aimed at some medium M - come to much the same.
�3. Circularity, Transitivity and the Terminal
It is true that many foundationalists and other terminal philosophers allow some forms of inferential justification that are not transitive. Even Descartes appears to allow that outside philosophy some non-deductive forms of inference can be counted on to justify their conclusions, including inference to the best explanation. Nonetheless, Descartes and other terminal philosophers, typically unconsciously, allow only transitive inferences about the deepest or most basic affairs - those within the philosopher's exclusive purview - thus remaining terminal philosophers about such deep matters even when they grant non-transitive inference elsewhere.
Let us call such philosophers deep-structural terminal philosophers. Deep-structural terminal philosophers are not committed to the terminal structure of all justification, only to the terminal structure of justification concerning suitably deep matters (the philosophical). In particular, as we see in detail below, the transitivity presupposition lives on in the conventional wisdom that if a categorial framework, conceptual scheme, final vocabulary, fundamental practice (reason included), or language had a ground, it could only be, absurdly, a ground known immediately or directly, on pain of either regress or circularity. The conventional wisdom is driven by deep-structural foundationalism even when it vilifies foundationalism.
It begins to appear that terminal philosophers share a crucial presupposition with Descartes, quite as unwittingly as Descartes himself, even those who congratulate themselves for having overcome his insidious influence. The presupposition can be seen at work in what is widely considered a commonplace of Cartesian dialectic. On this reading, Descartes concludes that we cannot, on pain of vicious circularity, justify our factual belief in an external world by appealing to facts which themselves could only be objects of such external-world belief. The idea is that our most basic external-world beliefs x - say the perceptual belief that here is a lump of wax - are what inferentially justify our less basic external-world beliefs y; these in turn inferentially justify our still less basic external-world beliefs z; so if we appealed at last to these still less basic beliefs z to inferentially justify our most basic external-world beliefs x, it would follow that our most basic external-world beliefs x inferentially justify themselves - a tight little closed circle prohibited by the principle that nothing can inferentially justify itself.
This argument has the form: Jxy (x inferentially justifies y)and Jyz; were it the case that Jzx, it would follow that Jxx, contrary to the prohibition of Jxx for any x. The argument presupposes that J is transitive (since otherwise it would not follow that Jxx); the point is not that all circularity arguments have this form, or that they all presuppose transitivity, only that this one does. Since it never occurred to Descartes to question this transitivity presupposition - indeed in these philosophically fundamental contexts he appears oblivious to it - he was forced to break out of the threatened circle, or try. To this end he argued from his very idea of God to the conclusion that the light of reason or "faculty of knowledge which God has given us can never disclose to us any object which is not true ... inasmuch as it apprehends it clearly and distinctly."(18)
The latter move is widely supposed to have landed Descartes in what we call the Cartesian Circle, pointed out by Mersenne and Arnaud, which arises when Descartes relies on clear and distinct ideas to prove the existence of God, and on the existence of God to validate clear and distinct ideas. This familiar Cartesian Circle is a further circle, one which arises as a result of his attempt to avoid the circularity he assumes must result if one tries to justify belief in an external world ultimately by appeal to basic external-world beliefs like "Here is a lump of wax." And this prior circularity arises on the above Cartesian argument only if the philosophically relevant forms of inferential justification are all transitive.
It is often said that Descartes's mistake here is to require too strong a form of inferential justification, namely deductive inferential justification. This is indeed a mistake. But suppose instead that he had allowed some non-deductive inferential justification at this deep philosophical level. Still the problem would remain if such non-deductive inferential justification must be transitive, as we see next, in connection with a 20th-century non-deductive parallel to Descartes's dialectic - parallel, anyway, up to the bit about God.
Recall from �1 the (neo-)Wittgensteinian conclusion that there can be no noncircular grounding of our factual language. The idea, to repeat, is that the very use of factual language, even if not itself a belief or belief-like, nonetheless is a practice presupposed by our most basic factual beliefs, and thus is a required part of any would-be justification of them. They in turn would be necessary for justifying our more complex factual beliefs, including any about the relations between language and the world. Therefore, if we appealed to the more complex beliefs to justify the very use of factual language - say on the ground that the language, or a key aspect of it, somehow conforms to the world as it is in itself - it would follow that the very use of factual language is part of what justifies itself, a manifestly vicious circularity.
The argument here has the form: suppose x is part of what inferentially justifies y, and y is what inferentially justifies z; therefore, were it the case that Jzx, it would follow that x is part of what inferentially justifies itself, which is viciously circular. The transitivity presupposed here has the form, "if x is part of what justifies y, and y justifies z, then x is part of what justifies z."(19) To be sure, this is not transitivity in the strict sense of a relation R such that whenever x bears R to y, and y bears R to z, then x bears R to z. Rather, we have here a sort of "extended" transitivity, in which whenever x is part of what bears R to y, and y bears R to z, then x is part of what bears R to z. Let us henceforth call this sort of transitivity partial extended transitivity, or P-transitivity for short.
But, you may well demand, what on earth is wrong with P-transitivity, or indeed with any other sort of transitivity of inferential justification? The short answer is that there are counter-examples to the alleged P-transitivity - cases in which x is part of what inferentially justifies y, y inferentially justifies z, yet x is not part of what inferentially justifies z. Here is one, developed in detail in the next chapter. By 1820 there was extensive observational and other evidence for Newton's theory TN of gravitation. As applied to the known solar system - that is, in conjunction with certain auxiliary assumptions and initial conditions A, including recently observed irregularities in the orbit of Uranus - TN entailed that there is another and unobserved planet X beyond Uranus, and entailed also certain key properties of X's orbit, including its semi-major axis, eccentricity, longitude of perihelion, and such further specifics as
O. Planet X is at orbital position p at clock time t.
Indeed, in conjunction with A, TN entailed O, and by the 1840s TN was seriously offered as justification for O. When astronomers subsequently turned their telescopes to look at p at t, they observed a planet where it should be, called it Neptune, and thereafter offered O as a reason for, hence as part of what justifies TN. In fact they regarded O as a strong reason for TN, hence as a strong part of its justification, since O involved novel prediction. Nor did they withdraw their previous offer of TN as justification for O. To the contrary, they now regarded O as having dual justification, from TN and from observation. O was part of what justified TN, TN justified O, yet O did not justify itself even in part, nor did TN, contrary to what transitivity would require.(20) Also, even though TN (like O) thereby occurred in its own inferential ancestry, TN `s degree of justification or support was increased by O. Parallel examples can easily be multiplied (and suppress for now any objections to using descriptions of scientific or other inferential practice as a source of examples that are meant to have normative epistemic force; the matter will be considered in Chapter 3).
In addition to suffering counter-examples, a number of crucial kinds of inferential justification - perhaps all kinds - can be shown to be non-transitive by way of more general arguments; so too for epistemic relations of evidence, warrant, support, dependence, confirmation, basing, and more (as we'll see in detail in Chapter 2). Meanwhile, let us continue exploring how far the transitivity presupposition is implicated in modern and postmodern philosophy alike. �4. Terminal Irrealists
Realism about a class of affairs is the view that such affairs exist and have certain features independently of our evidence, theories, categories, conceptual schemes, frameworks, practices, forms of life; they are found, not made. Correlatively, realism about truth is the view that truth and falsity are likewise independent of these, hence evidence-transcendent (among other things); even a belief that is epistemically ideal could be false. Irrealism, as the term will be used here, includes the view that the dispute between realists and their opponents is senseless, as well as the view - sometimes called anti-realism - that realism is simply false.
Recall again the general form of argument - a kind of circularity argument - shared by Descartes, (neo-)Wittgenstein and many between. According to the argument, vicious circularity results if we try to use some philosophically fundamental medium - perception, reason, language, framework, practice, categorial or conceptual scheme, or perhaps something presupposed by their very possibility - to get outside itself, in the sense of trying to determine whether its own renditions conform to relevant features of things as they are in themselves. The very attempt must presuppose what would be at issue, namely that the medium does so conform. Many think that much the same sort of argument is at work in Kant's treatment of the categories of the understanding: they cannot be used to get outside themselves, because, on pain of circularity, they can neither be justified nor criticized by appealing to matters themselves intelligible only through the self-same categories. Rorty, wittingly or not, merely generalizes this hackneyed form when he concludes that there must be final vocabularies beyond which there can be no noncircular argumentative recourse, only helpless passivity or a resort to force. To get a recipe for today's typical irrealist, merely historicize the categories, or relativize them to vocabularies or to forms of life, frameworks, practices, or methods of warrant or verification, or think of them as conceptual, or grammatical, or as infrastructural matters of différance, or as presuppositions of the very possibility of one or more of these. In each instance, the underlying circularity argument against using some such medium to get outside itself is presumed fatal to realism.
The typical irrealist thereby presupposes that the relevant forms of inferential justification are transitive. For not only do most irrealists run circularity arguments that parallel Wittgenstein's (or the one many read in him), which we lately explored in �3. Most if not all of them run variations on the "God's-eye-view" or "look-see" objection to realism: realists could justify their view only by stepping outside language to compare it with reality, looking to see that they fit, rather like comparing a picture with the real thing by staring at them side by side, which is absurd.(21)
Despite their popularity, there is less to look-see objections them than meets the eye. Granted, we must conceded that manytraditional realists do imagine some such unmediated intuitive vision or mental grasp of word-world fit - a God's-eye view, a spectator theory of knowledge, a metaphysics of presence. G.E. Moore is not unrepresentative of the type. According to Moore's early epistemology,(22) truth is a simple, unanalyzable, intuitable property of propositions. Any other view, he argued, presupposes that we can somehow get beyond concepts to a reality that sustains them, which is impossible in principle. Here Moore evidently was appealing to - or at least was strongly constrained by - a circularity argument: If we tried to justify the claim that our concepts are sustained by reality (correspond to it), we would have to use the very concepts at issue, clearly a viciously circular procedure, hence impossible in principle. Even when Moore later gave up propositions, and adopted the view that truth is correspondence of a belief to reality, he held fast to a claim of immediate, non-inferential knowledge, namely the knowledge that there is such correspondence. What was driving him to hold fast to immediate, intuitive knowledge of such correspondence was primarily the circularity argument, according to which there can be no inferential justification of belief in such correspondence, on pain of circularity. If there is to be any knowledge at all of correspondence, it must be non-inferential knowledge, that is to say, immediate or intuitive knowledge, which Moore duly concluded. Having so concluded, he was able to conclude further that what he thought of as the traditional problem of epistemology - namely, how to get outside the circle of our ideas or sensations - is no problem at all. To have a sensation is already to be outside the circle; "it is to know something which is as truly and reallynot a part of my experience, as anything which I can ever know."(23)
But what does any of this really show about realism? It shows only that traditional realists are in the grip of the same argument-form that lies behind the look-see objection against them. The idea in both cases is as above: if there were some inferential justification of word-world fit - some further argumentative recourse - it would give rise to a tight little closed circle; hence there can be no noncircular inferential justification or even criticism at this terminal level of fundamentality. If there is to be any justification at all, it must be non-inferential, that is to say by way of some immediate insight into the relation of representation to represented. Look-see. Traditional realists embrace this conclusion, their opponents call it reduction to absurdity. Both are mired in a terminal stance that presupposes transitivity of the relevant forms of inferential justification.
Why? Largely because they reject any sequence of inferential justifications that looks like this: M is part of what justifies x, x justifies y, y justifies M. And they reject it because they think that a sequence like this essentially amounts to the case in which we have a would-be justificatory inference from a premise P to a conclusion C, followed by a would-be justificatory inference from C back to P. They regard the latter sort of sequence as a paradigm case of vicious circularity. But the former sort of sequence amounts to the latter only if the kind of inferential justification involved in the sequence is transitive, or at least P-transitive. Given only "M is part of what justifies x, x justifies y, and y justifies M " it does not follow that (i) M is part of what justifies itself, let alone that (ii) M is the whole of what justifies itself; nor does it follow that (iii) we are committed to a would-be justificatory inference from premise M alone to the conclusion x, followed by a would-be justificatory inference from x back to M. Clearly, (i) would follow only if the form of inferential justification involved is P-transitive, and (ii) and (iii) would not follow even if it were. These are just some of the reasons why large circles of inferences do not automatically reduce to small ones, contrary to the hackneyed taunt against opponents of terminalism that they accept large circles just not small ones.
The same look-see dialectic underlies the widespread terminal conviction, urged by Putnam, that "our language cannot be divided up into two parts, a part that describes the world 'as it is anyway,' and a part that describes our conceptual contribution";(24) hence we cannot use language or conceptual scheme to get outside language or scheme. For suppose we used one part of our language to describe the world "as it is anyway," and another part to describe our conceptual contribution, in order to see whether they match. Since the part we use to describe the world "as it is anyway" must itself include or express some concepts Ci, we cannot appeal to it in order to justify those Ci, or even to criticize them, on pain of circularity; and such circularity would infect our remaining concepts if we appealed to Ci in order to justify them. Much the same line underlies Rorty's related slogan that, on pain of circularity, "one cannot use a part of one's present theory to underwrite the rest of it."(25) So too for Thomas Kuhn's rejection of scientific realism: "there is no theory-independent way to reconstruct phrases like 'really there'; the notion of a match between the ontology of a theory and its 'real' counterpart in nature now seems to me illusive in principle."(26)
Failing some such use of theory, language or scheme to get outside itself, the line continues, all that would remain is a God's-eye view - or a view from Nowhere, a disembodied spectator, a metaphysics of presence, the given, a view "sideways on" - which is absurd. As George Levine says, "Antirealism, even literary antirealism, depends on a sense of the impossibility of unmediated knowledge."(27) We must acquiesce in the necessity of having a language or conceptual scheme that is neither justifiable nor criticizable, as regards conformity to the world in itself, by appeal to matters themselves understood, if at all, only by means of the self-same language or scheme. Derrida puts it this way: "There is no sense in doing without the concepts of metaphysics in order to shake metaphysics. We have no language - no syntax and no lexicon - which is foreign to this history; we can pronounce not a single destructive proposition which has not already had to slip into the form, the logic, and the implicit postulation of precisely what it seeks to contest."(28)
Just how fundamental is the look-see argument? Very. For one thing, when some other argument against realism gets into trouble, irrealists typically fall back on a variant of the look-see argument - or on what drives it, namely, "You cannot use language or scheme to get outside language or scheme." For look-see and this its driver are widely supposed to be the ultimate realism-killers, all-purpose and implacable; when all else fails, they work. Putnam falls back on them when his model-theoretic argument against realism is threatened by a positive theory of how actually to identify the correct mapping (correspondence) between words and the world. He insists that anyone who advances such a theory - say a causal theory - would have to hold up the relevant sign ('CAUSE', 'REFERS') and look to see how this sign relates to its would-be object in THE WORLD, since "'the way the theory is understood' can't be discussed within the theory."(29) You cannot use the theory to get outside the theory, certainly not so far as to appraise the would-be reference of its own words to their would-be objects in THE WORLD. This look-see line is what lies behind Putnam's belief that
elements of what we call "language" or "mind" penetrate so deeply into what we call "reality" that the very project of representing ourselves as being "mappers" of something "language-independent" is fatally compromised from the start.... Realism is an impossible attempt to view the world from Nowhere.(30)
Rorty "whole-heartedly concurs,"(31) nor is he alone. Many philosophers take it for granted that realism necessarily involves an attempt to view the world from Nowhere, or that it involves a God's-eye view or some other heroic disembodiment or attempt to climb outside our own minds or practices, some place outside history, or at least some sort of immediate intuition or insight or view "sideways-on" into the relation of representation to represented, some absurd "metaphysics of presence" or spectator theory of knowledge. But look-see arguments against realism rest on the major premise that you cannot use language or scheme to get outside language or scheme, which rests in turn on a circularity argument that presupposes the transitivity of all the relevant forms of inferential justification. It follows that however widespread and entrenched, look-see arguments are no stronger than the fallacious transitivity presupposition on which they rest.
True, there might be ways of restoring the lost circularity arguments against realism even if the relevant forms of inferential justification are not all transitive. But Chapter 2 will argue that there are not. And yes, of course, look-see arguments are not the only sort of objection to realism. Among the further objections, the most common are those according to which realism must be miserably essentialist or totalizing (there is just one right way of describing and explaining, one right vocabulary, the one true theory); that realism must be noumenal, scientistic, reductive, confrontational, or inimical to freedom, expressivism and the sense of wonder; or that it must suppose we can escape regimes of power and knowledge by fleeing into some transcendental realm of truth and freedom. But realism as such entails none of these; there are versions of realism that simply are not essentialist, totalizing, noumenal, scientistic, reductive, confrontational, or inimical to freedom, expressivism and the sense of wonder, or transcendentalist with regard to truth and freedom; quite the reverse.(32) Irrealists who run such objections have set up a scarecrow.
Another sort of objection to realism starts from the empirical-equivalence thesis: for any theory, there are always indefinitely many empirically equivalent but logically distinct rivals (where "empirically equivalent" means that the theories have the same empirical consequences). From the empirical-equivalence thesis, it is supposed to follow that no possible evidence could count for one theory that does not also count for its empirically equivalent rivals; even our total evidence is not enough to determine which theory is correct or even more likely true than its rivals. Hence it strains belief to suppose, as does the realist, that from among these indefinitely many empirically equivalent alternatives one can be shown to be true. Furthermore, any evidence for the realist's theory is equally evidence for its empirically equivalent anti-realist rivals, so that the dispute between realists and anti-realists is senseless; irrealism is the preferred position. Or so we are told. As we see in Chapter 6, contrary to conventional wisdom, there is far less to the empirical equivalence thesis than meets the eye, and even less to the inference from it to underdetermination.
In any case, a number of other celebrated arguments against realism prove on examination to rest on a look-see objection, which, as seen, rests on a terminal stance, which rests finally on the transitivity presupposition. We've already noted how Putnam defends his model-theoretic argument against realism by falling back on a look-see argument. The following is a further kind of argument against realism, historically more important. As Nicholas Wolterstorff remarks, "fundamental to contemporary anti-realism is the affirmation of nominalism - that is, of ... anti-realism with respect to universals outside the mind." The radical world-making variety of anti-realism "is the resolute spinning out of the implications of nominalism," which has its roots in the medieval view according to which "natures as grasped by the mind are fundamentally altered from how they are in things."(33) I would add only that when we inquire why we must not suppose there are real universals or natures or kinds outside the mind, the typical modern reply - and post-modern - is that we could never justifiably believe there are any. For in order to do so, the reply goes, we would have to adopt the absurd method of taking in one hand some segment of reality, and in the other the universals in the mind, then look to see whether the universals in the mind conform to anything in reality.
When we inquire further why there is supposedly no alternative to this absurd method of look-see, what we find is the idea that any attempt at an indirect or inferential justification of belief in real universals must be viciously circular. For suppose we try to describe some segment of reality "as it is in itself," in order to infer, from this description, something about whether reality filtered by universals in the mind conforms to reality "as it is in itself." Obviously we cannot describe anything at all (say snow as white) without applying universals in the mind (that is, without applying snow-as-grasped-by-us and whiteness-as-grasped-by-us). So universals in the mind - even if not themselves beliefs or belief-like - are presupposed by, hence part of what justifies, any description whatever, including any would-be description of some segment of reality. Therefore, if we appeal to the latter description to justify certain universals in the mind (on the ground that they conform to real universals), we land in a tight little circle: universals in the mind are part of what justifies the would-be description of a segment of reality "as it is in itself"; this would-be description justifies the universals in the mind; therefore, these universals in the mind are part of their own justification. It follows, we are told, that no noncircular inferential or argumentative justification is possible here (and no critique either); if there is to be any justification at all, it must be by way of some non-inferential, immediate or look-see method of justification for belief in real universals. Since this is absurd, we can never know or verify whether there are real universals, and it is otiose to suppose there are. Realism about universals is not even false, it is senseless.
'Trouble is, a crucial move in this irrealist war horse has the form, "x is part of what justifies y; y justifies x; therefore, x is part of what justifies x." And this is just to presuppose that the relevant kinds of inferential justification are P-transitive. �5. Track-Record Arguments
The transitivity presupposition lurks as well in William Alston's rejection of what he calls "the track-record argument" for the reliability of sense experience as a source of belief.(34) The track-record argument is "a simple inductive argument," in which one begins by laying out "a large, carefully chosen sample of perceptual beliefs" Pi, then reports of each such belief Pi that Pi is true:
(1) At t1, subject S1 formed the perceptual belief that P1, and it was true that P1.
(2) At t2, S2 formed the perceptual belief that P2, and it was true that 2.
(n) At tn, Sn formed the perceptual belief that Pn, and it was true that Pn.
Let us suppose that a sufficiently high threshold percentage of the perceptual beliefs Pi are true - say 97% (Alston's figure). That is, 97% of the premises (1)-(n) are true. Then we may inductively infer from premises (1)-(n) that
R. Sense experience is a reliable source of belief.
Hence premises (1)-(n) inferentially justify the principle R of the reliability of sense experience.
The catch, of course, is that (1)-(n) justify R only if (1)-(n) are themselves justified. And how could they be justified? It looks as though the reliability of sense experience stated by R would have to be presupposed in any attempt to justify (1)-(n). For we would have to justify, for at least the threshold percentage of (1)-(n), the belief that it was true that Pi. And to be justified in claiming that a perceptual belief Pi was true, we would presumably ourselves have to acquire the needed perceptions and form the perceptual belief Pi, on the basis of which we take it that Pi was true. Yet this presupposes that the perceptual belief Pi is reliable in the sense of being substantially more likely to be true than not. According to Alston, "It is only by taking sense perception to be reliable that I can regard it as reasonable to believe that there is a tree in front of me when there visually appears to me to be a tree in front of me." So we must presuppose the reliability R that was to be shown. The reliability of sense experience is presupposed by, hence a required part of, the would-be inferential justification of the threshold percentage of (1)-(n).
Thus the reliability principle R would be part of what inferentially justifies the needed percentage of (1)-(n), and the needed percentage of (1)-(n) would be what inferentially justifies R. According to Alston, this must result in "epistemic circularity," as he calls it, in the sense that R - even if itself not a belief but an inference rule - would be part of its own inferential justification. Or at least it would be part of its own justification if and insofar as we are attempting to give a "fully reflective justification" of R, meaning a justification in which every belief employed in showing that R is justified is itself justified. The inevitable price to be paid for attempting a fully reflective justification here is circularity.
This argument is a special case of what Alston calls the epistemic circularity argument: "Where basic sources [of belief] are concerned, it will be in principle impossible to attain full reflective assurance with respect either to the principles of justification or reliability, or to the particular beliefs that fall under those principles."(35) That is, fully reflective reason-giving at this level of fundamentality must terminate. The price of any further reason-giving at this level is epistemic circularity. We can avoid the circularity only by going terminal.
However, this circularity argument works only if the inferential justification involved is P-transitive. Otherwise, given only that (i) the reliability principle R is part of what inferentially justifies the needed percentage of (1)-(n), and (ii) the needed percentage of (1)-(n) inferentially justifies R, it does not follow either that R inferentially justifies itself or that R is part of what justifies itself. If the needed relation of inferential justification is not P-transitive, this type of circularity argument against justifying basic principles of reliability fails. So too for the circularity argument behind Barry Stroud's view that if one's goal is the justification of grounds of science, it is circular to use other science as justification; we need information independent of science about whether the sense data used by science are true of the world, before we can rely on science.(36)
Wait a minute, you may say. Grant for the sake of argument that some sort of transitivity of inferential justification is presupposed here. Grant even that the relevant forms of inferential justification are not transitive. Wouldn't there still be a sense in which the reliability principle R is used in the course of its own justification? And wouldn't this be question-begging? Furthermore, isn't there some epistemic relation E other than inferential justification which is transitive and would thereby result in R's bearing E to itself, hence in R's depending epistemically on itself? For example, what about the relation "x is necessary for the inferential justification of y"? Surely this relation is transitive, and surely (i) R is necessary for the inferential justification of the needed percentage of (1)-(n), and (ii) the needed percentage of (1)-(n) is necessary for the inferential justification of R. So, given (i) and (ii), it follows that the reliability principle R is necessary for its own justification, which is question-begging if anything is.
Yes, of course, there are several relations that must be considered in addition to inferential justification, including the relation of being necessary for justification,(37) and also relations of evidence, warrant, support, dependence, confirmation, basing, whatever. And yes, there would still be a sense in which the reliability principle R is used in the course of its own justification. But we shall see in Chapters 2-4 that not much comes of any of this if you are looking for a relation that will restore traditional circularity and regress arguments in the face of the non-transitivity of the relevant relations of inferential justification. It proves exceedingly difficult to show, if it can be shown, that the brave new relation (i) has the needed epistemic force in particular inferential justifications, (ii) is transitive, and (iii) produces the wanted vicious circularity when something bears the relation to itself and/or appears in its own inferential ancestry. In fact I will be arguing that none of the alternative relations on offer satisfies (i)-(iii), and that until some such relation shows up, not only the targeted circularity arguments and regress arguments, but the closely related Agrippa's trilemma and look-see arguments are fallacious by virtue of presupposing the transitivity of all the relevant forms of inferential justification, evidence, warrant, support, dependence, confirmation, basing or the like. �6. Internalism, Deontology and Transitivity
The transitivity presupposition can predetermine much of one's philosophy by generating a kind of "internalist" stance - a tendency of many philosophers to concentrate on what is graspable by unaided reflection on language, concepts, intuitions, practices and more. Unaided? The reflection is unaided in the sense that it has no need to consult any external view for guidance let alone correction, the sciences certainly included; nothing truly fundamental is ultimately hidden from a sufficiently supple and diligent analysis or phenomenology. Such internalism is the characteristic stance of phenomenology, much analytic philosophy, and most contemporary epistemology. The focus is on what can be ascertained by unaided reflection. The transitivity presupposition tends to confine us to the flat plane of how things appear to the attentive self from its armchair.
Why? Largely because of the kinds of reason-giving the transitivity excludes. One kind is inference to the best explanation, roughly what Peirce called abduction, the non-transitivity of which was argued briefly at the end of �3 and is argued in detail in Chapter 3. Abductive inference is quite common, as when we explain the footprints in the garden, the shattered window, rifled safe and empty beer can on the kitchen table by positing an unhurried burglar. Such inference is also involved when one concludes that certain affairs are best explained by, hence support, the theory that later species evolved from earlier by way of natural selection. Darwin went so far as to say of his theory that "It can hardly be supposed that a false theory would explain, in so satisfactory a manner as does the theory of natural selection, the several large classes of facts above specified."(38) In these kinds of cases, x is said to inferentially justify y in virtue of y's being the best explanation of x (or of the affairs described by x); y provides a better explanatory story - a better understanding - than any competing account. (And stifle for now the usual-suspect objections to inference to the best explanation; they will be addressed in Chapter 6, in connection with a new account of just what sort of beast inference to the best explanation is.)
Now consider again Descartes's conviction that we cannot, on pain of circularity, justify our factual belief in an external world by appealing to facts which themselves are objects of such external-world belief. What if Descartes had allowed the use of abductive inference to justify belief in an external world, instead of requiring deductive or other transitive inferential justification at this level of fundamentality? Then he could have gone "external," positing a richly textured external world as (part of) what best accounts for certain of our experiences - such as seeming to see a lump of wax - without landing in circularity, since as seen circularity would follow by Descartes's argument (�3) only if abductive inference were transitive (though we may want to add the proviso that Chapters 2 and 3 succeed in showing that there are no other epistemic considerations that would restore the lost circularity).
Granted, it is perhaps conceivable that there exists some form of transitive inference which, starting from the flat plane of what is accessible to unaided reflection, yields justified belief in something beyond, something other. But until some such form of transitive inference turns up, the transitivity presupposition blocks the one customary and plausible way of doing epistemic justice to the other, namely inference to the best explanatory story about what might lie behind the passing show, including affairs behind whether something counts as warrant for a belief. Thus in practice, at least, the transitivity presupposition leaves no alternative to a broadly internalist stance, not only in epistemology but in philosophical method generally, where internalism often takes the form of a phenomenology or analysis of what lies epistemically closest to us. The transitivity presupposition constrains the philosopher to concentrate on internal matters about which, supposedly, nothing is ultimately hidden from unaided reflection - including matters of warrant, logic, mind, meaning, and value - and to believe that the hard-won philosophical analyses of such things are epistemically prior to, hence not correctable by, any external account, including above all any from the sciences.
This diagnosis of what generates internalism contrasts with Alvin Plantinga's insightful account, indeed appears inconsistent with it. According to Plantinga, the source of internalism is deontology, "the notion that epistemic permission, or satisfaction of epistemic duty, or conforming to epistemic obligation, is necessary and sufficient ... for warrant. Deontology generates internalism."(39) For (to make Plantinga's good, long story short) I cannot be held epistemically responsible, says the deontologist, for matters about which I have (and can have) no appropriate awareness. Such warrant as my beliefs can have for me is determined by conditions which I can be aware of by unaided reflection.
Plantinga's account strikes me as basically historically correct. The classical deontologism of Descartes and Locke, together with its descendants and variants to this day, have strongly motivated many philosophers to be internalists of some stripe.(40) Where, then, does that leave my diagnosis, which seems inconsistent with Plantinga's? Well, deontology need not be the only thing that has generated or at least motivated internalism, and not necessarily the most basic. For, to begin with, it is possible to be an internalist without being a deontologist. I can believe that what Plantinga calls internalism is true - that "in order for one of my beliefs to have warrant for me, I must have some sort of special or privileged access to the fact that I have warrant, or to its ground"(41) - without believing that the notion of epistemic permission or duty is necessary and sufficient for warrant. This means that one's internalism can be independent of one's deontologism, or even prior to it.
In addition, we need to be clear about what it means to call something internalism. As Plantinga says, "The term is in considerable disarray."(42) He uses the term to denote the view that a belief x is warranted for a person S only if S can be more or less immediately aware via unaided reflection that x is warranted, or aware of what determines that x is warranted; "in order for one of my beliefs to have warrant for me, I must have some sort of special or privileged access to the fact that I have warrant, or to its ground." It seems clear that internalism in this sense is an instance of internalism in the more general sense adopted at the beginning of the present section. According to the more general stance, when considering matters at a suitably deep level of philosophical fundamentality we should confine our attention to affairs graspable by unaided reflection. It follows that when considering specifically the fundamental matter of the conditions under which a belief is epistemically warranted for me, I should confine myself to conditions graspable by unaided reflection - unaided, indeed uncorrupted, by any reliance on alleged knowledge of some external condition to which I could not have the requisite privileged access. One can therefore be driven to internalism in Plantinga's sense by the general terminal stance that presupposes transitivity, all without being a deontologist.
Because internalism in Plantinga's sense is an instance of the more general internalist stance, we should expect that much the same terminal dialectic which, at deep levels of philosophical fundamentality, discourages reliance on belief in external conditions, would likewise discourage such reliance when it comes to the philosophically fundamental matter of the conditions under which a belief is warranted. This expectation is borne out, indeed it is borne out by Plantinga's own account of warrant. By his account, I am warranted in believing x only if (among other things) those of my epistemic faculties that produce x are functioning properly and in an epistemic environment for which they were designed.(43) Moreover, neither the faculties, nor whether they are functioning properly, nor whether the present environment is one they were designed for, is accessible by unaided reflection; we do not have the requisite privileged access to any of these, according to the internalist. How, then, could we ever be justified in positing any of these external conditions allegedly necessary for warrant? Presumably it would have to be by inferring them from matters to which we do have privileged access. And what kind of inference would that be? Suppose that the only kind of inference allowed were transitive, in accord with the general internalist tendency (however unwitting). As seen, this would exclude the one customary and plausible way of going external, namely abductive inference. In the present case, this would exclude positing epistemic faculties that often enough operate as designed and under design conditions, even if positing them would best account for certain key properties of warrant, normative properties included. Yet this abductive positing, near enough, is what Plantinga does.
Thus the transitivity presupposition would frustrate Plantinga's argument that his notion of warrant, arrived at by a kind of analysis of our concept of warrant,(44) is not empty but instead is realized in actual epistemic faculties that often enough operate as designed and under design conditions. The terminal dialectic which presupposes transitivity would block Plantinga's externalism, leaving no choice but to be an internalist.
In this way it could easily happen that what generates a philosopher's internalism in Plantinga's sense is not deontologism, or not only deontologism, but the exclusion, however unwitting, of non-transitive forms of inferential justification at these deep levels of fundamentality. Both diagnoses could be correct, one in terms of the transitivity presupposition, the other in terms of deontologism. In many philosophers, the deontologism might be the dominant constraint, in others it might play little or no role (especially if no deontological conception of justification provides a good reason for endorsing internalism, as Bergmann (2000) argues). Also, a number of philosophers may have a further motive for endorsing internalism, as we see toward the end of �8, namely a way of exalting human being over all the rest of nature.
Furthermore, there is a sense in which the diagnosis in terms of transitivity is somewhat more basic than Plantinga's. The transitivity diagnosis applies not only to the species of internalism Plantinga targets, which is epistemic, but to the genus that includes both this species and those outside epistemology. It explains what generates, or at least rationalizes, the general tendency among many philosophers to concentrate on what is graspable by unaided reflection about language, concepts, intuitions, practices and more, even when they are not deontologists, and even when they are doing metaphysics or ethics rather than epistemology.
So long as we're talking about Plantinga, it is worth recalling a familiar tension between internalism and anti-naturalism. Plantinga, perhaps surprisingly, now declares that he is a naturalist; his view, he says, "indisputably falls under the rubric of naturalistic epistemology,"(45) indeed of "a radical naturalism," according to which "the very sort of normativity involved in warrant" is to be found in descriptive psychology, with its (implicit) generalizations about the "properly functioning human organism in an appropriate environment."(46) This appeal to an external descriptive discipline is anathema to anti-naturalists, and not only because they think that the crucial normativity of the epistemic would necessarily be eliminated in any descriptive approach. In addition, they argue that any appeal to external descriptive affairs at this level of fundamentality must be circular; any such appeal would be referring to such affairs as evidence or warrant, which would be to presuppose the very relation of evidence or warrant one was trying to justify or recommend.
This circularity argument against naturalized epistemology is old, entrenched, and widespread. To take just one contemporary example, consider Harvey Siegel's rejection of Ronald Giere's proposal for a naturalistic epistemology of science:
any answer to the question of the relationship between evidence and a justified theory, if arrived at scientifically, would depend upon exactly the same relationship between it and the evidence for it as it recommends for the relationship between any justified theory and the evidence for it. Because these general questions about the epistemology of science cannot be answered naturalistically without begging the question, they cannot be so pursued.(47) The idea is that (i) any appeal to an external description as evidence about the evidence relation would be part of what justifies the epistemologist's claim about the nature of the evidence relation, and (ii) the epistemologist's claim about the nature of the evidence relation would be presupposed by, hence part of what justifies the appeal to the external description as evidence about the evidence relation; hence (iii) the epistemologist's claim about the nature of the evidence relation would be part of what justifies itself, an egregious begging of the question at issue. But once again the charge of question-begging follows only if the needed relation of partial inferential justification is P-transitive.
Another measure of internalism's pervasive influence is how Descartes and many of his descendants, under the unwitting yet self-imposed exclusion of any philosophically relevant non-transitive inference, tend to concentrate on the cogito and its structures. For the cogito is accessible to the attentive thinker without abductive or other non-transitive inference to something external that might explain the characteristic structures of the cogito or the Cartesian theater. By contrast, Daniel Dennett's Consciousness Explained(48) amounts to one long series of inferences to the best explanation of why the lived features of the Cartesian theater are as they are; such features are to be explained in terms of what Descartes would call a kind of "hidden constitution," which he denies mind can have. The internalist tendencies Dennett targets are enabled if not generated by the transitivity presupposition, which enables the circularity arguments that discourage the search for external constraints on the mind's activity, indeed discourage the very idea. External constraints are likely to be found, and belief in them justified, only or largely by means of abductive or related non-transitive inference, since any attempt to justify belief in them by means of transitive inference will likely be viciously circular. Among such external constraints would be a mind-independent world about which some of our beliefs are true - correspondence true - or a world in which there are mind-independent samenesses or kinds or universals which our concepts might either track or fail to track.
Once doubt is cast on external constraints, the obvious place to look for constraints is in the mind itself. Leibniz claims to find them, or some of them, in rules for constructing experience of space and time, as subsequently in his own way did Kant (though both may have been anticipated by Descartes, for whom ideas may function as rules). Influenced by transformations of Kantian themes, many later philosophers regard concepts as but rules for organizing experience or for constructing sameness and difference in a world that would not otherwise have them. Leibniz and Kant regard the rules as necessary, unchanging and innate, while many of their descendants regard them as contingent, changing and cultural. We can make them up and we can decide which ones to adopt. Or they can just happen. Rules as changeable internal constraints figure importantly in the philosophies of Carnap, the later Wittgenstein, J. L. Austin and in several philosophies of mathematics (especially in Wittgenstein's and in formalism and intuitionism). Carnap explicitly requires transitivity of reducibility for the purposes of a constructional system, and regards external questions about the system or framework not as questions of fact but as practical questions, "a matter of a practical decision concerning the structure of our language."(49) In all these philosophies, the supposed internal constraints are terminal, in that they can be neither justified nor criticized, as regards whether they might conform to relevant aspects of the world, by inference from anything further - anything to which we do not have the requisite special access.
So too for reason itself: "We cannot so much as raise the question of the reliability of reason," says Plantinga, "without taking the reliability of reason for granted.... [O]ne cannot sensibly try to determine whether reason is trustworthy, before relying on it, since one has no recourse but to rely on it in trying to make that determination."(50) This is a foundationalist speaking, but any of the other terminal philosophers would say much the same. For example, among the sociologists of knowledge, we find Barnes and Bloor confidently asserting that
justifications of deduction themselves presuppose deduction. They are circular because they appeal to the very principles of inference that are in question. In this respect the justification of deduction is in the same predicament as the justifications of induction which tacitly make inductive moves by appealing to the fact that induction 'works'. Our two basic modes of reasoning are in an equally hopeless state with regard to their rational justification.(51)
Like Plantinga and anyone else persuaded by the relevant circularity argument or arguments, Barnes and Bloor presuppose that the inferences we could plausibly use to justify or even criticize reason, or to argue for or against its reliability, must be transitive. Thus one might wonder whether Plantinga's presupposition of transitivity here is inconsistent with his tacit use, noted above, of non-transitive abductive inference in arguing that his notion of warrant is actually realized. But there is an inconsistency here only if abductive inference could be used without vicious circularity to determine whether reason is trustworthy. That it could be is the gist of much of the next section. Even so, there will remain some logical wiggle-room for a Plantinga, and he rarely needs much. Perhaps he could come up with non-ad hoc reasons why non-transitive inference may legitimately be used in connection with warrant but not in connection with the reliability of reason. Or he might argue that his argument for the impossibility of determining the reliability of reason does not after all presuppose transitivity of the relevant forms of inferential justification.
Sociologists of knowledge like Barnes and Bloor differ from foundationalists only as regards whether the terminus that is reason has some non-inferential or immediate justification (perhaps a priori) - as most foundationalists believe - or instead must be taken either on faith or not at all except relative to some language game or practice of reasoning. Nietzsche saw this, as had Kierkegaard, and concluded not only that "The world appears logical to us because we have made it logical,"(52) but that (as William Bluhm puts it), "On Descartes' assumptions, to embrace reason - as an objective and universal order beyond the self - must be an act of faith."(53)
As Bluhm also remarks, "This line of thought opens up the possibility that it is our fiat that creates the world." But of course the possibility opens up earlier, as soon as one concludes that there must be fundamental matters beyond which there is no noncircular argumentative recourse, no further reason-giving. For this conclusion is enabled if not driven by the transitivity presupposition, and the transitivity presupposition also encourages the internalism that leads so many philosophers to look for the termini in the cogito, in our understanding or language, or in their structures; and all these can appear unconstrained by anything external, hence could be the result of our fiat. Bluhm says in connection with Nietzsche that "The 'Overman' would ... know that no objective order of the world is given; she would have to create order from her own will." Indeed. But it needs to be added that the Overman would "know" this - would have reason to believe this - only if the forms of inference by which one might find an objective order, or justify belief in it, are all transitive (unless of course there is some other way to restore the lost circularity arguments, contrary to Chapter 3).
By the same token, the conclusion that order would have to be created from the Overman's own will follows only granted the transitivity presupposition. The reason is that requiring transitivity excludes the one customary and plausible way of going external, namely inference to the best explanation. Otherwise, deprived of such non-transitive means of reason-giving, the self is left to brood on itself and on what is accessible to a supple phenomenology or analysis of the-world-as-seen-from-within. In such a mood - and stimulated further perhaps by Descartes's claim that there is nothing which truly belongs to me except the control of my volitions(54) - one may be struck by a kind of primacy of the will, according to which it is the will's self-sufficient, self-confident freedom to doubt received theories, concepts, rules and vocabularies - a freedom that creates a space, breathing room, wherein the will can replace the received detritus with something else. The only constraints are genealogical; shatter them and transvaluation can then follow, where the new values, like the old, conform to nothing external, nothing independent of the valuing self. Then there is Rorty: "the only thing we can be certain about is ... our own desires.... The only cosmology we can affirm ... is our own world picture, our own way of setting things up for manipulation, the way dictated by our desires."(55) We are not to be bullied by any so-called transcendent universals, kinds, truths or values, or by any other surrogates for a dead God.
Philosophers today largely reject Cartesian and other foundationalism, whether by way of rejecting clarity and distinctness or by way of rejecting a priori knowledge, certainty, immediate justification, ahistorical canons of logic and evidence, philosophy as the mirror of nature, or any metaphysics of presence. Despite these manifold differences, most such philosophers are just as terminal as the foundationalists they reject. For they assume that with regard to the philosophically deepest matters, there can be no inferential justification - or critique - on pain of question-begging circularity. Strip away all differences of tradition, style, method, language, emphasis - from Continental to pragmatic to the most obtuse analytic empiricist - there remains the shared conviction that pure philosophy is terminal, a conviction which rests on arguments that presuppose that the forms of inferential justification in pure philosophy are transitive.
The transitivity presupposition figures as well in the broadly internalist stance of much modern and post-modern philosophy, according to which nothing is hidden and the metaphysician's "craving for depth" or externalist theory or explanation is impossible, pointless or pernicious. There is also a link here with post-modernist painting and architecture, which are distinguished by their "depthlessness" and emphasis on the facade.(56) There is a link too with political theories that focus on the surface of political life, not some underlying social contract or other founding principle.(57) �7. Using Language to Get Outside Language
You may well object that realists owe us far more than negative critiques of anti-realist arguments, including the crucial circularity arguments and their presupposition of transitivity. In addition, realists owe us some transparently noncircular positive account of how our concepts or words - or enough of them, enough of the time - track objects and universals or kinds or samenesses found not made, how and in what sense truth can be a matter of correspondence, how might one "ground" inductive and deductive reason, including above all the law of non-contradiction, and more.
Yes, certainly. The realist must offer not only critiques of anti-realist arguments, but a positive account of the needed relations between the world and our concepts, language, theories, practices of reasoning, whatever. On the other hand, developing such an account with the thoroughness required goes beyond the scope of this book. Instead, I'll content myself with sketching an account of how one might advance non-question-begging reasons for thinking that the law of non-contradiction conforms to a structure of the natural world, contrary to Nietzsche's "The world appears logical to us because we have made it logical." The point is not that the account is true, not even that it is the only initially plausible path to attempt. Rather, the point is that the account can be constructed without being viciously circular. Here is one way such an account might go, starting with a question of philosophical method.
It has been some time since philosophers dared to pit their intuitions and analyses against scientific theories of matter, space, time, the origin of species, gender differences and more. Not only have philosophers frequently proved spectacularly wrong about such matters - so have scientists - and not only have they frequently been wrong-headed, from Aristotle on gender to those brave souls who dismissed time travel as impossible, contrary to the twins effect in special relativity and the possibility of closed time-like curves in general relativity. In addition, partly in reaction to this history of lost territories, philosophers have tended to develop an image of themselves and their enterprise as largely independent of whatever the sciences might turn up. The philosopher's exclusive expertise is widely presumed to concern phenomenological and conceptual issues, including, perhaps above all, issues about language and its legitimate uses and limits. The philosopher's focus is on terminal matters that can be grasped by kinds of reflection that have no need to consult the sciences for guidance let alone correction.
Furthermore, the results of such reflection are deemed prior to scientific investigation, in a crucial sense - not necessarily a priori, since it is often by way of a kind of historical and/or sociological phenomenology or analysis of lived experience or consciousness or meaning or concepts that the philosopher is to proceed, but prior anyway to whatever results the sciences might produce, and not refutable by them; philosophy checks the tools science merely uses. It is in this spirit that Wittgenstein says, "Darwin's theory has no more to do with philosophy than any other hypothesis in natural science."(58) Within the exclusive province of philosophy our analyses and intuitions reign supreme.
The connection between this broadly internalist stance and the transitivity presupposition, as seen, may be understood in terms of the kinds of reasoning the presupposition excludes, a special case being abductive inference to conditions that might discipline philosophers' armchair reflections. By the same token, and anticipating the further argument of chapters ahead, rejecting the transitivity presupposition enables us to ask, without vicious circularity,
What happens when we consider knowledge, concepts, language and inference not from the point of view of how they seem to us on considered reflection from within, but from the point of view of how they appear from without, say to biological science? The relevant biological science is the nonreductive, holistic biology of historically evolved living organisms in relation to their normal environments and to each other.(59) For example, consider the honey bee, and suppose we want to understand the role of bee dances in the life of the hive, which role, biologists theorize, contributed to the evolutionary success of the bees. Careful observation and experiment show that each dance is a complex that consists not only of wing beats, steps and thoracic waggles having specific frequencies and orientations, but of bits of nectar and pollen that adhere to the dancing forager bees; the complex tells the waiting bees about the distance and direction to nectar of a certain kind, quality and quantity. That is, dance-complex variations map onto (correspond to) specific combinations of direction, distance, kind, quality and quantity of nectar, the mapping being a necessary causal factor in the waiting bees' being enabled to find the nectar.(60) This in turn provides significant support for the hypothesis that the mechanisms which produced past dance complexes that did map, often enough, were selected for (or at least the genotypes responsible for these mechanisms were selected for); granted the hypothesis, one of the things the complexes are for - one of their functions - is to map onto direction, distance, kind and quantity of nectar, even when a particular combination of these happens to be new under the sun.(61)
Note that the bee dances involve a kind of intentionality - proto-intentionality, if you like, but intentionality nonetheless. For traditionally, the intentional is what is supposed to stand in relation to something else - something which it intends or means or is about or is meant to do - even if the something else does not exist or never happens. The bee-dance complex displays intentionality in this sense. An aspect of the dance, in virtue of a natural-selective history of the mechanisms that produce it, has the function of - is for - standing in relation to something else, say to peach-blossom nectar at a certain distance and direction. It is supposed to stand in this relation even if somehow the nectar happens not to be peach-blossom or not to exist.(62) The same can be said of indefinitely many other signing behaviors - the call vervet monkeys use for pythons, their call for martial eagles, their call for leopards, and so on. Note also that the signings need not map "reliably," even in the weak sense of mapping more often than not; they need only have mapped often enough, in the appropriate environments, to enhance survival and reproduction. Externalism does not require reliabilism, contrary to many epistemologists' tendency to equate the two.
To be sure, animal signing behaviors, systematic and articulate though some may be, are a long way from language, and their intentionality falls short of fully human intentionality. But it is hardly a new thought that many human abilities have ample anticipations in the prehuman. Why not intentionality and language, and in particular how certain aspects of language seem to be determinately about, and some of them to map onto, affairs in the world? Irrealists must stop this entering wedge from biology if they want to preserve their characteristic ideas about language and avoid another lost territory. This is not at all easy to do, as we see next.
Meet Wittgenstein of the bees. Wittgenstein of the bees is famous for his philosophical method, a kind of bee-dance linguistic phenomenology. Philosopher bees are to reflect on what it makes sense to dance; phenomenology is grammar.(63) What it makes sense to dance is a matter of what we bees would dance in light of our intuitions as competent dancers. If we bees want to get clear about what can sensibly be danced, and why, we should investigate our bee-dance practice, not anything external to it; philosophy is not to offer any theory. As Wittgenstein says, "grammar is not indebted to reality. Grammatical rules first determine meaning (constitute it) and are therefore not responsible to any meaning and are to that extent arbitrary."(64) The grammar is not imposed on us bees by reality or by any superior necessity, but by our conventional language games. We cannot bee-dance describe a pre-existing situation and show it forces us to adopt a particular grammatical rule, since to do so would require us to use bee dances to get outside bee dances.
Wittgenstein of the bees not only gets it wrong. Given his internalist stance he is bound to. We humans, from our biology-informed viewpoint external to the bees, can understand how the bee-dance grammar is indebted to reality, even imposed by such reality, in the sense that the rules, including the mapping rules, are not only determined by external affairs in a natural-selective history, but determined in such a way that the received mappings - of aspects of the bee-dance complex onto certain affairs - are what enabled the bees to adapt their activities to affairs in the world beyond the veil of bee dances. The grammatical rules are not arbitrary but designed by selection for an environment that has certain fundamental pre-existing features, including distance and direction - and the nectar's being not both peach-blossom and apple, an instance of non-contradiction. The world does not appear logical because bee-dance grammar made it so, but because independently it is so.(65) That the world is logical, in the relevant sense, best explains the relevant features of the grammar, not the other way around.
We can also understand, from our abductive biological stance, that bee-philosophers of meaning who reject what Wittgenstein calls "theory" - which includes rejecting inference to the best explanatory theory of certain key features of language - are unlikely so much as to guess what their situation is. And we can also see how the transitivity presupposition lies behind the insistence by Wittgenstein of the bees that we cannot use bee dances to get outside bee dances, or by a Nietzsche of the bees that the world only appears logical. Their internalism is enabled by the transitivity presupposition, if not driven by it, because the presupposition excludes abductive or other non-transitive inference to something beyond the veil of bee dances which could account for key features of bee-dance grammar.
But what about the signing behavior that is language? Recently evidence has been accumulating to the effect that "the ability to use a natural language belongs more to the study of human biology than human culture. It is a topic like echolocation in bats or stereopsis in monkeys, not like writing or the wheel.... Language has been shaped by natural selection."(66) Contrary to David Premack and those we may take him as speaking for, it is no longer plausible to suppose that "Human language is an embarrassment for evolutionary theory because it is vastly more powerful than one can account for in terms of selective fitness."(67) Our species has been characterized from its origins by toolmaking; by complex social relations, essential for cooperation; and by a folk biology of stunning sophistication. The effectiveness of all of these is enormously amplified by language. "Devices designed for communicating precise information about time, space, predicate-argument relations, restrictive modification, and modality ... [plus] socially relevant abstract information such as ... beliefs, desires, tendencies, obligations, truth ... hypotheticals, and counterfactuals" are hardly wasted efforts for a gatherer-hunter organism so deeply dependent on complex social relations, tool-using and more. Recursion is especially useful, allowing its user to "specify reference to an object to an arbitrarily fine level of precision" and to express beliefs about the intentional states of others.(68)
Furthermore, evidence from empirical linguistics testifies to the existence of certain features common to all languages. These include major lexical and phrasal categories, phrase structure rules, rules of linear order, anaphoric elements, and mechanisms of complementation. For example, for any language, if it has the verb before the object, as in English, it will also have prepositions; if it has the verb after the object, as in Japanese, it will have postpositions. Indeed a language will be such that its phrases will be all either head-first, as in English, or all head-last, as in Japanese, a sweeping regularity - in fact a universal - that makes a language far easier for the child to acquire.
In addition, every language is characterized by a phrase-structure grammar in which, at some deep level, the sentences of the language consist of a noun phrase and a verb phrase. In this sense, the sentences are subject-predicate in form. Also, the function words of a language include an auxiliary word (or other device) for negation, which, like the other auxiliary words, has to do with the truth of the sentence as the speaker conceives it. Significantly, people add new content words to the language all the time, but the function words form a closed club that resists change. Thus every language employs sentences that at a deep level are subject-predicate in form and subject to negation. This is just one way the specific expressive abilities of a language turn out to form a well defined set that persists through enormous surface diversity. There is in addition strong evidence for an extensive genetic component in one's grammatical ability, including documented genetically transmitted syndromes of grammatically defective speech.
Finally, many of our sentences, like bee dances, have among their functions certain biological functions or purposes; and novel sentences - sentences never produced before - have derived functions. The sentences produced by a Pleistocene gatherer-hunter scout on returning to the band - including sentences never produced before - have the purpose or function of "adapting" the listeners to certain world-affairs or conditions, such as food or shelter over the horizon. This enables the listeners to pursue their purposes in line with just those conditions, thus enhancing their chances of success. The sentences perform this function by virtue of mapping onto certain affairs in conformity to certain mapping rules. The mapping rules, not unlike those for bee dances, are rules in conformity to which "a critical mass of sentences have mapped onto affairs in the world in the past, thus producing correlation patterns between certain kinds of configurations of sentence elements and certain kinds of configurations in the world."(69) The correlation patterns enable the listeners to adapt their activity to the configuration or world-affair thus mapped. "Truth rules are rules that project, from the parts and structure of sentences in a language, the conditions under which these sentences would be true."(70)
Granted all this, though only for the sake of argument, consider the deep-structure subject-predicate form of sentences whose function is to map affairs in the world. The form appears to conform to a structure of the world in which language evolved, much as the syntax of bee dances conforms to a structure of sameness and difference, found not made, in the environments in which they evolved. The subject-predicate form conforms to a structure of substances that not only instantiate universals from among ranges of naturally admissible universals, but instantiate none that are contrary to each other in this world (as a matter of natural law).(71) The best explanation of why the form is present in all languages is that language evolved to adapt us to our environment, and our environment has this structure.
Furthermore, it looks as though a true sentence which contains negation - say, "Snow is not green" - maps not the absence of a world affair (the absence of snow's being green) but the obtaining of some contrary affair (snow's being some other color, say white), so that negation turns a definite predicate into an indefinite one (indefinite in the way "a dog" or "some bat" is).(72) Or rather let us suppose so, for the sake of argument. Then the Law of Non-Contradiction (LNC) "reflects a thoroughly natural structural principle rather than an a priori metaphysical or rational principle, or the workings of our language ... or ... of a constructive Kantian understanding."(73) The best explanation of why the law occurs (in the relevant sense) in all languages is that language evolved to adapt us to our environment, our environment has a structure of substances instantiating universals in such a way that no two universals instantiated by the same item are contrary to one another in this world (as a matter of natural law), and LNC helps to adapt us by conforming to this structure.
On this account, therefore, knowledge of the law of non-contradiction, and of the kind of natural necessity and possibility involved, is a posteriori natural knowledge enabled by abductive inference. So too is knowledge of logic, though of course only so far as the kind of logic we are talking about is not an abstract a priori science, but a study of certain real relations among concepts that track samenesses and differences in the world.(74)
Those who are not already so inclined - and they may be legion - are unlikely to be persuaded by any of this to accept some such account of the law of non-contradiction. But of course that was not my intent. As I emphasized at the outset, the point is not that the account is true, not even that it is the only initially plausible path to attempt. Rather, the account can be constructed without being viciously circular. For suppose that LNC is used or presupposed, as presumably it must be, in the course of developing the argument and evidence involved in the inferential justification of the conjunction M of the evolutionary and linguistic matters that are to be explained - those sketched above - including in particular the matter of LNC's occurring (in the relevant sense) in all languages. Then (i) LNC is part of what inferentially justifies M, even if LNC is not itself a belief or belief-like. Suppose further, as lately suggested, that the best explanation of why LNC occurs or is reflected in all languages is that language evolved to adapt us to our environment, our environment has a structure of substances instantiating universals in such a way that no two universals instantiated by the same item are contrary to one another in this world (as a matter of natural law), and LNC helps to adapt us by conforming to this structure. Accordingly, by way of inference to the best explanation, (ii) M inferentially justifies LNC (in the sense of justifying belief in its conformity to a structure in the world). Thus we have (i) LNC is part of what inferentially justifies M, and (ii) M inferentially justifies LNC. It would follow that LNC is part of what inferentially justifies itself only if the type of inferential justification involved here were transitive. But it is not transitive, as already noted toward the end of Ё1.2 and1.3 and as argued in some detail in the next chapter.
In this way one could construct a non-question-begging series of reasons for believing that the world does not appear logical to us because we have made it logical, but because in the relevant sense it is. To insist on the contrary would be as ludicrously self-important as a Nietzsche of the bees insisting that the world appears logical because bee-dance grammar has made it so. �8. No Higher Viewpoint?
The clamor from the back of the room is the irrealist's: Hey, wait, your sketch speaks of so-called pre-existing features, and of a so-called reality, which surely is question-begging. For such features and reality are such - or we can intelligibly say they are such - only according to the categorizations of the language or sign system we humans employ when we do biology. How do we humans get outside our sign system? However it may go with the bees, when it comes to language, "There is no outside," as Wittgenstein puts it; "outside you cannot breath."(75) There is no God's-eye view, no higher viewpoint. In order to apply to ourselves the moral of the bee-dance case, we would have to use language to describe the so-called pre-existing affairs to which our language allegedly evolved to adapt us. Yet we cannot possibly justify our descriptive or factual language by appealing to facts which can only be stated in it. Putnam is right that it makes no sense to say that our concepts answer to a ready-made world. Categories and concepts are but rules for introducing sameness and difference into a world that would not otherwise have them.
No higher viewpoint? This move, like so much of Western philosophy, exalts human being over the rest of nature. Other animals' sign systems may conform to rules imposed by natural-selective pressures to adapt to a pre-existing structured reality, a ready-made world. Ours does not. Other animals' categories, concepts and signing behaviors may track samenesses that exist independently of their sign systems. Ours do not, or at least we cannot intelligibly say that they do. For it is only relative to language that it makes sense to say that these very samenesses are there for the animals' categories and concepts to track. Indeed it is only relative to language that it makes sense to say there are the samenesses or kinds we call the animals. If, as Putnam would have it, there is no ready-made world, then there are no ready-made spotted owls.
This irrealist privileging of human being follows from an internalist account applied to us combined with an externalist account applied to the animals; we are not so very far from Descartes on "thoughtless brutes." The resulting picture, modulated by historicist and linguistic transformations of Kantian themes, is a picture of a free play of sameness-creating categories and concepts for us, sameness-imposed categories and concepts for all other creatures; conceptual relativism for us, realism for them; world-making for us, a ready-made world for them, indeed for them a world made by us. Putnam lets the cat out of the bag when he says, "Kant's Copernican revolution means we are in charge." There may be no more revealing a remark in all of modern and post-modern philosophy.(76)
Because the privileging of human being is furthered by the internalism, one suspects that the privileging is a big part of what motivates the internalism. So too may it be a big part of what motivates presupposing the transitivity of inferential justification, which transitivity, as seen, enables the internalism. The transitivity enters with the claim that however it may go with the bees, for us and our language there is no outside, no higher viewpoint, since we cannot possibly ground our factual language by appealing to facts which can only be stated in it, on pain of circularity; you can't use language to get outside language. One suspects that the internalism and the irrealism, together with the transitivity that enables them, amount to special pleading for the point of view of our species. After all, irrealism means we are in charge.
Thus irrealism about kinds, categories or creatures seems a bad idea from the point of view of environmental ethics and much theology, which tend to reject any such privileging of human being. And yet many environmental ethicists and theologians embrace philosophies that prove irrealist - Heideggerian, for example, or conceptual relativist, or pragmatist warranted assertability, or form-of-life fideism. They embrace these philosophies largely in reaction against realism, or rather against what they take to be objectionable features of realism - its totalizing essentialism, as they see it, or its supposed reductivism and incompatibility with normativity and enchantment in the world. But to repeat what was said toward the end of �4, even though some traditional realisms may entail these evils, realism as such does not; irrealists have set up a scarecrow.
Of course the considerations outlined in this and the previous section on behalf of a positive alternative to terminal philosophies are unlikely to persuade anyone not already sympathetic and informed. But again, that is not the point, which is to sketch how one might plausibly advance non-question-begging reasons for thinking that the law of non-contradiction conforms to a structure of the natural world, rather than the other way around. For the time being, it is enough to have exposed further, by contrasting it with a positive alternative, the deep-structural terminal philosophy that owes so much to the 16th-Century dialectic in which it is still mired, and to rattle the cage of those still in thrall to the internalism and the irrealism enabled by terminal philosophy and by the fallacious transitivity presupposition that underlies it.
1. Annis (1978) is a linear contextualist, as apparently is Williams (1996) insofar as his contextualism presupposes what on p. 114 he calls formal foundationalism. Cf. Brady (1998).
2. E.g. Henderson (1994), or so I interpret him. Cf. Brady (1998), n3.
3. Williams (1996), 64.
4. Cling (forthcoming) argues that there are cases in which this can happen without vicious circularity.
5. Rorty (1989), 73.
6. Pears (1988), 14; cf. 255, on rules.
7. Quoted in translation by Noë (1994), 23, from Wittgenstein's Big Typescript and Philosophical Grammar.
8. Hacking (1975), 187, who of course means "knowledge" only insofar as the notion makes sense.
9. Posterior Analytics, Book I, Ch. 3.
10. Cf. Popkin (1979) on the predominant role of skepticism in the 16th and 17th Centuries, and on the impact of the revived texts of Sextus Empiricus in generating the Pyrrhonian crisis.
11. Cf. Popkin (1951), (1959).
12. Post (1996a) calls terminal philosophers "structural foundationalists," which can be misleading. Just as structural engineers are engineers, one would expect structural foundationalists to be foundationalists, whereas by my usage many of them are relativists, fideists, skeptics or other anti-foundationalists.
13. Sanford (1975), Post (1980), (1992).
14. Why (1)-(4) entail (6) may be left as an exercise for the reader, or one may consult Black (1988).
15. This too may be left as an exercise, or cf. Post (1996a), 243, Post (1980), and Black (1988). I realized only recently that Sanford (1975) was the first to note in print that regress arguments require transitivity of the relation at issue, here the relation of inferential justification (see also Sanford (1984)). Sanford and I have engaged in hilarious negotiations over my proper punishment, ranging from giving my next lecture in sack cloth and ashes, to walking to the next APA meeting on my knees, to ever more imaginative acts of contrition. My plea that the title of his 1975 paper was insufficiently informative ("Infinity and Vagueness") has been of no avail.
16. As does Plantinga (1993a), 68-69, 74-77.
17. Post (1980).
18. 18HR, I, 231; CSM, I, 203; AT, VIII, 16.
19. It is presupposed twice, once to get from "x is part of what justifies y, and y justifies z" to "x is part of what justifies z," and again to get from "x is part of what justifies z, and z justifies x" to "x is part of what justifies x."
20. See also Post and Turner (2000).
21. Hooker (1987), 332-333, uses the term "God's Eye argument" for a closely related bit of dialectic, arguing rightly that contrary to Putnam and his ilk, realism does not require a God's-eye stance. But whereas Hooker sees God's-eye-view arguments as arising from an "overly strong attachment to a verificationist semantics," I see them as arising far more fundamentally from the transitivity presupposition that lies behind terminal philosophy, which in turn lies behind not just verificationism but most if not all other forms of irrealism.
22. See especially Moore (1899). See also Passmore (1957), pp. 205-211, to bits of which I have helped myself in this paragraph.
23. Moore (1959), 27.
24. Putnam (1992), 123.
25. Rorty (1979), 294.
26. Kuhn (1970), 206
27. Levine (1993), p.13. The underlying look-see argument against realism is related to what Searle (1995), 173-174, calls the Ding an Sich argument. But his treatment of the Ding an Sich argument affords neither diagnosis nor critique of what enables the shibboleth that you can't use language or scheme to get outside language or scheme, namely the transitivity presupposition.
28. Derrida (1978), 280-281.
29. Putnam (1978), 126-127, 134.
30. Putnam (1990), 28.
31. Rorty (1993), 443.
32. As argued at length in Post (1987), (1990), (1991).
33. Wolterstorff (1990), 55, 63.
34. Alston (1989), 326-329.
35. Alston (1989), 342-343.
36. Stroud (1984), 211, 215, 242.
37. As urged in detail by Tim McGrew, Lydia McGrew, and Jeff Tiel, in correspondence, for which I am deeply indebted. They might deploy this relation in order to argue that the would-be "justification" of R must be circular after all. We'll see about that in Ch. 4 ("Regress Revisited"), and also about a relation E of positive evidence proposed by McGrew and McGrew (2000), which Post and Turner (2000) argue is not transitive after all.
38. Darwin (1962), 476. See Lipton (1991) for an accessible account of abductive inference and a defense of it against recent criticism.
39. Plantinga (1993a), 182; see also 5-25.
40. That they were thus motivated is consistent with the argument in Bergmann (2000) that "no deontological conception of epistemic justification provides a good reason for endorsing internalism" (my italics).
41. Plantinga (1993a), 183.
42. Plantinga (1993a), 5.
43. Plantinga (1993a), 64, 211-215; Plantinga (1993b).
44. Plantinga (1993b), ix.
45. Plantinga (1993a), 215.
46. Plantinga (1993b), 46.
47. Siegel (1989), 369.
48. Dennett (1991).
49. Carnap (1967), 8; (1952), 210. Alston (1989) likewise turns pragmatist about basic principles, such as the principle of the reliability of sense experience lately noted in �5.
50. Plantinga (1993a), 103.
51. Barnes and Bloor (1982), 41-42.
52. Nietzsche (1968), 625, 484, 521.
53. Bluhm (1996), p. 327.
54. HR, I, 401; CSM, I, 384; AT, XI 445.
55. Rorty (1991), 29.
56. Jameson (1984).
57. Cooper (1993), 57.
58. Wittgenstein (1961), 4.1122. He retains the theme in the Investigations, Wittgenstein (1958), II xii. Cf. Pears (1988), 450-459.
59. On this use of "nonreductive" see Post (1995). An account of a property P is nonreductive when according to the account P need not be equivalent to some lower-level property or compound of lower-level properties even when the matter of whether an individual x has P is determined by some lower-level conditions or other.
60. Kirchner and Towne (1994) provide an accessible introduction with references. None of this is to suggest that the bee dances form a language. Cf. Millikan (1984), 40ff, 96-97; Wenner (1990). I am indebted to Richard Burian and Karen Neander for calling attention to problems with an earlier presentation of this example.
61. Cf. Brandon (1990) for a rigorous account of adaptation and what one is for; Millikan (1984), (1993), with regard to cases new under the sun, including those nowhere encountered in the organism's evolutionary history; and Post (1996b), (1998).
62. Millikan (1984).
63. Cf. Noë (1994), 20; Pears (1988), 455.
64. Quoted in translation by Noë (1994), 23, from Wittgenstein's Big Typescript and Philosophical Grammar.
65. On this account of the law of noncontradiction cf. Millikan (1984), Ch. 18.
66. Pinker and Bloom (1992), 451-452, which contains further references, and from which I draw much of the present paragraph and the next.
67. Premack (1985), 281-282. If Premack were right, then by the same token so too, absurdly, would the elephant's trunk be an embarrassment for evolutionary theory, according to Pinker (1994), 332-333..
68. Pinker and Bloom (1992), 482-483.
69. Millikan (1984), 99.
70. Millikan (1993), 231.
71. Millikan (1984), 239-296.
72. Millikan (1984), 228.
73. Millikan (1984), 257; see also 221-229, 297-310.
74. Millikan (1984), 273. See also Sher (1999).
75. Wittgenstein (1958), I �3.
76. I have the remark firmly in my notes as occurring in Putnam, but alas not the location, which I have been unable to find. If the remark is not in Putnam, I apologize. On the other hand, it strikes me as so Putnamian that it must be there staring up at us from between the lines, and in any case it epitomizes an all-too dominant theme in Western philosophy after Kant.
Alston, William P. (1989). Epistemic Justification : Essays in the Theory of Knowledge (Ithaca: Cornell University Press) .
Annis, David (1978). "A Contextualist Theory of Epistemic Justification," American Philosophical Quarterly 15, 213-219. Reprinted in Empirical Knowledge: Readings in Contemporary Epistemology, ed. P. Moser (Totowa, N.J.: Rowman & Littlefield, 1986), 203-213. (Page references in the present text are to Moser.)
Barnes, B. and Bloor, D. (1982), "Relativism, Rationalism and the Sociology of Science," in M. Hollis and S. Lukes, (eds.), Rationality and Relativism. Cambridge, MA: MIT Press, 21-47.
Bergmann, Michael (2000). "Deontology and Defeat," Philosophy and Phenomenological Research, 60, 87-102.
Black, O. (1988). "Infinite Regresses of Justification," International Philosophical Quarterly, 28, 421-437.
Bluhm, William T. (1996). "Political Theory and Ethics," in Discourse on Method and Meditations in First Philosophy: René Descartes, ed. David Weissman (New Haven: Yale University Press, 1996), 306-329.
Brady, Michael S. (1998). "Can Epistemic Contextualism Avoid the Regress Problem?" The Southern Journal of Philosophy, XXXVI, 317-328.
Brandon, Robert N. (1990). Adaptation and Environment (Princeton: Princeton University Press).
Carnap, Rudolf (1952). "Empiricism, Semantics, and Ontology," in Semantics and the Philosophy of Language, ed. Leonard Linsky (Urbana: University of Illinois Press).
Carnap, Rudolf (1967). The Logical Structure of the World, translated by R. A. George (Berkeley: University of California Press).
Cling, Andrew (forthcoming). "Justification-Affording Circular Arguments."
Cooper, David E. (1993). "Postmodernism and 'The End of Philosophy'," International Journal of Philosophical Studies, 1, 49-59.
Darwin, Charles (1962). The Origin of Species (New York: Collier).
Dennett, Daniel C. (1991). Consciousness Explained (Boston: Little, Brown).
Derrida, Jacques (1978). "Structure, Sign, and Play in the Discourse of the Human Sciences," in Writing and Difference, trans. Alan Bass (Chicago: University of Chicago Press), 278-293.
Hacking, Ian (1975). Why Does Language Matter to Philosophy? (Cambridge: Cambridge University Press).
Henderson, David K. (1994). "Epistemic Competence and Contextualist Epistemology: Why Contextualism is Not Just the Poor Person's Coherentism" The Journal of Philosophy, 91, 627-649.
Hooker, C. A. (1987). A Realist Theory of Science (Albany: SUNY Press).
Jameson, Frederic (1984). "Postmodernism, or the Cultural Logic of Late Capitalism," New Left Review, 146.
Kirchner, Wolfgang H., and Towne, William F. (1994). "The Sensory Basis of the Honeybee's Dance Language," Scientific American, 270, 74-80.
Kuhn, Thomas S. (1970). The Structure of Scientific Revolutions, 2d ed. (Chicago: University of Chicago Press).
Levine, George (1993). "Looking for the Real: Epistemology in Science and Culture," in G. Levine, ed., Realism and Representation: Essays on the Problem of Realism in Relation to Science, Literature and Culture (Madison: University of Wisconsin Press).
Lipton, Peter (1991). Inference to the Best Explanation (London: Routledge).
McGrew, Lydia, and McGrew, Tim (2000). "Foundationalism, Transitivity and Confirmation," The Journal for Philosophical Research, 25, 67-82.
Millikan, Ruth Garrett (1984). Language, Thought and Other Biological Categories (Cambridge: MIT Press).
Millikan, Ruth Garrett (1993). White Queen Psychology and Other Essays for Alice (Cambridge: MIT Press).
Moore, G. E. (1899). "The Nature of Judgment," Mind, 8, 176-193.
Moore, G. E.(1899). Philosophical Studies (Patterson, NJ: Littlefield, Adams).
Nietzsche, F. W. (1968). The Will to Power, trans. Walter Kaufmann and R. J. Hollingdale (New York: Vintage Press).
Noë, Robert A. (1994). "Wittgenstein, Phenomenology and What it Makes Sense to Say," Philosophy and Phenomenological Research, 54, 1-42.
Passmore, John A. (1957). A Hundred Years of Philosophy (London: Duckworth).
Pears, David (1988). The False Prison: A Study of the Development of Wittgenstein's Philosophy (Oxford: Clarendon Press).
Pinker, Steven (1994). The Language Instinct (New York : Morrow).
Pinker, Steven, and Paul Bloom (1992). "Natural Language and Natural Selection," in The Adapted Mind: Evolutionary Psychology and the Generation of Culture, ed. Jerome H. Barker, Leda Cosmides and John Tooby (Oxford: Oxford University Press), 451-493.
Plantinga, Alvin (1993a). Warrant: The Current Debate (Oxford: Oxford University Press).
Plantinga, Alvin (1993b). Warrant and Proper Function (Oxford: Oxford University Press).
Popkin, Richard H. (1951). "Hume and Kierkegaard," The Journal of Religion, 31, 274-281.
Popkin, Richard H. (1959). "Kierkegaard and Scepticism," Algemeen Nederlands Tijdschrift voor Wijsbegeerte en Psychologie, 51, 123-141.
Popkin, Richard (1979). The History of Scepticism from Erasmus to Spinoza (Berkeley: University of California Press).
Post, John F. (1980). "Infinite Regresses of Justification and of Explanation," Philosophical Studies, 38, 31-52.
Post, John F. (1987). The Faces of Existence: An Essay in Nonreductive Metaphysics (Ithaca: Cornell University Press).
Post, John F. (1990). "Objective Value, Realism and the End of Metaphysics," The Journal of Speculative Philosophy, vol. 4, 146-160.
Post, John F. (1991). Metaphysics: A Contemporary Introduction (New York: Paragon House).
Post, John F. (1992). "The Infinite Regress Argument," in A Companion to Epistemology, ed. Jonathan Dancy and Ernest Sosa (Oxford: Blackwell).
Post, John F. (1995). "'Global' Supervenient Determination: Too Permissive?" in Essays on Supervenience, ed. Elias Savellos and mit Yalçin (Cambridge: Cambridge University Press), 73-100.
Post, John F. (1996a). Chapter on "Epistemology," Discourse on Method and Meditations in First Philosophy: René Descartes, ed. David Weissman (New Haven: Yale University Press, 1996), 236-271.
Post, John F. (1996b). "Post-Quinean Philosophical Investigations," in Quine and Wittgenstein, ed. Robert Arrington and Hans-Johann Glock (New York: Routledge), 252-279.
Post, John F. (1998). Review of Ruth G. Millikan, White Queen Psychology, for Philosophy and Phenomenological Research, 58, 233-237.
Post, John F. and Turner, Derek (2000). "Sic Transitivity: Reply to McGrew and McGrew," The Journal of Philosophical Research, 26.
Premack, David (1985). " 'Gavagai!' or the Future History of the Animal Language Controversy" Cognition, 19, 207-296.
Putnam, Hilary (1978). "Realism and Reason," in his Meaning and the Moral Sciences (London: Routledge and Kegan Paul).
Putnam, Hilary (1990). Realism with a Human Face (Cambridge: Harvard University Press).
Putnam, Hilary (1992). Renewing Philosophy (Cambridge, MA: Harvard University Press).
Rorty, Richard (1979). Philosophy and the Mirror of Nature (Princeton: Princeton University Press).
Rorty, Richard (1989). Contingency, Irony, and Solidarity (Cambridge: Cambridge University Press).
Rorty, Richard (1991). Essays on Heidegger and Others (Cambridge: Cambridge University Press).
Rorty, Richard (1993). "Putnam and the Relativist Menace," Journal of Philosophy, XC, 443-461. Sanford, David (1975), "Infinity and Vagueness," Philosophical Review, 84, 145-158.
Sanford, David (1984), "Infinite Regress Arguments," in Principles of Philosophical Reasoning, ed. James H. Fetzer (Totowa, NJ: Rowman-and-Allanheld), 93-117. Searle, John (1995). The Construction of social Reality (New York: The Free Press)
Sher, Gila (1999). "Is There a Place for Philosophy in Quine's Theory?" Journal of Philosophy, XCVI, 491-524.
Siegel, H. (1989), "Philosophy of Science Naturalized? Some Problems with Giere's Naturalism," Studies in History and Philosophy of Science, 20: 365-375. Stroud, Barry. (1984). The Significance of Philosophical Skepticism (Oxf U Press, 1984)
Wenner, Adrian (1990). Anatomy of a Controversy: The Question of "Language" Among the Bees (New York: Columbia University Press)
Williams, Michael (1996). Unnatural Doubts: Epistemological Realism and the Basis of Scepticism (Princeton: Princeton University Press).
Wittgenstein, Ludwig (1958). Philosophical Investigations, tr. G. E. M. Anscombe, 3rd edn., (New York: Macmillan)
Wittgenstein, Ludwig (1961). Tractatus Logico-Philosophicus, tr. D. F. Pears and B. McGuinness (London: Routledge)
Wolterstorff, Nicholas (1990). "Realism vs. Anti-Realism," in Reality in Focus: Contemporary Readings in Metaphysics, ed. Paul K. Moser (Englewood Cliffs: Prentice Hall), 50-64.
Home Place || firstname.lastname@example.org