In his paper "Empiricism, semantics, and ontology" ( 1956), Carnap distinguishes two kinds of questions, and accordingly distinguishes scientific questions and philosophical ones. On the other hand, Quine proposes in his paper "Epistemology Naturalized" that epistemology should be a part of science ( 1994). In this paper, I would like to ask the question whether Quine's naturalized epistemology successfully eliminates the boundary between science and philosophy, like Carnap's distinction between 'internal question' and 'external question'. My analysis concentrates on Cherniak's criticism of Quine by way of "positive skepticism". My conclusion will be that even though Cherniak's argument has succeeded in causing a trouble with Quine's position, Cherniak does not show that positive skepticism is about the relationship between external world and conceptual scheme.
1. Carnap's external question
We can trace the origin of clear distinction between science and metaphysics in Kant's distinction between "concepts of understanding" and "ideas of reason" ( 1977). Concepts of understanding are a priori categories which is necessary to transform our perception into experience. For example, causation is a concept of understanding. Ideas of reason are, on the other hand, about things beyond our experience. The belief that "there is something real outside us" is one of the ideas of reason (77). Even though there is something beyond our experience in these ideas, "we are not at liberty to abstain entirely from inquiry into them; for experience never satisfies reason fully but, in answering questions, refers us further and further back and leaves us dissatisfied with regard to their complete solution" (92). Here is the role of metaphysics as an independent enterprise from science. Namely, metaphysics can clear up confusions in science by "a formal determination ... of the boundary of the use of our reason" (ibid.).
Carnap ( 1956) follows up Kant's distinction by distinguishing two kinds of questions. Internal questions are ones "of the existence of certain entities of the new kind within the framework" (206; emphasis in original). External questions are ones "concerning the existence or reality of the system of entities as a whole" (ibid.; emphasis in original), and Carnap identifies these questions with choice of linguistic frameworks. For example, the question "do unicorns exist?" is an internal question, while "do physical objects exist?" is an external one, and the latter is identical with the question about the adoption of physical object language. The answers to external questions are "a matter of decision" (207), and the criteria for judgement are expediency, fruitfulness and conduciveness to the aim for which the language is intended (214). Carnap's analysis of external questions is obviously motivated by verifiability principle; since external questions are not verifiable, no ordinary scientific way is applicable to them.
Thus, both Kant and Carnap contrast scientific questions with philosophical, metaphysical questions. Their reasons for the contrast are also similar, namely they think that philosophical questions go somewhat beyond our experience.
2. Development of Quine's view on philosophical questions
Quine's view on philosophical questions is radically different from Kant's and Carnap's. In this section, we trace the development of his view looking at three of his works, namely "Two dogmas of empiricism" (1953), Word and Object (1960), and "Epistemology naturalized" ( 1994).
2-1. Quine in "Two dogmas"
Before we can look at Quine's opinion in "Two dogmas of empiricism" (1953) about Carnap's external question, we need to summarize Quine's overall argument in this paper. Quine attacks two unwarranted presuppositions accepted by Carnap and other empiricist philosophers. First dogma is the distinction between analytic and synthetic sentences. He calls this distinction a dogma because the notion of analyticity is hopelessly inexplicable. Attempts to explicate analyticity using the notions of synonymy, definition, interchangeability, semantic rules etc. fail because these notions are no clearer than analyticity.
Verification theory of meaning seems to be able to solve the problem with analyticity by saying that "statements are synonymous if and only if they are alike in point of method of empirical confirmation or infirmation" (37), but verification theory of meaning presupposes another dogma of empiricism, namely reductionism. The basic idea of reductionism is that "to each statement, or each synthetic statement, there is associated a unique range of possible sensory events" whose occurrence confirm or disconfirm the statement (40-41). This is a dogma because "our statement about the external world face the tribunal of sense experience not individually but only as a corporate body" (41). 
Now, what happens if we abandon these dogmas? First, if we abandon analytic-synthetic distinction, we need to admit that every statement has "a linguistic component" and "a factual component" (36-37). If we abandon reductionism in addition, talking about these components in the truth of individual statements becomes nonsense. Rather, science as a whole should be associated with these components (42). Quine describes the resulting image vividly: "the totality of our so-called knowledge or beliefs... is a man-made fabric which impinges on experience only along the edges" (ibid.). Any statement in the fabric can be held true come what may if we make drastic enough change elsewhere, and at the same time no statement (including logical laws) is immune to revision.
Now, our question is, what is the consequence of this picture to Carnap's external questions? Quine is quite explicit on this matter: "ontological questions, under this view, are on a par with questions of natural science" (45). Namely, ontological statements are just another parts of the network, and can be revised because of empirical inputs (44). Quine's following remark is also illuminative: "Now Carnap has maintained that this is a question not of matters of fact but of choosing a convenient language form, a convenient conceptual scheme or framework for science. With this I agree, but only on the proviso that the same be conceded regarding scientific hypotheses generally" (45). Thus, not only both scientific and ontological statements are somewhat empirical, but also both of them are somewhat matter of convenience. I think this is a quite relativistic position, but we need to be careful to understand what he says. Since for Quine ontological questions are a part of the conceptual scheme, they cannot be equivalent to the choice of the conceptual scheme itself. In this sense, ontological questions are not 'external' at all for Quine (I will keep using the phrase 'external questions' to refer to these questions, though). Then what does he mean when he says that he agrees with Carnap? My interpretation is this. Since ontology is fairly central to our conceptual scheme, a change in ontology can cause a drastic change to the entire scheme. Of course change in more peripheral statements can also cause a drastic change. I think that Quine is talking about this drastic change when he admits the choice between conceptual schemes.
2-2. Quine in Word and Object
Quine develops the above holistic view on knowledge in Word and Object (1960). The most significant change from the view in "Two dogmas" is the combination of the above argument with Neurath's boat metaphor. Because of this change, the entire picture becomes conservative rather than relativistic. Let us start with summarizing relevant parts of the book, and then think about the status of external question in this view.
In the first chapter of Word and Object, Quine develops a behavioristic theory of language acquisition (1-25). We learn words by uttering sentences (including one-word sentences like "ouch") in certain situations and by society's approval of the utterance. Even though we learn language in this manner, our language cannot be reduced to protocol language or other sensory language, just by the reason Quine argued in "Two dogmas". Since people may have quite different internal experience in the same situation, the relation between language and experience can be different from an individual to another. In Quine's vivid metaphor, they are "like different bushes trimmed and trained to take the shape of identical elephants" (8).
Therefore we cannot determine truth value of each sentence separately. Rather, we need to consider the entire framework to determine truth value of sentences in the framework: "where it makes sense to apply 'true' is to a sentence couched in the terms of a given theory and seen from within the theory, complete with its posited reality" (24). We need to use law of least action, standard of simplicity, principle of familiarity, principle of sufficient reason along with experience in such consideration (19-21). Quine calls these considerations "scientific method" (23). This scientific method end up with a conservative view of science, as is expected from the use of law of least action etc.: "we continue to take seriously our own particular aggregate science, our own particular world-theory or loose total fabric of quasi- theories, whatever it may be" (24). In this sense, our language is like Neurath's boat: "Neurath has linked science to a boat which, if we are to rebuild it, we must rebuild plank by plank while staying afloat in it" (3). Quine even rejects Peircean notion of truth, namely truth as "the opinion which is fated to be ultimately agreed to by all who investigate" (Peirce  1958, 133). Quine's reason is that we cannot meaningfully talk about the ideal theory or distance from the ideal theory because of underdetermination of theories by evidence and indeterminacy of translation (Quine 1960, 23-24). Thus we are strictly confined to our own particular theories.
Then, where is the place of philosophers in this picture? Right after the Neurath's boat metaphor, Quine answers the question: "The philosopher and the scientist are in the same boat" (Quine 1960, 3). But what does he mean by this metaphorical statement? The following quotation is illuminating: "The philosopher's task differs from the others', then, in detail; but in no such drastic way as those suppose who imagine for the philosopher a vantage point outside the conceptual scheme that he takes in charge. There is no such cosmic exile" (275). Compare this quotation with the above quotation from "Two dogmas", in which Quine says that he agrees that external questions (and, he adds, internal questions, also) are matter of choosing conceptual scheme. Because now we need to keep the boat afloat, we lost the freedom to make a drastic change Quine was talking about in "Two dogmas".
This conservative turn is also reflected in the way Quine talks about philosophical questions: "What reality is like is the business of scientists, in the broadest sense, painstakingly to surmise; and what there is, what is real is part of that question. The question how we know what there is is simply part of the question" (22-23). The last arbiter for all these questions is "scientific method" (23). This statement is somewhat ambiguous to me. One interpretation is that traditional metaphysical questions, including Carnap's external questions, should be investigated in scientific method. Another way to interpret this is that philosophers should ask questions answerable in scientific method. In this latter interpretation, Quine is rejecting external questions themselves. The following passage suggests the former interpretation: "what distinguishes between the ontological philosophers' and all this [other non-fiction genres] is only breadth of categories.... On the other hand it is scrutiny of this uncritical acceptance of the realm of physical objects itself, or of classes etc., that devolves upon ontology" (275). Here, Quine clearly admits that there is a role to be played by Carnap's external questions. So what Quine claims in these quotations is that external questions should be investigated in scientific method.
2-3. Naturalized Epistemology
In the paper entitled "Epistemology naturalized" ( 1994), Quine goes one step further. In this paper, Quine observes that foundationalist attempts to establish theory of truth has failed, by the reasons Quine discussed in the above two works. To save epistemology from this situation, Quine proposes a new way to do epistemology. Now epistemology is not only conducted with scientific method, but epistemology is a science, or, more exactly, a part of psychology; "epistemology, or something like it, simply falls into place as a chapter of psychology and hence of natural science. It studies a natural phenomenon, viz., a physical human subject" (25). So, basically what we study in epistemology is relation between inputs to a human subject and outputs as descriptions of the external world by the subject. Since epistemology is supposed to provide basis for all kinds of science, including psychology, this makes epistemology a circular enterprise. But since foundationalist attempts have failed, maybe there is nothing wrong with circularity. Indeed, Neurath's boat metaphor implies this kind of circularity (26).
What happens to Carnap's external questions in naturalized epistemology? For example, let us consider the question, Is there external world? Soon we notice that this question is already answered before we start naturalized epistemology, because, as a part of science, naturalized epistemology starts from taking external world for granted. Thus, traditional philosophical questions just cease to exist. In Quine's own words in a later paper: "what evaporate is the transcendental question of the reality of the external world -- the question whether or in how far, our science measures up to the Ding an Sich" (Quine 1981, 22).
3. Evaluation of Quine's arguments
In the previous section, we have traced the development of Quine's argument, and especially the change in his view about Carnap's external questions. To summarize crudely: in "Two dogmas", both internal and external questions are matter of choice; in Word and Object, both of them are to be investigated with scientific method; and in "Epistemology naturalized", external questions cease to exist. In all the three works, Quine does not admit any significant difference between philosophical questions and scientific questions, and maintains that "external" questions are not external to the conceptual scheme at all.
Now it is time to evaluate. Does Quine succeed to show that there is no philosophical questions independent from science? In this section, first I summarize Cherniak's interesting argument from "positive skepticism" (Cherniak 1986). then I analyze Cherniak's argument and consider what Cherniak's argument really establishes.
3-1. Cherniak's "positive skepticism" argument
As we saw above, the doubt about the correctness of our entire conceptual scheme does not make sense to Quine (especially after Word and Object). We just accept it as it is. But according to Cherniak, naturalized epistemology may have a consequence about the nature of the conceptual scheme itself. Namely, our scientific picture about "man's place in nature" suggests that "it is deeply implausible that man can have a complete and completely correct theory of the universe" (Cherniak 1986, 124). This is the position Cherniak calls "positive skepticism" (132).
Cherniak uses two arguments for positive skepticism. First is concerning to Occam's Razor (125-127). From our knowledge about evolution, we can assume that our cognitive tool kit, like ability to judge about simplicity, is designed for dealing with "middle-sized objects like bananas and lions" (125). On the other hand, science have to deal with much bigger astrophysical objects and much smaller microphysical objects, and these objects obey quite different laws. Is there any reason to assume that our intelligence designed for middle-sized objects is also sufficient to study these things? Cherniak's reply is negative, thus he concludes: "it seems to us somewhat more likely than not that what we judge to be the simplest -- or most natural, plausible, elegant, or useful -- theory will not always be the correct one" (125; emphasis in original).
The second argument is from finitary predicament (127-129). We seems to have "fixed finite limits on our cognitive resources" (127). On the other hand, the universe seems to us to be indefinitely diverse and complex. Then, is there any reason to assume that the universe is finitely axiomatizable in the way we Homo sapiens can understand? Even if it is axiomatizable, it is still possible that the theory may be so complex that it is not manageable to us. Thus Cherniak concludes: "we may be unable to have complete theories, because some interesting true theories (as opposed to mere exhaustive enumerations, and so forth) are likely to be too complex" (127).
These arguments brings us to the conclusion that the scientific world view itself implies that our scientific world view has significant limit. Quine's argument shows that our conceptual scheme is indispensable to us, but "it's indispensability does not make it true" (129). Thus, "Quine does not thereby evade the question of the correctness (or, of course, of the completeness) of the total scheme" (129-130). External questions, in the sense of questions about the choice of conceptual scheme, have returned in naturalized epistemology.
3-2. What does Cherniak establish?
Even though Cherniak's argument makes an important point, we have to be careful what exactly he established. First of all, does his argument apply to the questions like "does external world exist?"? Both of his skeptical arguments are directed to what Hacking calls "realism about theories", namely the view that our scientific theories are (or will become) approximately true (Hacking 1983, 27). This kind of realism is quite distinct from the realism about external world itself, and I do not think positive skepticism can argue against the latter, metaphysical realism. We may not be able to have a correct theory about the universe, but this does not mean that our current theories are not about the universe, or we have no evidence about the existence of universe. When we start naturalistic epistemology we have given up these kind of questions. Therefore, Cherniak's argument has resurrected external questions, but Carnap's paradigmatic cases of external questions, like the question about the existence of physical object in general, are not resurrected.
Second, maybe Cherniak does not even establish that there are external questions. Quine rejects the correspondence theory of truth, because it does not make sense in his holistic view of language (Quine 1953, 79). Each word in the language does not correspond to something in the world in isolation. It has meaning only relative to the context it is used in the language. If we accept this view of language, I doubt if it makes sense to say that our word 'conceptual scheme' correspond to the conceptual scheme. Science suggests that it is implausible that our 'conceptual scheme' is complete and correct, but is this a statement about our conceptual scheme? To say that 'conceptual scheme' refers to conceptual scheme, we need to reestablish the correspondence theory of truth. Before we do this, we cannot say that "science itself, in motivating doubts about the future of progress, still distinguishes between mere correctness of game-playing within its own disciplinary matrix, and truth" (138). Or, more exactly, we can say this, but it does not mean what it is supposed to mean.
The above consideration suggests a possible reply from Quine to Cherniak. Namely, Cherniak's positive skepticism is just another part of the conceptual scheme, which has consequences on the truth values of other statements , not of the conceptual scheme itself. Some of these statements contain words like "conceptual scheme" or "external world", but mere use of these words does not mean that these statements are about the scheme itself or about the relationship between the scheme and external world. All these things takes place inside the scheme, nothing goes out the scheme. I think that Quine's "no cosmic exile" argument implies this reply.
However, this possible reply can make Quine's enterprise vacuous. Carnap argues that ontological questions are questions about choice of conceptual scheme, and Quine disagrees with him saying that we cannot get out of the scheme. But if Carnap's word "conceptual scheme" does not refer to conceptual scheme, is there anything wrong with Carnap's argument? Of course his identification of ontological questions with external questions is problematic, but at least talking about "external questions" should be totally legitimate. Moreover, we may be able to talk about, think about, and even attain a "drastic change" of "our conceptual scheme" without drastically changing the conceptual scheme. In short, if we interpret Quine's argument as I did above, Carnap and early relativistic Quine can return, with exactly the same linguistic forms and possibly with new interpretations. If Quine rejects this interpretation, then, of course, Cherniak's positive skepticism regains the power.
4. Concluding remarks
In this paper, I have traced the development of Quine's holistic view of language and the change in his view on Carnap's external questions. At the end of the development, external questions are totally eliminated. To evaluate Quine's position, I analyzed Cherniak's positive skepticism argument. I think that Cherniak has succeeded to cause a trouble to Quine. On the one hand, if Quine tries to reject positive skepticism using his "no cosmic exile" argument, then he has to allow the resurrection of Carnap and early relativistic Quine inside the conceptual scheme. On the other hand, if Quine does not use this argument, then he has to admit that there is a kind of external question about the correctness of the conceptual scheme itself.
 Even though Quine cites Duhem to support his claim, Duhem's argument for underdetermination is much more moderate than Quine's (Duhem  1954). First, Duhem's arguments are limited to specific system of hypotheses like Newton's optics, so Quine's talk about totality of knowledge is not supported by Duhem's argument (183-190). Second, when an experimental result contradicts with the prediction of the system of hypotheses, a physicist with "good sense" can choose a proper hypothesis to abandon (216-217). Duhem does not agree with Quine's relativistic conclusion.
 By the way, Quine asserts that Carnap agrees that analytic-synthetic distinction is essential to the distinction between internal and external questions (45-46). But in the footnote Quine cites, Carnap says nothing like that (Carnap  1956, 215, n.). All Carnap says is that Quine argues so, and the idea that there is no clear boundary between logical and factual truth "seems to deviate considerably from customary ways of thinking" (ibid.).
 Exactly speaking this is not a correct statement, because the connection between Duhem's thesis and Neurath's boat appears in "Identity, ostension, and hypostasis", published before "Two dogmas" (Quine 1953, 78-79). But, for the full appreciation of conservative implication of the boat metaphor, we need to wait until Word and Object. This is why I describe the difference between "Two dogmas" and Word and Object as a change.
 Even though Cherniak talks only about spatial scales, but the same argument seems to apply to temporal scales. Namely, since our intelligence is tuned for middle range time span, it may not be sufficient for the study of long time span -- like study of evolution. Of course, this move can undermine the argument itself, which depends on our current theory of evolution, but I think that that is the point of Cherniak's argument.
Carnap, R. ( 1956) "Empiricism, semantics, and ontology", reprinted in Meaning and Necessity enlarged edition. Chicago: University of Chicago Press.
Cherniak, C. (1986) Minimal Rationality. Cambridge: The MIT Press.
Duhem, P. ( 1954) The Aim and Structure of Physical Theory. translated from the second edition by P.P. Wiener. New York: Atheneum.
Hacking, I. (1983) Representing and Intervening; Introductory topics in the philosophy of science. Cambridge: Cambridge University Press.
Kant, I. ( 1977) Prolegomena to Any Future Metaphysics That Will Be Able to Come Forward as Science. translated by P. Carus, revised by J.W. Ellington. Indianapolis: Hackett.
Peirce, C. S. ( 1958) "How to make our ideas clear", reprinted in P. P. Wiener ed., Charles S. Peirce: Selected Writings (Values in a Universe of Chance). New York: Dover Publications.
Quine, W.v.O. (1953) From a Logical Point of View. Cambridge: Harvard University Press.
--. (1960) Word and Object. Cambridge: The MIT Press.
--. ( 1994) "Epistemology naturalized", reprinted in H. Kornblith (ed.) Naturalizing Epistemology 2nd edition. Cambridge: The MIT Press.
--. (1981) Theory and Things. Cambridge: Harvard University Press.