Chapter 6 Toward a Realistic Rationalism
6 .1 The Path Back to Rationalism
6.1.1 Fregeand the Linguistic Turn The infl...
10 downloads
494 Views
6MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Chapter 6 Toward a Realistic Rationalism
6 .1 The Path Back to Rationalism
6.1.1 Fregeand the Linguistic Turn The influence of the linguistic turn , like that of other significant intellectual revolutions , continues even though the linguistic turn is now history. However , the nature of its influence has not been fully understood , because the linguistic turn is generally seen in terms of the linguistic doctrines expressed in the slogans and manifestoes of its leading figures, and those linguistic doctrines are no longer taken seriously. But the influence that the linguistic turn exerts on philosophy is a matter of the continuing ascendancy of the positivist , empiricist , and naturalist views behind those doctrines rather than a matter of the temporary ascendancy of the doctrines themselves. The linguistic turn is mistakenly thought of as a nonpartisan attempt to reframe philosophical questions in manageable linguistic terms, and, philo through linguistic analysis (of one form or another ), to eliminate ' s " Does . make no sense e. the that ( g , Heidegger sophical questions " . On the i.e. exists? exist because the Not the ) , , Negation , Nothing only surface, it was often such a nonpartisan attempt , but , at a deeper level , it was another of the partisan attempts , which occur in the history of philosophy from time to time , to cleanse philosophy of metaphysical positions taken to be the cause of all its ills . The linguistic turn steered philosophy away from the traditional rationalist enterprise of explaining new forms of positivism , synthetic a priori knowledge and toward " . naturalism , and empiricism The term linguistic " had little to do with " the deeper ends of the " revolution in philosophy . It expressed the means chosen for effecting change and reflected the perception that dictated the choice of means, the perception that the linguistic (especially , the semantic) foundation of the doctrine of synthetic a priori knowledge is the soft underbelly of traditional rationalism . It is ironic that it was Frege, the greatest rationalist and realist metaphysician of nineteenth and very early twentieth - century philosophy, who provided the first phase of the linguistic turn with the
178
Chapter6
linguistic and logical weapons for its assault on rationalist and realist metaphysics. In the first phase, the early Wittgenstein , Schlick, Carnap, and their Viennese followers took both their linguistic cue and their logico- semantic tools from Frege (and Russell), using them to criticize traditional philosophy - much of which Frege (and Russell) subscribed to - as cognitively meaningless, and, on the model of Frege' s Begriffsschrift, to develop a conception of an ideal language and the novel thesis that , as Ayer (1946, 57) expressed it , " philosophy is a department of logic ." In the second phase of the linguistic turn , logical empiricism itself came under attack in large part for its Fregeanism. The late Wittgenstein and Quine directed their criticisms against the Fregean semantics underlying the forms of empiricism and positivism developed in the first . Both saw Fregean semantics as the basis for the metaphysician ' s phase claims about synthetic a priori knowledge , and both used those criti cisms as a springboard for launching even more extreme forms of positivism and empiricism (see Kenny 1973, 113- 14; Quine 1961c). The late Wittgenstein continued to argue for a naturalistic and posi tivistic view of philosophy on which sentences about nature are the only meaningful sentencesand metaphysical sentences transcend the limits of language. Quine, eschewing positivistic naturalism , argued for a scientistic naturalism concerned with showing that much of traditional metaphysics is scientifically anachronistic speculation about the natural world . Quine ' s (1969a, 82) naturalism replaces the positivist idea of exposing metaphysical statements as nonsense with the idea that philosophy goes on within natural science, either as a form of metascientific analysis of the conceptual and linguistic side of the scientific enterprise or '' as a chapter of psychology and hence of natural science." In this section, we will examine both the role played by Fregean assumptions in the development of the philosophical positions in the first phase of the linguistic turn and the philosophical significance of the Fregean focus of Wittgenstein ' s and Quine ' s criticisms of those positions in the second phase. I will try to show that the preoccupation with Frege has made the linguistic underbelly of rationalism appear far softer than it is by linking rationalism to Fregean intensionalism , thus obscuring other, less vulnerable intensionalist ideas. As we shall see, an intensionalism that breaks radically with Fregean intensional ism can provide concepts of meaning and analyticity that are not open to criticisms directed against the semantics assumed in the first phase of the linguistic turn . These conceptions of meaning and analyticity are, moreover, too thin to be used for any positivistic purpose like those in the first phase. Thus, such an intensionalism has the potential to restore
Toward a Realistic Rationalism
179
explanation of synthetic a priori knowledge to its rightful place in philosophy . 6.1. TheVicissitudesof the SyntheticA Priori Passmore (1957, 157) writes : . . . in arguing that it is language that leads us astray or, again, in setting up the ideal of a perfect language which would not betray us becausein it every expression would have a fixed and definite sense, Frege, more than any other nineteenth century philosopher , anticipates the preoccupations of twentieth century positivism and its diverse progeny. To be sure, the metaphilosophical ideas with which Passmore credits Frege, together with the technical logical and linguistic machinery he invented and the important philosophical arguments he developed , provided the early Wittgenstein , Schlick, and other logical empiricists with the philosophical , logical , and semantic tools for their work . But Passmore's epitome of Frege as farsightedly anticipating twentieth ' century philosophy puts the wrong slant on Frege s relation to the revolution , missing his ambiguous role as the intellectual benefactor of the positivist , naturalist , and empiricist views that came to replace the traditional rationalist / realist view of philosophy as an autonomous a priori study of the most general facts about reality . It was Frege, the rationalist and realist, who made it all possible. ' Frege s logico mathematical innovations and his criticisms of Kant' s doctrines of analyticity and mathematical truth were the entering ' wedge for the early logical positivists attack on metaphysics, especially Kantian , neo- Kantian , Husserlian , and other forms of then current metaphysics. Before Frege, Kant' s criticism of Hume had stood as the bulwark against empiricistically and naturalistically inspired positivism . The vaguenessof Hume ' s notion of relations of ideas allowed Kant to explicate that notion as analyticity in his concept-containment sense. This linked mathematics with metaphysics as synthetic a priori knowledge ' . Since the former as well as the latter is outside both Hume s share , categories of relations of ideas and matters of empirical fact " they a common fate. If , as Hume advocates, metaphysics is consigned to the flames," then mathematics is too. This enabled Kant to argue that , since mathematics cannot be abandoned, metaphysics cannot be abandoned either. The task of philosophy must thus be the metaphysical one of explaining how synthetic a priori knowledge in mathematics, science, and philosophy is possible. Kant' s argument assumes that there is no other explication of the notion of relations of ideas that is preferable to his own notion of
180 Chapter6 analyticity . If there were another, one broad enough to bring mathematics under analyticity , it could cut the link between mathematics and metaphysics, thereby stripping metaphysics of the protection it receives from our unwillingness to abandon mathematics. This is precisely what ' Frege s explication of analyticity as truth based on laws of logic and definitions purports to do. Coupled with Frege' s criticism of the Kantian notion of analyticity , Frege' s logical explication of analyticity creates an entirely new situation for naturalists and empiricists generally. Rejecting Kant' s account of mathematics as synthetic a priori knowledge in favor of Frege' s account of it as analytic a priori knowledge , the ' logical empiricists could reject Kant' s reductio of Hume s positivism , his explanation of mathematical knowledge , and his rehabilitation of metaphysics. Dummett (1991, 111- 24) also casts Frege- specifically , the Frege of the Grundlagen- as the initiator of the linguistic turn . Dummett (1991, 111) claims that Frege' s significant philosophical move in initiating the linguistic turn was invoking the context principle to transform Kant' s about our knowledge of numbers into the linguistic question question about " how the senses of sentences containing numbers are to be fixed ." Dummett exaggerates the importance to the linguistic turn of this relatively obscure use of the context principle in the philosophy of mathematics. It was of significantly less importance than Frege' s general metaphilosophical ideas about the imperfections of language, his notion of an ideal language as a means for overcoming them, his criticism of the Kantian concept of analyticity as " unfruitful ," and his replacement of it with his own logical concept of analyticity . The logical empiricists seized on Frege' s concept of analyticity as showing that the Kantian concept of analyticity, due to its extreme narrowness, created the illusion of synthetic a priori knowledge which neo- Kantians , assorted rationalists , and Husserlian phenomenologists exploited to justify introducing special faculties such as intuition . Dubious appeals to intuition might have to be swallowed if Kantian analyticity were the only explication of the notion of relations of ideas, but , after Frege, there was no longer a necessity to swallow them. The early Wittgenstein and the logical positivists saw metaphysical explanations of alleged synthetic a priori knowledge as both misguided and unnecessary misguided because they falsely claim to provide a priori insights into the nature of things , and unnecessary because Frege had provided the means for an alternative , analytic , explanation of a priori . knowledge ' Frege s broadening of the category of the analytic opened up the prospect of accounting for all allegedly synthetic a priori knowledge as
Toward a RealisticRationalism
181
' analytic a priori knowledge and thereby resuscitating Hume s empiricism . Ayer (1946, 85), who popularized the new empiricism , wrote : Our knowledge that no observation can ever confute the proposition " 7 + 5 = 12" fact that the symbolic simply on the " " " 7 + 5" depends is synonymous with 12, just as our knowledge expression that every oculist is an eye-doctor depends on the fact that " " the symbol " eye- doctor" is synonymous with oculist . And the same holds good of every a priori truth . To the early positivists , the view that mathematical truths are analytic in the senseof holding in all possible casesmeant that those truths do not have factual content. Carnap (1963, 47) commented on this : . . . What was important in this conception from our point of view was the fact that it became possible for the first time to combine the basic tenets of empiricism with a satisfactory explanation of the nature of logic and mathematics. Previously, philosophers had only seen two alternative positions : either a non empiricist conception , according to which knowledge in mathematics is based on pure intuition or pure reason, or the view held , e.g ., by John Stuart Mill , that the theorems of logic and mathematics are just as much of an empirical nature as knowledge about observed events, a view which , although it preserved empiricism , was certainly unsatisfactory. Mathematics no longer falls between the two Humean stools of relations of ideas and matters of empirical fact, and hence the empiricist is no longer in the embarrassing position of consigning mathematics to the flames along with metaphysics. Wittgenstein and Quine, at first sympathetic to the program of accounting for allegedly synthetic a priori knowledge as analytic a priori knowledge along the lines of Fregean analyticity, came, for different reasons, to seethe program not as the solution for the ills of philosophy ' but as part of the problem . Both saw Frege s theory of meaning as the ' ' principal culprit in Quine s case, as presented in Carnap s semantics. Their criticisms were largely directed at it . For Wittgenstein , disenchantment resulted, as Allaire (1966) and Kenny (1973, 103- 19) have ' pointed out , from his losing faith in Frege s theory of meaning largely due to the difficulty about color incompatibility that Wittgenstein ([ 1922] 1961, secs. 6.375 and 6.3751) described in the Tractatus. Subscribing both to Frege' s conception of logical form and his view that logical necessity is the only "necessity there is, Wittgenstein could not explain" how sentences like This spot is (entirely ) red and (entirely ) green,
182
Chapter 6
which conjoin atomic (and therefore logically compatible [[ 1922] 1961, sec. 4.211]) propositions , could be necessary falsehoods. Wittgenstein (1953, e.g ., sec. 92) came to think that this problem revealed a basic inadequacy of Frege' s account of meaning in natural language. Since there is no basis for logical necessity in the atomic form of such sentences, the idea that the logical powers of sentencesderive from their logical forms , which are hidden beneath their surface syntactic forms , is wrong . Abandoning Frege' s semantics together with much of his own early philosophy , Wittgenstein developed a new account of meaning on which the logical powers of sentencesdo not have the force of necessity because they derive from the use of words . Nonetheless, he maintained his basic positivism on which all of traditional philosophy is to be rejected as nonsense. In the new account of meaning, however, the nonsense derives from the misuse of words . Wittgenstein thus continued to press his attack on metaphysics, though now on the basis of other, as he came to see them , more adequate linguistic means. ' Quine saw Frege s theory of meaning as the principal obstacle to an ' uncompromising empiricism . Quine s (1961c, 23) characterization of the notion of analyticity that he attacks in " Two Dogmas of Empiri . cism" - what a logical truth turns into when we put synonyms for ' s notion of a truth is that rests on synonyms - Frege logical truths and definitions . Quine ' s focus was on Camap , who used his formalization of Fregean analyticity as a basis for developing a compromising empiricism that rejects Mill ' s view that I' the theorems of logic and mathematics are just as much of an empirical nature as knowledge about observed events" as " unsatisfactory " . Quine, however, saw Mill ' s empiricism as essential to an uncompromising empiricism and Fregean analyticity as the last refuge for the rationalist doctrine that logical and mathematical knowledge is a priori knowledge . In his attempt to restore what he saw as the only true empiricism , Quine targeted the Fregean theory of meaning as updated in Camap ' s semantics, in particular on the notion of synonymy on which Frege' s extension of the class of logical truths to the class of analytic truths ' depends. Camap (1956a, 222- 29) had updated Frege s semantics with the introduction of meaning postulates. This promised to provide ' Fregeans with a way to get around Wittgenstein s problem about " ascribing modal properties to sentenceslike This spot is (entirely ) red and (entirely ) green" . On the meaning postulate approach, the logical powers of sentences arising from their extralogical vocabulary are handled in the same way as the logical powers of sentences arising from their logical vocabulary. Given a set of logical postulates that state the contribution of the logical vocabulary to the extensional structure
Toward a Realistic Rationalism
183
of L, the analytic sentencesof L can be characterized as sentencesof L that follow just from the logical and meaning postulates of L. Since meaning postulates can be formulated so that mathematical truths come out as analytic sentencesof the language, the formal side of the logical empiricist ' s resuscitation of Humean empiricism is complete . In " Two Dogmas of Empiricism ," Quine (1961c) tried to show that there is nothing in the areas of definition , linguistics , and logic- the areas to which , respectively, the notions of meaning , synonymy, and analyticity most naturally belong- that can be used to make objective sense of those notions . The story is familiar . Semantic definitions rest on prior synonymy . Abbreviation is irrelevant . An objective notion of linguistic synonymy requires a substitution test that distinguish es between synonymous and nonsynonymous pairs of expressions, but all such tests are circular. Meaning postulates are neither general enough nor explanatory enough to explicate analyticity . ' ' Quine s solution to Wittgenstein s problem about the logical powers of elementary sentenceswas simply to deny that such sentenceshave logical powers . Since the extralogical vocabulary of a language contributes nothing to the logical powers of sentences, there is no need for intensions or the apparatus to represent them. This nihilistic solution fits perfectly with Quine ' s epistemological outlook . From the perspective of his Millian instincts , Quine saw the introduction of intensions as compromising both the extensionality of logic and the purity of ' empiricism . Camap s (1963, 64) logical empiricism , as we saw in chapter 3, explicitly recognized that " the rationalists had been right in ' ' rejecting the old empiricist view that the truth of 2 + 2 = 4 is contingent " ' upon the observation of facts. On Camap s Humean empiricism , only nonformal knowledge depends on experience, while , on Quine ' s Millian empiricism , all knowledge depends on experience with the natural world . .One side of the Quinean coin was to eliminate the logical empiricist ' s restriction of empiricism to nonformal knowledge ; the other side was to provide an epistemology for knowledge generally on which the allegedly a priori truths of logic and mathematics are at bottom empirical . The sticking point for Carnap (1963, 64) was Millian empiricism ' s " unacceptable consequence that an arithmetical statement might possibly be refuted tomorrow by new experiences." Quine (1961c, 44) takes care of this with his account of logical and mathematical statements as central components of our total fabric of scientific beliefs. Even though logical and mathematical statements are revisable in principle in the light of new experiences, their centrality in the total fabric of our scientific beliefs makes their revision something that can occur only as a last resort.
184 Chapter6 Despite their clear differences, the Wittgensteinian and Quinean positions that displaced logical empiricism have much in common . Both adopt a form of naturalism , both are anti -intensionalist in the extreme, both reject absolute necessity, both deny the existence of synthetic a priori knowledge , and , as a consequence, both are implacably opposed to the traditional metaphysical conception of philosophy . As Quine (1960, 76- 77) himself notes, their common target was theories of meaning based on the concept of linguistically neutral meaning . Once those theories are removed and replaced with a conception of meaning based on holism a la Wittgenstein or holism a la Quine, the broad notion of analyticity , which served as the logical empiricists ' basis for a priori knowledge , disappears, and with it the traditional rationalist enterprise . Thus, the only issue left unresolved by the linguistic turn , the issue that today divides Anglo -American philosophy , is whether it is Wittgenstein ' s onto logical naturalism with its positiv istic and therapeutic emphasis or Quine ' s epistemic naturalism with its scientistic and pragmatic emphasis that goes in its place. I said at the outset that the path back to rationalism is through a critical examination of Frege' s role in the arguments of the philosophers who brought about the two phases of the linguistic turn . As we have seen, Schlick, Ayer, Carnap , and other logical empiricists used Frege' s broadened conception of analyticity as a basis for their unmetaphysical conception of the a priori' . And , as we also saw, Wittgenstein and Quine accordingly took Frege s conception of analyticity and ' his theory of ' meaning as their target . Now I concede that Wittgenstein s and Quine s arguments were entirely successful in refuting that conception of analyticity and that theory of meaning . What I do not concede is that , in ' ' successfully refuting them, Wittgenstein s and Quine s arguments do what they have to do to refute all of the theories of meaning on which the a priori and rationalism can be based. Here we return to the problem the apparent logical powers of extralogical words . If Carconcerning ' ' 's nap , Wittgenstein s, and Quine s solutions exhaust the approaches to this problem , then the arguments of Wittgenstein and Quine have done all they have to do to refute them, but if their solutions do not exhaust all of the approaches, the entire second phase of the linguistic turn is in question . An alternative approach has to be intensionalist , so it has to retain the idea that meanings are sensesof linguistic expressions, analyticity is a property of senses, and synonymy is a relation among them. Further, since Wittgenstein ' s problem arises in connection with logically elementary sentences, the sense structure of sentences cannot"always be - and in the case of logically elementary sentencessuch as Bachelors are single " isn' t- reflected in their logical syntax . Finally , the Frege
Toward a Realistic Rationalism
185
principle , which Wittgenstein ([ 1922] 1961, secs. 6.37 and 6.375) embraces in the Tractatus, that logic is the only source of necessity, has to be dropped . In opposition to Frege, we have to say that linguistic meaning is an independent source of necessity. We have to say that logically elementary sentencescan have richly structured senses. This move is the key to solving Wittgenstein ' s problem concerning the logical powers of logically elementary sentences(seeKatz [to appear]). The appearance that the powers are logical is misleading . The powers are not logical but semantic. Semantically rich nonlogical sensesmake it possible to overcome the logically elementary character of sentences such as " Bachelors are single ." To make this approach work , we require a non -Fregean definition of sense, one that characterizes the semantically rich senses of syntactically simple expressions nonlogically . I provided such a definition in SemanticTheoryand subsequent publications . I (1972, 1- 10) defined sense as the aspect of the grammatical structure of expressions and sentencesthat determines their meaningfulness, synonymy , antonymy, ambiguity , redundancy, analyticity, and other senseproperties and relations. Frege' s arguments that sensesare necessaryto explain the informativeness of identity statements, substitution into oblique contexts, and meaningfulness in the absence of reference motivate the introduction of a concept of sense, but they do not motivate the introduction of just Frege's concept. Our concept of sense does just as well for these purposes. And it does better against attacks on intensionalism first by Wittgenstein and Quine and later by Donnell an, Kripke , and Putnam. As I (1986, 1990b, 1994a, 1997) have argued, the successof these attacks reveals, not a weakness of inten sionalism , but a weakness of Fregean intensionalism . What makes my definition of sense radically different from Frege' s definition is that it uses no concepts from the theory of reference. In defining senseas the determiner of referential properties and relations such as denotation , truth , and logical equivalence, Frege defines the concept of sense in terms of the vocabulary of the theory of reference, thereby reducing the theory of senseto the theory of reference. On our definition , the theory of sense is fully independent of the theory of reference. It has nothing to do with the relation of language to the world . It is an autonomous theory about one aspect of the internal grammatical structure of sentences. With this new definition of sense, the Fregean grip on the notion of analyticity can be broken. Frege (1953, 4) had defined an analytic proposition as a consequenceof logical laws plus definitions without ' any assumptions from a special science. Frege s (1953, 99- 104) rationale /1 for this definition was that it is more fruitful " than the LockeanKantian definition . To be sure, fruit fulness is a virtue if analyticity is
186
Chapter 6
to serve as the basis for explaining logical truths as analytic truths in a logicist program to reduce mathematics to logic . Independently of such an ulterior motive , fruit fulness might well be a vice. If we are looking for a definition to explain analyticity in natural language, a Lockean-Kantian definition , just becauseof its unfruitfulness , might be preferable . I (1972, 171- 200, 1986, 1988, 1990b, 1994a, 1995, 1997, and in preparation ) have argued that a reconstruction of the Lockean-Kantian definition of analyticity fits the facts about sense structure in natural better than Frege' s and further that , since Wittgenstein ' s and language ' s criticisms were Quine very largely tailored to refute Fregean inten sionalism , their criticisms do not refute intensionalism per se, and that the differences between our intensionalism and Frege' s enable the former to meet their criticisms . The principle consequence of this argument , for our present concerns, is that the problem of explaining synthetic a priori knowledge for nearly the same broad range of propositions that Locke and Kant considered to be synthetic a priori is reintroduced . On our narrow concept of analyticity, a sentence is analytic in case it has a referring term with a sense that contains the sense of the entire sentence. Containment means literal , " beams in the house," containment , not the figurative , " plant in the seed," containment of Fregean analyticity . Analyticity is not a species of logical truth where , for example, every proposition contains the disjunction of itself with every other proposition . (See Katz [ 1986, 62 3, in preparation ] .) Since the breadth of a notion of analyticity and the breadth of the range of propositions that are synthetic a priorion that notion are , the narrower the inversely related- the wider the notion of analyticity ' s wide notion leaves a of range synthetic priori propositions Frege little room for synthetic a priori propositions and Carnap ' s far wider notion left (as he of course intended ) no room for them at all . In contrast, the narrowness of our concept of analyticity entails a wide range of such propositions . Essentially the same class of a priori mathematical and other formal truths are synthetic propositions on our analytic / synthetic distinction as on Kant' s. Thus, as I (1990b, 297) reinstatement of argued at the end of The Metaphysicsof Meaning, the the traditional analytic / synthetic distinction brings " twentieth century philosophy full circle round to the situation just prior to the Logical " Positivist , Wittgensteinian , and Quinean attacks on metaphysics. After nearly a century devoted to looking for ways to avoid Kant' s problem of explaining how synthetic a priori knowledge is possible, we have come face to face with it again. The problem reemerges in even more aggravated form becauseKan tian explanation is no longer an option . Attempts to ground synthetic a priori knowledge on our own nature are inadequate for various
Toward a Realistic Rationalism
187
reasons, the most salient of which being that logical and mathematical truths tell us not merely that something is so but that it must be so. Human nature , however transcendentalized, cannot explain their necessity. The argument in The Metaphysicsof Meaning thus seemed to me to leave us with but one approach to explaining synthetic a priori knowledge in the formal sciences. Such knowledge has to be grounded in the nature of the logical and mathematical facts. Further, since the epistemology of formal knowledge piggybacks on the ontology of the formal sciences, a rationalist epistemology must be based on features of logical ' and mathematical facts that explain why they couldn t be otherwise . of logical and mathematical facts that Finally , since the only conception ' is realism, realism had to be t be otherwise explains why they couldn available for the construction of a rationalist epistemology for logic and mathematics. The situation in the philosophy of mathematics was the other way around . Since causal contact with abstract objects is impossible , a rationalist epistemology had to be available for the construction of a ' realist position to answer Benacerraf s charge that realism cannot explain mathematical knowledge . The upshot was clear: not only does a realist ontology have to be available for the construction of a rationalist epistemology, but a rationalist epistemology has to be available for the construction of a realist ontology . These reciprocal demands dictated an approach to the explanation of synthetic a priori knowledge on which rationalist epistemology and realist ontology are combined into a single, unified metaphysical position . This position is partly developed in the earlier chapters of this book . In chapters 2 and 3, I showed how it solves the epistemic puzzle for realism induced by the causal inaccessibility of mathematical objects by explaining the a priori character of mathematical knowledge on the basis of a realist account of the necessity of mathematical truths . This explanation of how we can have knowledge of abstract objects without causal accessand how it can be synthetic a priori knowledge without resorting to Kantian idealism- on which it becomes knowledge of our own sensibility and understanding - is only a partial solution to the general problem of synthetic a priori knowledge . The remaining question is whether this explanation can be generalized to other kinds of synthetic a priori knowledge . How much of the overall problem of explaining synthetic a priori knowledge can be handled within the conception of a rationalist epistemology that has been sketched in the previous chapters? The rationalist epistemology developed thus far is incomplete along two dimensions . Horizontally , it needs to be spelled out more fully for the caseof synthetic a priori knowledge in other formal sciencesand in
188
Chapter 6
the natural sciences. The extension to logic seems straightforward , while the extension to linguistics , at least in its present state, is not. The extension to the natural sciences is trivial if the only synthetic a priori knowledge in them is the principles from the formal sciences they employ, but not otherwise . Vertically, our sketch of a rationalist epistemology has to be extended to synthetic a priori knowledge in philoso phy. We cannot deal with all of these questions here, but , as the issue of philosophical knowledge has been at the center of the arguments in the linguistic turn which ushered in the present naturalist / empiricist hegemony, I will address the question of synthetic a priori philosophical 1 knowledge . 6.2 From Philosophyof Mathematicsto Philosophy Two related factors have undermined the twentieth -century philoso ' pher s belief in the existence of a priori philosophical knowledge . One is the dissatisfaction that derives from the positivist ' s invidious comparison of progress in metaphysics with progress in natural science. (1963 , 44- 45) once put the comparison in these terms: Camap Even in the preVienna period , most of the controversies in traditional metaphysics appeared to me sterile and useless. When I compared this kind of argumentation with investigations and discussions in empirical science or in the logical analysis of language , I was often struck by the vagueness of the concepts used and by the inconclusive nature of the arguments . . . . I came to hold the view that many thesesof traditional metaphysics are not only useless, but even devoid of cognitive content. This comparison continued to be influential even after philosophers stopped taking the view that traditional metaphysics is senseless seriously. But as a consideration against synthetic a priori philosophical knowledge , it is a straightforward petitio, because it assumes, contrary to traditional metaphysics, that metaphysics and scienceare enough alike to be put on the same scale. The positivist cannot simply assume that 1. Strictly speaking, we do not require an account of natural knowledge on which it is uniformly a posteriori , since most philosophers who believe that there is synthetic a priori knowledge in natural science think that it consists in principles of systematization such as simplicity and principles of inference, which are, as it were, borrowed principles . Moreover, the Quinean holistic picture of knowledge , restricted to natural knowledge , would seem to support the claim that there are synthetic a priori truths native to natural science. This might be challenged, say on the basis of Koslow (1992), but this raises issues beyond the scope of this study .
Toward a Realistic Rationalism
189
scientific progress is an appropriate yardstick for judging progress in metaphysics, and, seeing that the latter does not measure up to the former , conclude that something is wrong with metaphysics. In The Metaphysicsof Meaning, I (1990b, 313- 17) took some initial steps in arguing further that they are not enough alike for such a comparison . This chapter will continue the argument . Even the brief account of synthetic a priori philosophical knowledge that I will present here will explain why philosophy is not sufficiently sciencelike for it to be judged by scientific progress. The other factor is the apparent failure of metaphysics in the face of philosophical skepticism . This is a far more serious threat to our confidence in the existence of a priori philosophical knowledge . Unlike the interpretation that the positivist puts on the facts about progress in metaphysics and science, the apparent failure of metaphysics to handle skepticism does not have to be argued for. Most philosophers would probably say that the failure is more real than apparent . The curious thing is that the threat of skepticism has been an impetus in converting many philosophers to naturalized epistemology. Stroud (1984, 209- 54) shows that Quine ' s naturalized epistemology offers no advantage of all' over traditional metaphysics in dealing with skepticism . Goodman s (1955) naturalized inductive epistemology, which is an attempt to escape Humean skepticism about induction , is equally unsuccessful. As I (1962, 48- 49) once argued, the Humean problem resurfaces as a problem of what right we have to think that the inductive practices to be explicated are better than other conceivable (counterinductive ) practices, that is, what right we have to think they are reliable in the long run . Goodman (1955, 64- 66) confidently assumes that Hume dissolved the problem , but this is not the case. Hume didn ' t show that there is no rational justification of induction , but only that there is none on the empiricist assumption that conclusions about unobserved events are either based on inferences from matters of fact in the past or on relations of ideas. Give ~ the dependency of Hume ' s ' argument on the empiricist theory of knowledge , Goodman s naturali zation is no more protection against the skeptic than Quine ' s. In the remainder of this section, I will try to show that an extension of the rationalist epistemology set out in the previous chapters to philosophical knowledge provides us with a way in which skepticism can be isolated from a large class of issues about the nature of our scientific and philosophical knowledge . Nagel (1986, 67- 71) rightly says that the problem of skepticism is an unavoidable consequence of the realist' s objective conception of the world , but I think that we can make living with the problem considerably easier. In the next section, I will try to show that Camap ' s critique of synthetic a priori
190 Chapter6
philosophical knowledge fails because the semantics on which it is based has the same problems as the multiform semantics of combinatorial approaches to mathematical truth (as discussed in chapter 2, section 3). The starting point for our rationalist account of synthetic a priori knowledge in the formal sciences is the supposition that such knowledge is knowledge of abstract objects. Since abstract objects are unchangeable in their intrinsic structures and the relations they have to one another in virtue of such structures, the way abstract objects actually are is the way they must be, and, accordingly, basic knowledge and much transcendent knowledge in the formal sciencescan be known a priori because the way they actually are can be determined by pure reason. The initial problem in extending this account to philosophical knowledge is that philosophy is not for the most part a study of abstract objects. This means that our account of basic knowledge in the formal sciencescannot be extended to philosophical knowledge wholesale, but only to certain limited casesof philosophical knowledge where intuition informs us about philosophically relevant properties of abstract objects. Philosophy can be about anything , but its main interest focuses on epistemological and metaphysical questions. What prevents the account in chapter 2 from applying to philosophi cal knowledge wholesale is that philosophy studies both the foundations of the sciences and the grounds for its own conclusions about them. But even though in the final analysis we may not be able to extend our account of synthetic a priori knowledge in the formal sciences to philosophical knowledge generally, we can use the account to explain synthetic a priori philosophical knowledge in the foundations of the sciences. What is required for this is a specification of the relation between philosophy in its role as a second-order discipline and the first - order sciences that it studies. This relation will explain how phi losophy can address the aspects of reality studied in the first -order disciplines . Thus, we ask: (Ql ) What is the relation of philosophy to the scienceson which philosophical synthetic a priori knowledge rests? (Q2) What is the nature of philosophical knowledge ? Our aim is to restore a conception of philosophy as an independent discipline providing knowledge of the most general features of reality and the nature of knowledge in the sciences. Adopting this aim does not commit us to the view that philosophy is logically prior to science or that it has primary jurisdiction over knowledge in the sciences. There ' is a sensein which the sciencesare first , though it is not Quine s sense of science as first philosophy . Philosophy is a second-order discipline
Toward a RealisticRationalism
191
with respect to the sciences, studying their onto logical commitments , conceptual structure , and epistemic character. There is also a sense in which philosophy is first . It concerns issues arising from the ontologi cal, conceptual, and epistemological principles on which scientific investigation depends. Since the sciencesrest on those principles , trying to settle such issues within the sciences themselves would be a bootstrap . As will be recalled from chapter 3, this is where 'operation s Quine neoempiricist epistemology comes to grief : an epistemology that makes scientific investigation settle all issues including those concerning its own constitutive principles is incoherent} An answer to (Ql ) can be given in terms of the conception of ontology as a foundational discipline of foundational disciplines presented in chapter 5. We distinguished three levels of investigation : the sciences, the deeper level of their philosophical foundations , and the still deeper level of general philosophy; At the first level , mathematics, logic , physics, and the like investigate numbers , sets, propositions , atoms, genes, and other objects in the world . At the next level , there are the philosophy of mathematics, the philosophy of logic , the phi losophy of physics, and the philosophies of the other sciences, which concern the foundations of their respective sciences at the first level. At the deepest level , there are the philosophical disciplines of pure ontology and epistemology. The disciplines at each level are concerned with questions that arise but are not answered in the prior investigations . Since the sciencesare continuous with common sense, they try to answer questions that arise in the course of common -sense reflection on the world but cannot be answered by common sense. These are questions like : How many 2. Contrary to the early positivistsand others- then and now- who havefallen under ' Wittgensteins spell, we maintainthat philosophyseeksknowledge, that is, philosophical statementsthe truth of which is adequatelygrounded. Schlick(1959,57) claimsthat " the '" was error of ' metaphysics " to supposethat the foundationsof [science ] was made up of " philosophical of and crowned a dome statements(the statementsof the theory knowledge), by of philosophicalstatements(calledmetaphysics ). Schlick's (1959 , 57) argumentfor this is: For if , say, I give the meaningof my words through . . . other words, one must askfurther for the meaningof thesewords, and soon. This processcannotproceed . It alwayscomesto an end in actualpointings. endlessly This argumentis too strong. If it showsthat the metaphysicianis caughtin an infinite , it will also show that the mathematician regressand must forswearmaking statements . No doubt, too is caughtin an infinite regressand must alsoforswearmakingstatements the responsewould be that mathematiciansare not engagedin a semanticenterprise , . but, of course, neitheris the metaphysician
192
Chapter6
numbers are there? What does it mean to say there are infinitely many of them? Why does dye turn liquids a uniform color? Why does water, unlike other liquids , expand on freezing ? Why does adhesive tape stick? What is the difference between the pronouns in " It is crawling on the wall " and in " It is raining outside " ? Why are " flammable " and " inflammable " synonyms rather than antonyms ? at the level of the foundations of the sciences, philosophies Similarly, of mathematics, physics, linguistics , and so on try to answer questions that arise in these sciencesbut perforce go unanswered in them. Typically , these questions concern the onto logical nature of the objects in their domain (more generally, the reality that the sciences studies) and the epistemological character of knowledge in the science. Mathematics tells us about the vastness of the realm of numbers , but does not tell us what kinds of things numbers are. To answer that question and to explain how we acquire the knowledge we have of them, of mathematics formulate philosophical theories, such philosophers as Godel ' s realism and Brouwer ' s conceptualism, and construct arguments to show that their theory is the most satisfying account of the mathematical facts and the relevant issues in the foundations of mathematics. Let us call the we obtain in the foundations of the sciences " foundational knowledge." Foundational knowledge knowledge conforms to the naturalist ' s claim that philosophers do not study scientific reality in the same hands- on way that scientists do. Even assuming that the natural ists were also right that scientific theories collectively cover reality completely, philosophical investigations can still inform us about reality , e.g ., about whether numbers are abstract or concrete objects or not objects at all . Accordingly , it is not , as some naturalists have wanted to say, that second-order philosophical investigation can contribute nothing to our knowledge of the world , but rather that its contribution to our knowledge of the world is indirect , in contrast to the contributions of science. Since such investigation takes place below the level of scientific investigation , it concerns objects as they are represented in scientific theories, and consists largely in interpreting scientific theories. Even though philosophical contributions to our knowledge of the world are indirect , they address questions about the nature of reality because the questions are about objects the existence of which has already been vouchsafed by the science in question . Finally , pure ontology and epistemology address questions that arise at the level of the foundations of the foundations of the sciences. The questions they address arise in the controversies between rival philo sophical theories in the foundations of the sciences, but are not answered in those controversies. While those philosophical theories are concerned with the nature of the reality studied and the knowledge
Toward a Realistic Rationalism
193
acquired in the sciences, pure ontology concerns questions about the nature of the categories that those theories employ for representing the most general aspects of reality. As illustrated in chapter 5, ontology explains categories such as Abstract and Concrete in a way that accounts for their relations to one another and to other general categories. Pure epistemology concerns questions about the grounds of the principles on which scientific and foundational knowledge rest. It is especially concerned with the validity of those principles in the light of skeptical challenges to them. I will call the knowledge in pure ontology and epistemology " metaphysical knowledge ." We have distinguished between scientific knowledge and philosophi cal knowledge and between foundational knowledge and metaphysical knowledge . There is a third distinction that is important for our conception of philosophical knowledge . This is a distinction between internal knowledge and external knowledge . In application to our conception of philosophical knowledge , internal knowledge is knowledge within a framework where justification only requires meeting internal standards of evaluation , and external knowledge is knowledge about a framework where the internal standards cannot play the role in justification that they do within the framework . This terminology is adopted from philosophers like Kant and Carnap, but it marks a different distinction . In particular , unlike theirs , my terms " internal " and " external" are not also connected with a particular philosophical view about what a framework is or with philosophical doctrines like the positivist doctrine that external questions are meaningless. Justification in the sciences only requires meeting the prevailing standards. It would be nutty to insist that quantum methodological ' physics isn t justified until the physicists can explain how they know the future will be like the past or that Darwinian biology isn' t justified until biologists can explain how they know the world didn ' t come into existence yesterday complete with apparent causal traces. This , I presume , goes without saying . But it is worth saying that justification in the foundations of the sciencestoo only requires meeting the prevailing standards. This is a point that emerged in our response to the epistemic challenge to realism in chapter 2. There we saw that the successof a philosophical position in the foundations of mathematics does not depend on its successagainst skepticism . Since the prevailing standards of reasoning are not in question in the controversy in the foundations of mathematical knowledge , empiricism does not have to satisfy the Humean skeptic and rationalism does not have to satisfy the Cartesian skeptic . Similarly, philosophical skepticism is not an issue in the case of explications in the area of metaphysical knowledge . For example, whether the account of the abstract/ concrete distinction presented in
194
Chapter6
chapter 5 is the best explication of the distinction depends on how well it handles various alleged counterexamples, on systematic considerations such as simplicity , and on extrasystematic considerations such as the light it sheds on foundational and scientific issues. The question is internal to pure ontology . The explication does not have to refute philosophical skepticism concerning the categories Abstract and Concrete or any others that enter into the explication . Similarly , explications such as the explication of rational choice or confirmation are internal matters within pure epistemology. The situation changes radically when we cross the line into external metaphysical knowledge . Here a large portion of our epistemic framework , even the entire framework in some cases, is called into question on the basis of a skeptical challenge to the epistemic standards that prevail within the framework . The skeptic adopts what is putatively a vantage point outside the framework and challenges us to show that we know what we take ourselves to know in spite of the skeptical possibility that the standards on which the knowledge rests led us astray. Skepticism unites philosophical foes. Although rationalists and empiricists , realists and nominalists , Cartesian dualists and materialist monists , and so on disagree about what internal knowledge we have, they perforce agree that we have external knowledge . Hence, they must now make common cause to defend the scientist' s claim to knowledge in the sciencesand their own claims to knowledge in the foundations of those sciences. With respect to a particular skeptical issue, external metaphysical knowledge can thus be of three sorts: (A ) Knowledge that there is internal knowledge . (B) Knowledge that there is no internal knowledge . (C ) Knowledge that we cannot know whether there is internal knowledge or not. In case (A ), we prevail over the skeptic . We have external knowledge that vindicates our belief in internal knowledge in the sciences and philosophy . In case (B), the skeptic prevails over us. We have external knowledge that vindicates the skeptic' s position that we do not have internal knowledge . In case (C), it' s a standoff . We have external is undecidable. And , of course, knowledge "that the issue of skepticism there is the none of the above" possibility . In this case, we fail to obtain external metaphysical knowledge . It is important to recognize that case (C) is a form of external metaphysical knowledge . Recognizing it prevents us from impatiently dismissing skepticism . Some philosophers dismiss skepticism out of
Toward a Realistic Rationalism
195
hand because they despair of providing a justification that will stand up to the skeptic . Even supposing that refuting the skeptic is hopeless, recognition of (C) shows that a reason is needed to claim that it is hopeless. Sometimes philosophers dismiss the issue of skepticism as pointless . Their sentiment here is that a reason to reject skepticism would only be a reason for believing what we cannot help believing anyway . The trouble with this sentiment is that , regardless of what our psychology might encourage us to believe, it is still worth knowing the truth about our beliefs if it can be known , even if it is the hard truth that what we believe is provably without rational justification . Although there is some plausibility in thinking that we can' t help believing that we have internal knowledge , there is no plausibility , as things stand, in thinking that we cannot know the truth about that belief. In this connection, one is reminded of Prichard ' s (1949) well -known argument for dismissing the issue of skepticism . He dismisses it because it depends on the Cartesian theory of knowledge and he takes its formulation of the epistemic task to be mistaken. He (1949, 15) argues that the theory of knowledge in this sensecannot exist because our customary knowledge " neither can be, nor needs to be, improved or vindicated by the further knowledge that it is knowledge ." A skeptical doubt about something we know is not a " genuine doubt but rests on a confusion ." We see that there can be no " genuine doubt " in a case where we know a proposition P once we notice that . . . what we are really doubting is not whether this consciousness [the knowledge that P] is really knowledge , but whether consciousness of another kind , viz ., a belief [that P], was true. [A doubt could only arise if ] we have lost hold of , i .e., genuine no longer remember, the real nature of our consciousness of yesterday. In this case, Prichard thinks that our only recourse is to regain that consciousnessby repeating the cognitive steps by which we justified the belief a day ago. The problem with Prichard ' s objection is that what he says is true only when the knowledge in question is internal . Since the prevailing standards do not come into question with common -sense knowledge , knowledge in the sciences, or foundational knowledge , the only issue in these areas is whether the standards were properly applied in the individual caseor whether something we have taken to be the standard either isn' t the correct standard or isn' t correctly formulated (e.g ., issues like those Hempel [1965, 3- 79] raises about confirmation or issues like the intuitionist challenge to double negation ). To be sure, the only
196
Chapter 6
that could arise in a case of internal knowledge results genuine doubt from having " lost hold of " the fact that the appropriate standards have been properly applied , and the only recourse for dispelling the doubt is to repeat the steps taken in their correct application . But when the ' knowledge in question is external, what Prichard says isn t true . His reflections do not even apply to the Cartesian claim that there is real skeptical doubt about propositions we ordinarily take ourselves to know and that the job of the theory of knowledge is to defend them ' against that doubt . Prichard s criticism of the Cartesian theory of knowledge runs afoul of the internal / external distinction , since, on that distinction , skeptical doubts arise in an area where the standards are open to challenge. We are now in a position to extend our account of synthetic a priori knowledge in the formal sciencesto synthetic a priori knowledge in the area of philosophical knowledge . On the basis of the previous considerations , we divide (Q2) into (Q2a) and (Q2b). (Q2a) What is the nature of internal philosophical knowledge ? (Q2b) What is the nature of external philosophical knowledge ? No question is begged, since we are making no antiskeptical claim at this point . As yet , the answer to (Q2b) is left open and hence no assumption is made about whether there is internal knowledge for our account of synthetic a priori philosophical knowledge to be an account of. We will turn to the issue of external philosophical knowledge below. I will consider two casesof synthetic a priori knowledge in connection with (Q2a): philosophical knowledge in the foundations of the formal sciences and internal metaphysical knowledge . Extending the explanation for synthetic a priori truths in the formal sciencesto synthetic a priori truth in their foundations is straightforward .3 The extension begins with the response we gave to the epistemic challenge to realism in chapter 2. That response is an account of logical and mathematical knowledge that is based entirely on intuition and reason. Knowledge in the foundations of the formal sciences, like knowledge in the formal sciences, divides into a range of basic caseswhere intui of how synthetictruth in 3. It is not at all a bivial matterto convertour explanation a into an of howsynthetictruth can be known and mathematics priori explanation logic in theformalsciences generallycanbe knowna priori. Evenif theonly othercasewe versionof linguistic haveto consideris linguistics , it will takea far betterdeveloped thanI haveproducedhereto explainhowour accountof synthetic realisticrationalism canbeadaptedto thecaseof synthetica priori a prioritruth in logicandmathematics . ThevariousthingsI havesaidaboutthecaseof linguisticsin this truth in linguistics . andearlierworksmustsufficefor thepresent
Toward a Realistic Rationalism
197
tion and reason can exclude every possibility of falsehood and another range of cases where they cannot. In the former cases, no empirical information is required to establish these synthetic truths and hence the account is an explanation of how they can be known a priori . In the latter cases, foundational truths can be treated in the way apodictic knowledge in the formal sciences was treated in chapter 2. Some mathematical statements, e.g ., Church ' s thesis, although they have no proof , still can be shown to be a priori mathematical knowledge on the basis of an a priori justification that shows them to be essential to the best systematization of a body of a priori mathematical knowledge . Although we are unable to prove that recursiveness is effective computability , accepting their equivalence can presumably be shown a priori to be essential for the best systematization of the overall body of mathematical knowledge . The same kind of a priori argument is possible for the apodictic status of propositions in the foundations of the formal sciences. To be sure, apodictic propositions in logic and mathematics belong to formal sciences that consist very generally of propositions we can prove , while propositions in the foundations of the formal sciences belong to a branch of knowledge that , for the most part , consists of propositions we make no pretense to be able to prove . But the body of knowledge with respect to which we determine what is essential to the best systematization can be taken broadly to include the knowledge in the formal sciencestogether with the knowledge in their foundations . When we have a body of knowledge that includes both the formal sciences and their foundations , we can argue that synthetic propositions in the latter that cannot be proved can nonetheless be known a priori becausethey can be shown on the basis of a priori argument to be essential to the best systematization of that comprehensive body of knowledge . As we noted, a complete account of the objects in the domain of the formal sciencesdepends on knowledge of their foundations . Without an onto logical account of the objects in the domain of a formal science and an epistemology for our knowledge of them , the complete understanding for which we strive cannot be obtained . Benacerraf' s ([ 1973] 1983) combined semantic and epistemic requirement on accounts of mathematical truth is an expression of this need for a more complete understanding than the formal sciencesthemselves provide . This is not to deny that mathematicians, logicians , and linguists can get their technical work done without having a complete picture of of the nature of mathematical , logical , and linguistic objects. But as Quine (1961b, 47) once said
198
Chapter6
. . . [the linguist ' s ignorance concerning the ontology of semantics ] is a theoretically unsatisfactory situation , as the theoretically minded among the linguists are painfully aware. This goes as well for ignorance about numbers , sets, propositions , sentences, and other objects of study in the formal sciences. Philosophi cal claims like Putnam ' s (1975c) that mathematics needs no foundations are misleading . While it is true that mathematicians typically do their technical work without enlightenment about the philosophical foundations of mathematics, it is not true that the fuller understanding we seek about mathematical reality can be achieved without philosophy . As we argued at the end of chapter 4, the issue of whether there are such things as numbers depends as much on philosophy as on mathematics. Our approach to explaining knowledge in the formal sciences and their foundations contrasts with the approach Chisholm (1977, 34- 61) calls " linguisticism ." That approach claims that a priori truths are true in virtue of rules of language. This doctrine is a holdover from the attempt during the linguistic turn to explain the truths of formal science as analytic a priori truths . The doctrine underlies Dummett ' s (1978, 214) view that the notions of reference and truth in mathematical practice do not depend on an objective realm of objects, but on conceptual connections within the language of mathematical practice. Hale (1987, " 124) also appears to endorse linguisticism when he claims that the obvious and natural account of our knowledge of truths about abstract " objects for Platonists who abandon a contact" epistemology' is that the truth of a statement about abstract objects is a matter internal to " ' language in the sense that truth is owed to conceptual liaisons. Although our conception of a priori formal knowledge takes the faculty of reason to be the means of recognizing analytic necessary connection, it also takes it to be the means of recognizing synthetic necessary connection. We agree with linguisticism that there is an area in which recognizing necessity is entirely a matter of linguistic understanding that that area is much narrower than the one we , but claim ' ' s and Camap s notion of analyticity . On our approach, get from Frege the area in which recognition of necessary connection is entirely a matter of linguistic understanding is just the class of metalinguistic " " ' Holmes ' sentences such as " ' John loves his mother is meaningful , ' ' " " ' ' Sister and female sibling' are dusted the table is ambiguous , " " ' Unmarried ' and ' closed' are " "' , antonymous ,' Open " " ' synonymous " ' is bachelor is redundant , , Squares are rectangles " and so on.analytic ' "' If such Squares are not rectangles is contradictory, suffice considerations which sentencesare the only casesfor linguistic
Toward a Realistic Rationalism
199
for knowledge of necessary truth , then the operation of reason is only in part linguistic , and the least significant part at that (see Katz 1997, in preparation ). Caseswhere recognizing necessaryconnections is more than a matter of understanding the meaning of sentences show that linguisticism ' s claim that recognition of necessary relations is entirely linguistic is too strong . For example, we can fully understand the meaning of the sentence " Every even number greater than two is the sum of two " primes but still not know whether it is true or false. In fact, the reasoning from which we could obtain knowledge of the truth or falsehood of Goldbach ' s conjecture can only get started after we grasp the meaning of the sentence. This objection cannot be deflected by moving to a stronger notion of understanding a sentence than the ordinary one of understanding its compositional meaning . Any notion ' strong enough to deliver the truth value of Goldbach s conjecture would be far too strong to count as a notion of linguistic understanding . The other case to consider in answering (Q2a) is that of internal metaphysical knowledge . In ontology, internal knowledge will consist of a priori explications of concepts that occur in foundational discussions (e.g ., the explication of " abstract" and " concrete" presented in chapter 5), and hence what goes for foundational knowledge goes for such internal metaphysical knowledge . In epistemology, internal knowledge will consist of explications and propositions stating that particular formal or foundational truths are a priori . Since intuition and pure reason are all that is required to show that an a priori truth is a priori , the propositions expressing the knowledge that truths in the formal sciencesand their foundations are knowable a priori are themselves knowable a priori . We turn now to (Q2b). Earlier we described three possible outcomes to the search for external philosophical knowledge . Since we are only trying to formulate a rationalist / realist position and defend it against naturalist / empiricist positions , (B) can be ignored . If (B) obtains, we all go under together- those on Neurath ' s boat as well as those sailing under realist and rationalist colors. Hence, assuming either (A ) or (C) obtains, the question is what can be said about external metaphysical knowledge . Our explanation of synthetic a priori internal philosophical knowledge cannot be extended to external metaphysical knowledge . On the one hand , external philosophical knowledge is never knowledge of necessary truth . Philosophical skepticism shows that it is possible for propositions such as that the future will be like the past, that there is an external world that causes our internal percepts, and that there are other minds to be false. Such external propositions are not only
200
Chapter6
supposably false, like logical and mathematical truths ; they are possibly false as well . Hence, explanation of external philosophical knowledge cannot be based on reasoning that rules out the possibility of its falsehood. On the other hand , not being internal knowledge , external philo sophical knowledge is not part of the system of knowledge embracing the formal sciences, their foundations , and the internal metaphysical foundations of their foundations . To be sure, propositions such as that the future will be like the past, that there is an external world that causes our internal percepts, and that there are other minds are assumed within the overall body of scientific knowledge , but that doesn' t transmute them into external knowledge . Further, external metaphysical propositions cannot be justified as apodictic within the overall body of internal knowledge , since the argument would have to be that they are essential to the best systematization . But such an argument begs the skeptic' s question . Skeptics have their own , quite different , ideas about the best systematization . One thing that can be said about external metaphysical knowledge , assuming that it exists, is that it is synthetic a priori knowledge . It is a small irony that the very skepticism that calls our internal knowledge into question shows that external metaphysical knowledge cannot be a posteriori . Experience is useless to defeat either Cartesian skepticism or Humean skepticism . The demon hypothesis can explain our experience , whatever the experience happens to be, just as our causal story can explain our experience, whatever it happens to be. Counterinductive hypotheses are compatible with the strongest evidence of regularity , just as our inductive hypothesis is compatible with it . In showing that experience cannot distinguish between the hypothesis that the actual world is a world in which the skeptic' s philosophical fantasy is fact and the hypothesis that it is not , skepticism establishes the irrelevance of experience to metaphysical investigation . Hence, external , if it exists, is a priori knowledge . metaphysical knowledge G. E. Moore ' s (1959) celebrated a posteriori " proof " of our knowledge of the existence of external objects challenges this conclusion. Moore thought that the best way to prove the existence of external objects is to exhibit them empirically for all to see. So, Moore shows us one hand and then another, claiming thereby to have proved that two external objects exist, and hence that external objects exist. The demonstration purports to show that we cannot doubt the existence of external objects becausethe fact of their existence is already part of our common -sense knowledge of the world . All that is required for us to realize that we have knowledge of their existence is to have our attention called to it
Toward a Realistic Rationalism
201
by an appropriate empirical exhibition . We have to show that this argument does not answer the Cartesian skeptic . From the viewpoint of prescientific thought about the world , Moore was certainly right to think that his argument is as strong a proof as might be given for anything (outside mathematics and logic , presumably ). Further, even from the viewpoint of the sciences, Moore is right . The Cartesian question of the existence of external objects is not like the question of why objects fall down , why dyes turn liquids a uniform color, or why water expands on freezing . Such questions arise in prescientific reflection and get answered in scientific investigation . The question of the existence of external objects does not arise in our common -sense reflections about the world , and it also does not arise in the practice of science or the philosophy of science. This brings out the full scope of Moore ' s argument . The argument not only applies to the area of common sense, it applies equally well , and for the same reason, both to science and to the foundations of science. Being built up from common -sense knowledge , science may revise our common -sense beliefs about the form or constitution of external objects. It can show that matter is discontinuous , that the earth is round , or that space is non -Euclidean. But science cannot, on pain of internal incoherence, revise the common -sense commitment to the existence of an external world . Moreover , given that Moore ' s argument is valid for the sciencesand given the relation between the sciencesand their foundations , his conclusion is valid for their foundations too. But Moore ' s argument has essentially the same problem as Prichard ' s. Moore (1959, 147) rightly says: . . . we all of us do constantly take proofs of this sort as absolutely conclusive proofs of certain conclusions - as finally settling certain questions, as to which we were previously in doubt . But the fact that Moore ' s proof is typical of the proofs we give in ordinary life and science- he cites proving that there are three misprints on a page as an illustration of what he means- should alert us to its vulnerability as a refutation of Cartesian skepticism . Moore ' s " " , but it is inconclusive externally proof is absolutely conclusive internally 's Stroud 110 criticizes Moore (1984 ) , . proof as follows : . . . the philosophical sceptic' s denial of our knowledge [is] the outcome of an investigation into the basis of all the knowledge or certainty we think we have about the world around us. That is why I think we feel it is not a " sufficient refutation " of that " scepticism simply [for Moore ] to bring forward a particular case
202
Chapter 6 . . . in which we do know of the existence of some material " ' object. The philosopher s assessmentof all of our knowledge of the world around us is meant to apply to every particular case in which we do think we know of some material object, so no case that could be brought forward would escape scrutiny .
Since the skeptic questions our entire system of internal knowledge including the prevailing standards that ground that knowledge , neither the beliefs comprising the syst~m nor the standards are of use against such an external challenge, and, as a consequence, Moore ' s reminder that the existence of objects outside of us is already part of our common -sensesystem of beliefs is of no help against the skeptic' s challenge to the system as a whole . Moore ' s premise that we know that here is one hand and there is another is certainly true. The Cartesian skeptic does not dispute that we have such knowledge about the existence of the two hands, but rather that it is knowledge of the existence of objects outside of us. The skeptic' s challenge is a challenge to the system as a whole because it provides a uniform alternative interpretation of our beliefs about hands and other objects of experience on which they are inner phenomenal objects rather than outer physical" objects. The Cartesian skeptic no more denies that hands exist in the vulgar sense" than Berkeley denies that stones exist in the " vulgar sense." Moore ' s holding up two hands no more refutes Descartes than Dr. Johnson' s kicking a stone refutes ' Berkeley. Hence, Moore s argument is not an objection to the claim that external metaphysical knowledge (if there is any) is a priori . An epistemology for internal knowledge need not answer the question of skepticism to be acceptable, but it cannot be acceptable if it would in principle prevent us from learning the ultimate outcome of our struggle with philosophical skepticism . If , on the supposition that there is external metaphysical knowledge , it is a priori knowledge , then ' ' Quine s epistemology does just that . Quine s blend of empiricism and epistemic naturalism commits him to the continuity of empirical knowledge from the sciencesto their foundations and from their foundations to metaphysical speculation . Empirical science is our first and only philosophy. Furthermore , as Quine (1974, 2) says in The Rootsof " , the skeptical challenge is a challenge to natural science that Reference " arises within natural science. And , in " The Nature of Natural Knowledge ," Quine (1975, 68) says: I am not accusing the sceptic of begging the question ; he is quite within his rights in assuming science in order to refute science; this , if carried out , would be a straightforward argument by
Toward a Realistic Rationalism
203
reductio ad absurdum. I am only making the point that sceptical doubts are scientific doubts . But if this is so and if scientific knowledge is a posteriori , as it is on ' Quine s epistemology, then his epistemology makes it impossible for us to determine whether (A ), (B), or (C) is the case. Since the point is the same for each of these cases, let us consider (A ). If we construct an argument that refutes, say, the Cartesian skeptic, it constitutes metaphysical knowledge of why we are justified in claiming to know that there is an external world . As we have seen, such an argument must be a priori . Cartesian and Humean skepticism show, as we have seen, that , if there is an argument that can meet Cartesian or Humean skepticism , it cannot be one based on experience. Since experience is irrelevant , and since ex hypothesithere is an argument that refutes the Cartesian skeptic, then it is an a priori argument . But, since Quine ' s epistemology allows only a posteriori arguments, it would rule out our knowing that (A ) is the case. A priori arguments against skepticism seem promising only up to a point . Perhaps the most promising of them are the' ones that attempt to exploit the rational presupposition of the skeptic s argument . Skeptics cannot just express their doubts about our epistemic claims. They must provide an argument that we have to refute to show that we really do know what we think we do. But, in providing such an argument , skeptics assume linguistic and inferential principles which we can try to turn against them. Moreover, even though skeptical arguments arise in the area of external metaphysical knowledge , they presuppose internal principles , principles belonging to linguistics and logic . Since skeptics cannot disavow those principles without undercut ting their skeptical challenge, they have a commitment to internal knowledge that offers us at least an entering wedge . The question is which forms of skepticism are vulnerable to this move and how vulnerable are they. I suspect that the use of such internal knowledge is one way to refute the logical skepticism that ' Descartesbased on his conception of God ' s omnipotence : Descartess skeptical argument against the principle of noncontradiction is selfrefuting because it presupposes the principle . It is not clear whether there are other forms of skepticism that are vulnerable to this kind of response. The question is how far the skeptic' s commitment to internal knowledge goes. Skeptics have to express their skepticism in language and sustain it logically , but they do not have to assume substantive principles about nature. Since the Cartesian challenge to our knowledge of the external world , the Humean challenge to our knowledge of the laws of nature, and so on do not involve a potentially
204 Chapter6 self-defeating commitment to the standards that they challenge, it doesn' t look like they are vulnerable to this kind of response.4 We have considered a number of kinds of response to skeptical challenges and will consider another kind in the next section. None seemspromising . The upshot for our rationalist epistemology - and for rival epistemologies too - is the unsurprising conclusion that nothing much can be said at present to show that we have the external metaphysical knowledge we take ourselves to have. But, in the present context, this conclusion is of less consequencefor philosophy than it is frequently taken to be. As long as option (B) is not established, the foundations of the sciencesand metaphysical explication can continue to provide a skeptic-free environment for one set of philosophical concerns. Skepticism has often been a central concern to philosophers , but at least as often philosophers have had other central concerns. For Socrates, Plato, Aristotle , Locke, Leibniz , Frege, and Quine, skepticism was of relatively marginal interest. It is ironic that it is the difficulty of the issue of skepticism , in particular the difficulty of establishing (B), that leaves us free to pursue business as usual in the areas of scientific and internal philosophical knowledge .5 6.3 CaTnap's Criticism of the SyntheticA Priori Failure to capitalize on this relation between the difficulty of the issue of skepticism and the freedom we have to pursue philosophical investigation into questions of internal knowledge has led to a degradation of reason and the consequent creation of a situation where naturalism , , and positivism are seen as saving philosophy from itself . empiricism Hume ' s ([ 1739] 1958, 183) views that our reasoning about causes and effects derives from custom rather than from " the cogitative part of our natures" and that " school metaphysics" is nothing but " sophistry and illusion " obtain much of their force from such a degradation of reason. Viennese positivism - which substitutes " literal nonsense" for " sophistry and illusion " as the sin for which books of " school metaphysics" 4. Descartes's own response was that our belief in an external world that causes our inner experiences must be true , since, if it weren' t, then, per impossibile, God would have done evil . To block skepticism about the external world , we have to prove the existence of God. As if that were not enough of a problem , Nozick (1981, 201- 2) points out that such a proof is not sufficient . 5. Moreover, new developments in science can be expected to add new dimensions to old problems in the foundations of the sciencesand perhaps to raise new ones. Besides current problems in the foundations of scienceand metaphysics, we can expect that new results in the foundations of the sciences will stimulate new topics for metaphysical explications . Also , progress in metaphysical explication can very well change the picture in the foundations of the sciences.
Toward a Realistic Rationalism
205
would be burnt - is like Humean positivism in also gaining its force from an exaggeration of the scope of skepticism . The early Wittgenstein ([ 1921] 1961, sec. 6.51) wrote : Scepticism is not irrefutable , but obviously nonsensical, when it tries to raise doubts where no questions can be asked. For doubt can exist only where a question exists, and an answer only where something can be said. Given what we have said about the scope of skepticism at the end of the last section, skeptically inspired pessimism about rational metaphysics no longer makes the positivistic quick fix seem irresistible . Nor does skepticism bear on issues in the area of internal knowledge , such as the rationalist / empiricist issue. Hume ' s skeptical doubts about our knowledge of causes and effects in and of themselves provide no support for a positivist criticism of internal metaphysical theories or an empiricist view of scientific knowledge . Hume ' s argument for his his empiricist view that " [a]11the objects of human positivism assumes " reason or inquiry are either relations of ideas, which are intuitively or demonstrably certain, or matters of fact and existence, which are based on sensory observation or on records and memory of such observations . Given this assumption , his empiricist view can receive no support from his skeptical arguments . If rationalists are helpless against Hume ' s point that induction is neither intuitively nor demonstrably certain, then empiricists are helpless against Hume ' s point that arguments for induction based on experience are circular. If Hume ' s skeptical doubts were to sound the death knell for rationalism , they would sound it for empiricism as well . Thus, those doubts play no decisive role in the controversy between empiricists and rationalists , which is about the nature of internal knowledge in the sciences. Similarly , skeptical doubts can play no role in the realism / antirealism controversies in the foundations of the formal sciences and natural sciences. In chapter 2, we argued for the irrelevance of skepticism to the realism / antirealism controversy in the foundations of the formal sciences. Skepticism is also irrelevant to the instrumentalist claim that observational terms designate objects in nature but theoretical terms only provide convenient computing machinery. Here, too, skepticism cuts both ways : instrumentalists cannot use skepticism against the realist' s view of the theoretical vocabulary of science without having it turned against their realist view of the observation vocabulary. Given that skepticism has no implications for scientific and philo sophical issues in the area of internal knowledge , positivist claims have to be assessed on the basis of the internal standards appropriate to them without consideration of the role that positivistic semantic
206
Chapter6
doctrines were intended to play in saving us from skepticism . The ' skeptical doubts behind Wittgenstein s concern with solipsism in the T;oactatuscannot provide any basis for his positivist claim that metaphysical sentences are nonsense. The soundness of that claim would first have to be adjudicated within linguistics and the foundations of linguistics . ' Similarly , the soundness of Carnap s (1952) claim that statements like (1)- (6) are nonsense has to be evaluated on the basis of the appropriate linguistic standards. (1) Mathematical (logical , linguistic ) objects are abstract. (2) Mathematical (logical , linguistic ) objects are concrete. (3) Mathematical (logical , linguistic ) objects are fictions . (4) There is an external world . (5) The future will be like the past. (6) There are other minds . Since the claim that those English sentences do not have a sense is a linguistic claim like any other claim about English sentences, we can compare sentences (1H6 ) with genuine cases of nonsense sentences, such as (7) - (9), to determine whether (1H6 ) are sufficiently similar to them to count as nonsense too. (7) Quadruplicity drinks procrastination . (8) Colorless green ideas sleep furiously. (9) The number seventeenloves the taste of home cooking . But speakers of English will judge (1H6 ) to be nothing like (7)-( 9). Further, it is hard to see how a sentence like , say, (5) could have no sense while sentenceslike (10) and (11) have a sense. (10) The future of London will be like the past of London . ' (11) Switzerland s future will be like Switzerland ' s past. Hence, Carnap ' s claim is dubious linguistics . This argument against assimilating sentenceslike (1)-(6) to casesof semantic deviance like (7) - (9) is parallel to the argument in chapter 2 against the combinatorialist ' s attempt to give a multiform semantics for English to justify construing mathematical sentences like (12) as having a different logico- grammatical form from sentenceslike (13). (12) There are at least three large cities older than New York. (13) There are at least three perfect numbers greater than 17.
Toward a Realistic Rationalism
207
' Our argument here is another development of Burgess s (1990, 7) point that the question of what semantics is appropriate for mathematical " sentencesbelongs to " the pertinent specialist professionals . The question of what semantics is appropriate for metaphysical sentencesalso " " belongs to the pertinent specialist professionals . Once the former question is put in the hands of the linguists , it is independent of the epistemological issues of skepticism , and once the latter question is put in their hands, it is independent of the onto logical issue of realism. As a consequence, there is nothing to counterbalance the force of the robust linguistic intuition that sentenceslike (1)-( 6) are not meaningless in the manner of sentenceslike (7)-( 9). From the first expression of his positivism in his 1928 book Pseudoproblems in Philosophy(1969) to its mature expression in later works like ' " " Empiricism , Semantics, and Ontology (1952), Carnap s claim that sentenceslike (1)-( 6) are meaningless was based on his own homespun concept of meaningfulness . Meaningful sentencesdivide exhaustively into factual sentences, which have an appropriate empirical relation to the natural world , and formal sentences, which are nonfactual but reflect features of the framework concerned with expressing and manipulating sentences. There can be no evaluation of this conception on the basis of the results of investigations in linguistics , since it is simply ' a proposal of a framework (or class of frameworks ) in Carnap s (1952, ' 30- 33) own sense of that term . Carnap s critics - Quine , of course, but also Oxford philosophers like Strawson (1963, 503- 18) - pointed out that Carnap ' s claims to address substantive philosophical issues are " " empty without a reason to identify meaningfulness and meaning lessnessin his special sensewith meaningfulness and meaninglessness in natural language. " In " Empiricism , Semantics, and Ontology , Carnap (1952) repackages this distinction between the meaningful and meaningless in the termi " " " " nology of internal questions and external questions. The former are the old factual and formal sentences, and the latter are the old metaphysical sentences. External questions concern the acceptance of a linguistic or theoretical system as a whole , and hence they are meaningless because their answers are unverifiable on the basis of reasoning within a scientific system. Carnap introduced this new way of drawing the old distinction to explain why theoretical systems containing terms referring to abstract without commit ting the user to objects can be used in formal systems Platonist metaphysics. Carnap ' s explanation was that there is no commitment because the question whether to adopt a term within a theoretical framework can be answered on the basis of arguments internal to the system. As he sees it , the question of Platonist metaphysics, in contrast, is an external question about whether to accept the theoretical
208 Chapter6 system as a whole , and such questions make no sensebecausethere is no explicit framework within which to answer them. External questions are thus noncognitive , and decisions about them are practical matters, to be made on the basis of the uses to which one wishes to put the system. Hence, Carnap can have his cake and eat it : internal questions about terms referring to abstract objects are meaningful and can be employed in formal systems such as those described in Meaning and , while external questions such as the one at issue in the debate Necessity about mathematical realism are meaningless. Given its intended scope, Carnap ' s meaningful / meaningless distinction is itself external. In this case, however, it is not an assertion and its " acceptancecannot be judged as being either true or false." It is " a " pseudo- statement without cognitive content. It is a proposal that , as " Carnap (1952, 31) says, can only be judged as being more or less expedient", fruitful , conducive to the aim for which the language is intended . 6 The point extends to Carnap ' s conception of logic . His principle of tolerance, which he (1937, secs. 17, 52) has espoused from The Logical Syntax of Languageon, removes every consideration , even a constraint on explications of the notion of a valid consistency, as " consequence: In logic, thereare no morals. Everyone is at liberty to build " up his own logic , i.e., his own form of language, as he wishes. With this denial of the normative in semantics and logic , choosing a framework within which to state principles of logic is liberty hall . Even a perverse whim to be logically inconsistent must be indulged , since there is no possibility of drawing a line between the permit ted and the forbidden within Carnap ' s amoral logical world . Allowing a moral preference for consistency would lead quickly to the conclusion, disastrous for Camap , that logic is a normative discipline . But if there is nothing normative in semantics or logic , acceptanceof a language system in which metaphysical sentences are meaningless has no cognitive status and the language system has no normative force. But, in such a totally amoral world , there is no normative ground on which Carnap can stand when he moralizes against metaphysics. The principle of tolerance leaves everything up to the individual . But if everyone can make philosophical choices entirely as he or she sees fit , it is hard to see how Carnap ' s statements about metaphysics can amount to more than positivist propaganda . It hardly needs saying that 6. But if this is the status of his meaningful / meaningless distinction , it is not only that ' Carnap never intended it to fall within linguistics , but that it couldn t. Within linguistics , a distinction between the meaningful and the meaningless is a claim about linguistic fact, not a proposal that we are free to accept or reject on the basis of our desires (whatever they may be). In linguistics , distinctions between classesof sentencescan only be judged as correct or incorrect depending on whether or not they fit the grammatical facts.
Toward a Realistic Rationalism
209
propaganda provides no reason for denying that there is synthetic a priori knowledge . The closest Carnap ever comes to appreciating these problems is his ' (1956b, 233- 35) concern about Quine s objection that , without the investigation " of natural languages, the semantic intension concepts, " even if formally correct, are arbitrary and without purpose . But Car' nap doesn t really get it . He (1956b, 235) says, I do not think that a semantic concept, in order to be fruitful , must necessarily possessa prior pragmatical counterpart . It is theoretically possible to demonstrate its fruit fulness through its application in the further development of language systems. But this is a slow process. ' Carnap doesn t seem to see that the slower process of developing language systems does nothing to reduce the arbitrariness of semantic concepts and Carnapian language systems becauseit derives from the arbitrariness of the whole process of developing such systems under ' the dubious guidance of the principle of tolerance. He doesn t seem to see that Quine , Strawson, and his other critics were questioning the relevance of an entirely unconstrained process of developing language systems to resolving linguistic issues and dissolving philosophical controversies.7 6.4 Conclusion The linguistic turn can be characterized by the problem it set out to solve, the adversaries it intended to defeat, and the means it chose to solve its problem and defeat its adversaries. Its problem was synthetic a priori knowledge . Its adversaries were positions that based metaphysical theories of knowledge and reality on the existence of synthetic a priori knowledge . The means for solving its problem and defeating its adversaries was a critique of the linguistic , particularly the semantic, assumptions on which such positions grounded their claims for the existence of synthetic a priori knowledge in science and philosophy . The linguistic turn made no mistake about the problem it had to solve or the adversaries it had to defeat. Its failure was due to the inadequacy of the means it used. The revolutionaries underestimated 7. On linguistic realism, there is a more sweeping criticism of the verificationism underlying Camapian and other forms of positivism . If sentences are abstract objects, then grammatical properties of sentences are intrinsic features of the structures of abstract objects, and hence sentencesare meaningful or meaningless independently of our cognitive capacities and the conditions in the natural world that determine what is and what is not empirically verifiable .
210
Chapter6
the resources of metaphysics and overestimated their own linguistic criticisms . Their metaphysical adversaries were not restricted to positions like Kant' s that were strong epistemologically but weak ontologi ' or cally positions like Frege s that were strong onto logically but weak epistemologically . As I have argued in the course of this book , there is a metaphysical position that is strong both epistemologically and ontologically . As I have also argued here (but more fully in TheMetaphysics of Meaning), this position is also strong semantically, avoiding both the problems that Frege exploited in criticizing Kant' s analytic / synthetic distinction and the problems that Wittgenstein and Quine exploited in ' criticizing Frege s intensionalism . The different positions on the issue of synthetic a priori knowledge were driven by the same desire to provide an adequate understanding of philosophy in the light of the impact on traditional philosophy of developments in nineteenth- and twentieth -century science. Except for the late Wittgenstein , the different sides on the issue shared a respect for scientific achievements such as Cantor ' s set theory, Frege' s logic , and Einstein' s physics, the consequent disillusionment with Kantian and neo-Kantian metaphysics, and an appreciation of the need to provide a new explanation of the relation between scienceand philoso phy. The positions that originated with the early Wittgenstein and Schlick and culminated with the late Wittgenstein and Carnap offered positivistically based explanations on which systematic knowledge is scientific. Quine ' s naturalized epistemology offered the explanation that philosophy is natural science. These positions present a unified front only in opposing conceptions of philosophy on which philosophy is an autonomous discipline concerned with explaining synthetic apri ori knowledge . Starting from the same desire for an adequate understanding of philosophy in the light of the developments in nineteenth- and twentieth -century science but embodying precisely this beleaguered metaphysical of conception philosophy, realistic rationalism provides a quite different explanation of the relation between science and philosophy . Unlike the positivist explanation , it recognizes the legitimacy of philo ' sophical as well as scientific knowledge , and, unlike Quine s explanation , it maintains that philosophical knowledge is distinct from scientific knowledge . Not only does it not assimilate philosophy to science in order to legitimatize philosophical knowledge , it maintains , contra Quine ' s epistemic naturalism , that scientific knowledge is heterogeneous . Realistic rationalism holds a dualistic view about the of science, claiming that the formal sciences study abstract ontology and the natural sciencesstudy concrete and composite objects. objects The aim of the former is to prune the supposable down to the necessary,
Towarda RealisticRationalism 211 while the aim of the latter is to prune the possible down to the actual. Given this difference between knowledge in the formal and natural sciences, the former is synthetic a priori knowledge and the latter is synthetic a posteriori knowledge . Philosophy addresses onto logical and epistemological questions in the foundations of the sciencesand the foundations of the foundations of the sciencesthat the sciencesthemselves do not address. The relation between philosophy and the scienceshas both a vertical dimension on which philosophy attempts to understand the nature of the sciences and a horizontal dimension on which it attempts to understand aspects of the same reality studied in the sciences. On both dimensions , internal philosophical knowledge is synthetic a priori knowledge . Such knowledge is not the product of successful encounters with the skeptic . It is the product of the continuing dialectic among nominalists , conceptualists, realists, positivists , empiricists , and rationalists .
,