It seems imaginable that there might be intelligent creatures whose minds operate within a different framework of concepts and an unfamiliar mode of thinking from our own such that their experience of the world could be substantially remote from our own. In fact, we are never in short supply of creative thought experiments from philosophers on such totally alien conceptual schemes. For instance, imagine that the Martians are equipped with very different sensory apparatus and their minds operate in terms of a conceptual scheme bearing no similarity to ours. Would they be disappointed to find that we, the alleged intelligent creatures with whom they disparately want to establish contact, are just noisemakers? Does not this, and other similar cases, prove that totally disengaged conceptual schemes are at least theoretically possible?
Nevertheless, Davidson is not impressed at all with those thought experiments for a good reason: They cannot be verified semantically. Given, they may be possible; but anything is possible in a logical sense. The questions with which Davidson is concerned are how to verify semantically the existence of a conceptual scheme, and how to distinguish semantically between two alternative conceptual schemes. Presumably we need a semantically verifiable criterion for the existence of a conceptual scheme (a criterion of schemehood/languagehood for the Quinean conceptual schemes: whether some form of sounds, writings, and recorded symbols made by some creatures represent the use of a language at all) and a criterion for distinguishing between two alternative conceptual schemes. As Davidson discovers, after examining a few available criteria of identity of conceptual schemes allegedly associated with three different forms of conceptual relativism, the Quinean notion of a conceptual scheme simply cannot pass the verifiability test.
For an extreme conceptual relativist, an alien conceptual scheme could be extremely remote from ours to the extent of being ‘mutually unintelligible’ or ‘forever beyond our grasp’ and ‘rational resolve’.1 Consequently, cross-language understanding between those schemes is unattainable in principle. Davidson quickly points out that such a criterion simply does not make sense. In order to identify an alternative conceptual scheme, it has to be somehow intelligible to us to the extent that we can recognize it as a state of mind. Since such a scheme does not even make any sense to us, what justification could we possibly have for believing it to exist as a state of mind? What would stop us to deny it as a state of mind at all?
To respond, we need to clarify the key assumption of the argument, i.e., the unintelligibility to us of an alien conceptual scheme. It could mean that it is unintelligible contextually or hypothetically for us if we stick to our own framework of concepts and way of thinking; or it could mean that it is unintelligible categorically for us even if we try to learn to think in the way of the alien conceptual scheme by grasping its concepts. Davidson is highly ambiguous as to what he really means by ‘unintelligibility’ here. In one place, he seems to mean the former sense when he complains that conceptual relativism ‘seems (absurdly) to ask us to take a stance outside our own ways of thought’ (Davidson, 2001a, p. 40). But in other places he clearly means the latter interpretation since unintelligibility is something ‘forever beyond our grasp’. Going along with Davidson, I believe that extreme conceptual relativism in the latter sense is untenable, but not for the logical reason given by Davidson. Even if there exists an alien human language that cannot be made intelligible through interpretation by our semantic and conceptual apparatus, we can always make it intelligible, if it qualifies as a human language at all, by learning it from scratch as a child does. If it is unlearnable in principle, we have no other way but to conclude that it is not a human language (the language of bats, for example).
Davidson’s ambiguity over the notion of intelligibility reveals a fundamental flaw in his reasoning: He often confuses alleged extreme conceptual relativism with a more promising form of conceptual relativism. What Davidson’s argument shows us is that total categorical unintelligibility of another language, culture, or tradition is not an option. ‘It slays the terrifying mystical beast of total and irremediable incomprehensibility. But what we suffer from in our encounters between peoples are the jackals and vultures of partial and (we hope) surmountable noncommunication’ (Taylor, 2002, p. 291). To the best of my knowledge, not any of the conceptual relativists targeted by Davidson, i.e., B. Whorf, W. V. Quine, T. Kuhn, C.I. Lewis, P. Feyerabend, or P. Strawson, really holds such an extreme position. Most of them actually belong to the camp of radical conceptual relativists who contend that conceptual schemes or the languages associated with them can be and actually are, in many cases, radically distinct or massively different without any significant overlap, even to the extent of being incommensurable and leading to massive, even complete, communication breakdown. But such a failure of cross-language understanding is contextual and can be partially overcome, although it remains questionable whether full communication could be carried out without residue between two conceptually remote language communities. Davidson unfairly paints extreme and radical conceptual relativism with the same brush and treats them both as a case of complete failure of translatability (1984, p. 185; 2001a, p. 40). However, they are actually two totally distinct breeds; one holds that some cross-scheme understanding is in principle unattainable while the other believes it is attainable. It is necessary to make the distinction in order to defend radical conceptual relativism.
How do we effectively identify a radically distinct conceptual scheme? How do we distinguish and compare radically distinct conceptual schemes? The intelligibility criterion does not work here since intelligibility is not an issue for radical conceptual relativism. A different criterion is needed. The identification of the Quinean conceptual schemes with languages prompts us to consider cross-language translation. As Quine realized, it is ‘a measure of what might be called the remoteness of a conceptual scheme but what might better be called the conceptual distance between languages’ (Quine, 1981, p. 41). But what else can be used to measure such conceptual distance between languages better than the possibility and degree of intertranslation between them? (Quine, 1969, p. 5). Thus in the case of radical conceptual relativism, intertranslatability between languages becomes, as Davidson construes and attributes to the Quinean conceptual relativism, a criterion of conceptual schemes, as both a criterion of languagehood and one of different conceptual schemes. Lack of shared common parts between two radically distinct conceptual schemes makes intertranslation between the two associated languages impossible and leads to total translation failure. ‘No significant range of sentences in one language could be translated into the other’ (Davidson, 1984, p. 185). Thus, the complete failure of intertranslatability between two languages becomes a necessary and sufficient condition for difference of conceptual schemes associated with them, respectively. Thus we have the following criterion of identity and difference of conceptual schemes:
The criterion of intertranslatability: If a set of languages can be translated into one another, then they share the same conceptual scheme; otherwise they have different conceptual schemes. Accordingly, a form of activity represents the use of a language for an interpreter if and only if it could be translated, at least partially, into the interpreter’s own language.
The belief that there are some distinct alien languages untranslatable into a known language simply does not make sense if translatability is a necessary criterion of languagehood.2 If we cannot translate a language at all, how can we even be able to recognize it is a language, not just noises? Radical conceptual relativism apparently falls into the similar dilemma faced by extreme conceptual relativism. Of course, Davidson fully realizes that this objection in terms of translatability is too much of a ‘cheap shot’ that does not have any substantial persuasive power to conceptual relativists. They can quickly offer rebuttal that translatability is not, and should not, be a necessary condition for languagehood. To answer these possible objections, Davidson asks, could we make sense of there being a language that we cannot translate at all but that we could still recognize as a language by other criteria? In other words, could we have ‘a criterion of a languagehood that did not depend on, or entail, translatability into a familiar idiom’? (Davidson, 1984, p. 192) Where do we look for such a criterion? The opponents can easily supply a bundle of criteria of languagehood independent of translatability. At least interpretation in a loose sense or other criteria focusing on communicative functions will do the trick.3 But Davidson will not give in to these suggestions. For Davidson, the problem at hand is not whether there are other creditable criteria of languagehood available (I do not think Davidson would deny that there are such criteria) but whether Quinean radical conceptual relativism itself can supply such a criterion. The question becomes, could we derive a criterion of languagehood independent of translatability from Quinean scheme-content dualism?
As Davidson observes, Quinean scheme-content dualism is associated with two popular metaphors: A language either (Rl) organizes its empirical content through its referential apparatus or (R3) fits its content in terms of its descriptive apparatus. The organizing model Rl cannot be made intelligible and thus cannot supply the criterion needed. The fitting model R3 simply means that a language fits all potential experiences. But ‘the point is that for a theory to fit or face up to the totality of possible sensory evidence is for that theory to be true. ... [t]he notion of fitting the totality of experience ... adds nothing intelligible to the simple concept of being true’. Therefore, ‘our attempt to characterize languages or conceptual schemes in terms of the notion of fitting some entity has come down, then, to the simple thought that something is an acceptable conceptual scheme or theory if it is true’ (Davidson, 1984, pp. 193-4). Then the proposed criterion of languagehood becomes that a form of activity represents the use of a language for an interpreter if and only if it is largely true from the interpreter’s perspective. However, according to Davidson’s truth-conditional theory of meaning/translation, to be true and to be translatable always go hand in hand. Hence, the proposed truth-criterion slips right back to the translatability criterion: A form of activity represents the use of a language for an interpreter if and only if it is translatable into the interpreter’s language. Davidson concludes that Quinean conceptual relativism cannot supply a criterion of languagehood independent of translatability.
By comparison, the translatability criterion of alternative conceptual schemes appears to be logically coherent, if not tenable, from a third-person perspective, simply imagining that one is observing two competing languages, somehow recognized as such, which cannot be translated into one another. But remember that Quinean content-scheme dualism can be boiled down to the claim that a conceptual scheme different from the interpreter’s is largely true. Then ‘the criterion of a conceptual scheme different from our own now becomes: largely true but not translatable’ (Davidson, 1984, p. 194). But again according to Davidson’s truth-conditional theory of meaning/translation, truth cannot be divorced from translation. In conclusion, the translatability criterion of conceptual schemes, both as a criterion of languagehood and as a criterion of alternative conceptual schemes, is incoherent.
A radical conceptual relativist can respond to Davidson on two fronts, either to defend the Quinean translatability criterion—for a Quinean conceptual relativist, many fundamental assumptions of Davidson’s arguments are either unsound or questionable—or to separate radical conceptual relativism from the Quinean relativism as Davidson construes by removing the translatability criterion out of the equation. The first route is a well-worn path by many critics4 that I will not belabor except to remark on briefly, below. The second route, to me, is more effective and will be discussed in the next section.
One can and should ask the question: Why does translation have to be a necessary condition of languagehood? The basic assumption lurking behind such a rather strict requirement is that having justification for believing in the linguistic/conceptual character of someone’s behavior requires being able to understand their language/concepts. At first glance, the assumption seems to be quite innocent since if we cannot make another’s language intelligible, how could we even know whether a certain form of activity represents a linguistic act. However, whether such an assumption is acceptable depends upon the kind of cross-language understanding involved here. Among many different notions of cross-language understanding, the translatability criterion of languagehood assumes the translatability notion of cross-language understanding. That is, we can understand an alien language only by mapping it onto our own language.
What makes the case in hand more provocative is the fact that such a translatability notion of cross-language understanding is actually Davidson’s own stand, not just the Quinean conceptual relativist position as Davidson construes and criticizes. To see why, one needs only to recall Davidson’s truth-conditional account of understanding. According to it, to know the (Tarskian) truth conditions of the sentences of a language is both necessary and sufficient to understand it. But since truth and translation always go hand in hand, it amounts to a translatability notion of cross-language understanding: To be able to translate a language (in the sense of truth-functional translation) it is necessary to understand it. Ironically, what Davidson attacks is actually his own position. In other words, Davidson’s criticism itself is incoherent. More importantly, as I will argue in chapter 12, it is the knowledge of truth-value conditions (under what conditions a sentence has a truth-value), not the knowledge of truth conditions (under what conditions a sentence is true), that plays an essential role in cross-language understanding.
In addition, Davidson’s arguments against the translatability criterion do not work without help from his interpretation of Quinean content-scheme dualism and his own truth-conditional theory of translation; for only the combination of both can make the notion of a conceptual scheme, which is supposed to be largely true but untranslatable, incoherent. Recall that Davidson concludes that the distinction can be boiled down as follows: ‘Something is an acceptable conceptual scheme or theory if it is true’ or ‘largely true’. Being largely true from whose perspective? Davidson answers, from the interpreter’s notion of truth! This is simply another application of his principle of charity. As to be argued below, the principle is untenable. More to the point, as I will argue later, it is a false assumption that an interpreter would hold the sentences of an alien language embodying a radically distinct conceptual scheme as true.
Modest conceptual relativism recognizes the existence of the common part between two partially distinct conceptual schemes. Because of this, partial translation between the two associated languages is always possible. As such, translatability would not suit for the identity criterion in the case of modest conceptual relativism.
The basic supposition of modest conceptual relativism must be, Davidson contends, that the difference between two conceptual schemes can be identified by reference to the common part shared by both. The common part that is intertranslatable must represent shared concepts and beliefs/cognitive contents, and the other parts are presumed to constitute the difference between two schemes, namely, difference in meanings or concepts, not just difference in beliefs. Since those parts are not translatable, we can only hope to discover the difference in meanings and concepts through interpretation. In addition, since those parts are not shared, the interpretation cannot presuppose shared beliefs, meanings, or concepts. This is actually the case of so-called radical interpretation in which the interpreter has as evidential basis for interpreting a subject only the subject’s physical behavior in their environmental context. Thus, the identity criterion of alternative partially distinct conceptual schemes becomes:
The criterion of interpretability: An alien conceptual scheme is different from an interpreter’s when the interpreter can identify the difference in meanings or concepts in terms of radical interpretation.
Obviously, the above interpretability criterion assumes that the interpreter, in the case of radical interpretation, could decisively separate the speaker’s concepts from his or her beliefs and could determine, when the speaker thinks and speaks differently from the interpreter, whether the difference lies in his or her concepts rather than his or her beliefs. If such a difference were to occur between the two languages, ‘we should be justified in calling them alternations in the basic conceptual apparatus’ (Davidson, 1984, p. 188), not simply alternations in beliefs. But the assumption is, Davidson argues, not justified given the underlying methodology of radical interpretation—the principle of charity.
According to Davidson, a central source of trouble in interpreting others in the case of radical interpretation is the way beliefs and meanings conspire to account for utterances. There is no way to disentangle completely what aliens means from what they believe. We do not know what alien believe unless we know what they mean; we do not know what they mean unless we know what they believe. Therefore, we are always captured in such an interlock between meanings and beliefs. This is especially the case when we encounter an alleged alternative conceptual scheme (Davidson 1984, pp. 27, 142, 196). To break into such a vicious circle of radical interpretation, we have to hold one end steady while studying the other end. Following Quine, Davidson suggests that we should fix beliefs constant as far as possible while solving for meanings. In fact, this is actually not much of a choice for the interpreter since meanings are unobservable for the interpreter while some belief-related behaviors are observable within specific contexts. Of course, we cannot hold all beliefs constant for it is neither possible nor feasible. Fortunately, we do not need to do so. Among the behavioral facts ascertainable by observation prior to interpretation are notably the speaker’s beliefs of holding certain token-sentences to be true through consent when those token-sentences are actually true according to the interpreter’s truth conditions. For Davidson, it is the only legitimate evidential basis for radical interpretation. Therefore, what we need in radical interpretation is to fix a specific part of beliefs, namely, holding constant sentences of the speaker’s language to be true by the speaker from the interpreter’s perspective while deriving meanings from it.
The notion of ‘holding true’ plays the most essential role in Davidson’s project. But why does the interpreter have to determine whether the speaker holds a sentence in his or her own language to be true by attributing the interpreter’s own holding-true conditions of the sentence to the speaker? The answer lies in Davidson’s truth-conditional interpretation of the principle of charity, i.e., in radical interpretation, putting the speaker in general agreement on beliefs with the interpreter according to the interpreter’s point of view. But in order to complete a radical interpretation, only holding some of the speaker’s sentences true is not sufficient. Radical interpretation can only be accomplished through more generous ‘charity’, that is, through the interpreter ascribing to the speaker’s beliefs as largely true by the interpreter’s own lights. In other words, most sentences in an alien language that are held to be true by the speaker must be actually true according to the interpreter’s truth conditions.
A main reason for this strict interpretation of the principle of charity is this: Radical interpretation can only be accomplished if the interpreter knows what the speaker’s subject matter is. If the speaker’s saying is about anything at all, it is necessary for the speaker to possess a massive network of true beliefs about the subject matter in question. So the interpreter has to assume that the speaker’s beliefs are largely true of the speaker’s own environment as the interpreter believes them to be; otherwise if the speaker’s beliefs bear little or no relation to reality, then the interpreter could not even start the first step of radical interpretation, which is purely based on the speaker’s reaction to his or her immediate environment. Similarly, in order for the interpreter to identify what the speaker’s subject matter is, he or she has to share with the speaker a massive network of true beliefs relating to that subject matter (Davidson, 1984, pp. 155-70). Furthermore, for communication to be possible there must not only be a high degree of agreement in factual judgments, i.e., the shared beliefs of holding true, there must also be agreement in definitions, meanings, or concepts. Davidson thus concludes that radical interpretation requires the discovery of massive agreement between the interpreter and the speaker in both beliefs and concepts. This is why Davidson is convinced that we should prefer a theory of interpretation that maximizes the agreement and minimizes disagreement between the speaker and the interpreter. In order to understand others, we have to suppose that others are like us, to attribute our own beliefs to others, and to interpret others’ utterances based on our own beliefs.
If the charity principle is a methodological prescript forced on us in radical interpretation and not an option as Davidson strongly believes, nothing can force us to allocate a difference in respect to holding a sentence true to a disagreement in beliefs, rather than to a disagreement in concepts. ‘We could not be in a position to judge that others had concepts or beliefs radically different from our own’, for ‘when others think differently from us, no general principle, or appeal to evidence, can force us to decide that the difference lies in our beliefs rather than in our concepts’ (Davidson, 1984, p. 197). It is always possible to conclude, through a variety of semantic maneuvers, that others share our scheme but have different beliefs than ours. Therefore, as with other previous criteria, the interpretability criterion fails to identify the existence of alternative conceptual schemes.
Typical responses to Davidson’s above argument are to focus on its shaky foundation, namely, the principle of charity. I agree with those critics that the principle of charity is untenable, theoretically flawed, and practically unproductive.5 However, my interest here is not in the principle of charity in general, but in Davidson’s truth-functional interpretation of the principle, which will be discussed in detail below.
Davidson’s arguments from inverifiability against the Quinean notion of conceptual schemes are fully loaded with many unwarranted assumptions, theories, principles, or notions, such as his truth-conditional theory of meaning/translation, the translatability criterion of alternative conceptual schemes, the principle of charity, and the so-called fitting model of conceptual schemes. They are at least as shady and controversial as the target notion he wants to overthrow. Instead of examining those assumptions one by one, my strategy is to examine and challenge the two basic assumptions of the Quinean linguistic model on which Davidson’s arguments rely. One assumption is explicit; that is, the identification between conceptual schemes and sentential languages, which prompts Davidson to focus on sentences, truth-values of sentences, and sentential translation in distinguishing alternative conceptual schemes. His alleged fitting model of conceptual schemes, which is indispensable to his inverifiability argument, is also based on such an interpretation. Nevertheless, if the thesis of scheme-language identity is unfounded, the translation criterion goes out of the window and his argument of inverifiability becomes baseless. This is, I will argue, exactly the case. The other assumption is hidden, tacitly subscribed to by Quine, Davidson, and many others, conceptual relativists and anti-relativists alike, according to which we should focus on, in the discussion of alternative conceptual schemes, the notion of truth (whether truth is relative to a conceptual scheme or relative to a language) and redistribution of truth-values crossing two distinct languages. Application of Davidson’s truth-functional theory of interpretation/translation and his version of the principle of charity presuppose the second assumption. Its removal would undermine Davidson’s project.
Presumably, conceptual schemes are about concepts. However, the notion of concept is notoriously murky and slippery, often referring to a shadowy and problematic entity of some obscure sort, such as Platonic mind-independent abstract entity, disposition, mental ability to perform certain activities, or whatever composes the propositional content of one’s thought or belief. ‘The linguistic turn’ in the last century seems to point a way out of those confusions. After all, it is widely recognized that thinking is essentially a linguistic activity. Language is intimately associated with categorization and conceptualization, the two primary functions of concepts. Thinking in terms of language can certainly help us better grasp concepts, better detect conceptual variation, and better measure conceptual distance between alternative conceptual schemes. Through the association with concepts, conceptual schemes have commonly come to be associated with languages. For these reasons, many friends of conceptual relativism as different as R. Carnap, B. Whorf, and W.V. Quine have been thinking of the notion of conceptual schemes along such a linguistic line.
The exact relationship between conceptual scheme and language depends mainly upon two factors: What the linguistic counterparts of concepts are thought to be on the one hand and kinds of language involved on the other. Since concepts are linked to meanings, the primary linguistic vehicle of meaning would presumably be the primary bearer of a conceptual scheme. In the first half of the twentieth century, we witnessed an important reorientation in semantics, seen in Frege’s and Russell’s works as well as in the verification theory of meaning of logical positivism, whereby the primary vehicle of meaning came to be seen no longer in the term but in the sentence. Along with such a reorientation in semantics came an ontological shift from concepts to sentences as the primary elements of conceptual schemes. Through the link to sentences, conceptual scheme becomes more closely connected to language. This is partly why R. Carnap started to speak of linguistic framework when others talked about conceptual framework/scheme.
However, Carnap’s linguistic framework, which mainly refers to the meaning postulates of a language or theory, is still not a language itself. Quine pushes us further down the path to assimilate conceptual scheme into language. Quine contends that a conceptual scheme is not merely associated with a language, but is, rather, identical with it. Based on Quine’s holistic semantics, even in taking individual sentences as the unit of meaning, we have drawn our grid too finely since ‘our statements about the external world face the tribunal of sense experience not individually but only as a corporate body’ (Quine, 1980, p. 41). Instead, the unit of linguistic meaning with empirical significance is neither terms nor sentences, but the whole language. In some sense, once meaning is linked to language as a whole, meanings themselves, as obscure intermediary entities, may well be abandoned. One might wonder whether the notion of conceptual schemes should go by the wayside along with the dissolution of meaning and its counterpart, the concept, in the language. But Quine, as an empiricist who is unwilling to part with the scheme-content dichotomy (Quine, 1981, pp. 38-40), continues to use the term ‘conceptual schemes or frameworks for science’ and regards it as ‘a tool’ or ‘a device for working a manageable structure into the flux of experience’. Thus, ‘science has its double dependence upon language and experience’ although this duality is not significantly traceable into the duality of analytic sentences used to describe a scheme on the one hand and synthetic sentences used to describe experience (the dogma of the analytic-synthetic distinction) on the other (Quine, 1980, pp. 42-6). Thirty years later, Quine, in response to Davidson’s criticism, admitted ‘where I have spoken of a conceptual scheme I could have spoken of a language. Where I have spoken of a very alien conceptual scheme I would have been content, Davidson will be glad to know, to speak of a language awkward or baffling to translate’ (1981, p. 41). Considering the possibility that not all languages are conceptually distinct, for Quine, a conceptual scheme eventually becomes a set of languages that share the same conceptual make-up, whatever it may be.
Furthermore, in Quine’s mind, the language identical with the conceptual scheme is not a language in a special technical sense, such as the logical positivist theoretical language, but rather ‘ordinary sentential language, serving no technical function’ (Quine, 1981, p. 41). Obviously not all declarative sentences in an ordinary language are true. However, for a conceptual scheme to organize experience and face reality its sentences have to be true, at least from the speaker’s perspective. Therefore, a conceptual scheme eventually becomes identical with a sentential language with all its sentences held to be true. This is exactly the way in which Davidson construes the Quinean notion of conceptual schemes: ‘We may identify conceptual schemes with languages, then, or better, allowing for the possibility that more than one language may express the same scheme, sets of intertranslatable languages’ (Davidson, 1984, p. 185).
While thinking in terms of language may promise to clarify our understanding of the notion of conceptual schemes, it is not immediately clear how that clarification should be carried out. In other words, while it is relatively unproblematic and widely accepted now that conceptual schemes are and should be somehow associated with languages, it is controversial as to how to associate conceptual schemes with languages; since there are certainly many different ways in which a conceptual scheme could be connected to a language. To identify a conceptual scheme with a sentential language is certainly not the right way.
Which kind of language do Quine and Davidson have in mind when they identify a conceptual scheme with a sentential language? Strictly speaking, a notion of sentential language itself does not make much sense. Any linguist would tell us that it is obviously misguided to think of a language as a totality of sentences, not to mention as a totality of all declarative sentences held to be true. A language is a grammar and a vocabulary, not just sentences. But let us suppose, for the sake of argument, that the notion makes sense. What kinds of sentential languages then? Among the advocates of conceptual relativists identified by Davidson, Whorf exclusively focuses on natural languages while Kuhn and Feyerabend primarily concentrate on scientific languages. In his criticism, Davidson mixes and treats them both as sentential languages held to be true by their speakers without further distinction. In Quine’s naturalized epistemology, it is ‘the total science’ or a network of comprehensive scientific theories that is supposed to be a conceptual scheme facing the totality of experience. Thus, we have two possible candidates for sentential languages that Quine and Davidson have in their mind: natural language and scientific language.
A natural language per se such as English or Chinese is in no sense a conceptual scheme; otherwise, it would be a bizarre way to construe the notion of conceptual schemes. Does any conceptual relativist seriously think that all Chinese would inherit a different conceptual scheme from that of all English simply because they speak different natural languages? To put it another way, in what sense is the Chinese language throughout its 5,000 years of history a conceptual scheme and over time the same conceptual scheme? A natural language is not a theory either. A natural language like English does not schematize experience or even metonymically predict or fit reality. Although part of a natural language, i.e., its grammar, does in some sense determine the logical space of possibilities (Whorf, 1956), it is the theoretical assertions made in the language that predict and describe reality and in so doing assert which logical spaces are occupied, i.e., which logical possibilities are actualized in the world. In addition, a natural language as such is not even a worldview as many are convinced. ‘If every language is a view of the world, it is so not primarily because it is a particular type of language (in the way that linguists view languages) but because of what is said or handed down in this language’ (Gadamer, 1989, p. 441). What constitutes the so-called worldview proper is not natural language as language, whether as grammar or as lexicon; the worldview consists in the coming into language of what has been handed down from the cultural tradition associated with the language. Only in this sense do we say that a language-view is a worldview.
Many have treated a scientific language as a conceptual scheme. Although it is more closely related to a conceptual scheme than a natural language is, a scientific language construed as the totality of all its sentences is not a conceptual scheme. A conceptual scheme is, briefly put, supposed to be a conceptual framework which schematizes our experience in terms of its metaphysical presuppositions of existents, states of affairs, modes of reasoning, and categorization to form a theory. Many parts of a conceptual scheme, such as a categorical framework, usually a lexical structure of a scientific theory, are simply not a set of sentences, but rather a categorical system. In addition, a conceptual scheme, which serves as the conceptual framework of a theory, cannot in itself be the theory or the language expressing the theory. Nor will it improve matters to stipulate that a conceptual scheme is the totality of sentences held to be true by its speaker. A conceptual scheme does not describe reality as the Quinean fitting model R3 suggests; it is rather the theory it formulates that describes reality. A conceptual scheme can only ‘confront’ reality in a very loose sense, namely, by becoming in touch with reality in terms of a theory. Therefore, a conceptual scheme cannot be said to be true or largely true. It is simply not the bearer of truth. Only the assertions made in a language and a theory couched in the language can be said to be true or largely true.
To treat a conceptual scheme as a sentential language is to confuse language with language scheme, i.e., the conceptual core of language. When Whorf compares a natural language to a scientific theory, he actually means that it is the grammar of a natural language (construed as the rules for the use of expressions to determine what states of affairs are expressible and what expressions make sense) that functions just like a scientific taxonomy. Grammar or taxonomy fixes the logical space of possibilities that the world may or may not occupy, in as much as grammar determines what it makes sense to say. Any sentences couched in the language with that grammar or taxonomy will have truth-values, and they are true or false depending upon whether things are as they are asserted to be. It is exactly the role of the later Kuhn’s notion of lexical structure played out in a scientific theory. Similarly, for Carnap, a linguistic framework is not a language, but mainly refers to the meaning postulates of a scientific language. We can call those different conceptual apparatus of a language the language scheme. The language scheme of a scientific language is the conceptual scheme of the corresponding theory. Strictly speaking, neither a language nor a theory can be a conceptual scheme.
If a conceptual scheme were identical to a sentential language, construed as the totality of sentences held to be true by its speaker, then it would seem to be natural to deal with a conceptual scheme by the notions of truth and meaning in Tarski’s style. This is exactly the strategy used by Davidson in his inverifiability argument against the notion of conceptual schemes. The Tarskian notions of truth and meaning are based on the classical bivalent logic, in which a declarative sentence has a definite truth-value; that is, it can only be either true or false and cannot be neither true nor false (having no truth-value at all). If this is so, why do we not distinguish alternative conceptual schemes in terms of the redistribution of truth-values?
According to Davidson’s reading of Quinean content-scheme dualism, especially the fitting model R3, this dualism can be boiled down to the claim that a form of activity represents the use of a language for an interpreter if and only if it is largely true from the interpreter’s perspective. Davidson finds that, similar to this truth-value criterion of languagehood, a criterion of alternative conceptual schemes lurks behind conceptual relativists’ rejection of a sharp analytic-synthetic distinction. For many philosophers of science, Kuhn and Feyerabend in particular, to give up a sharp analytic-synthetic distinction as basic to the understanding of language is to give up the idea that we can clearly distinguish between theory and language/meaning. When switching from one theory to another, with the change of meanings of the terms that are determined by some principles or meaning postulates of the corresponding language, a sentence containing those terms, which were accepted as true within an old theory (‘motion’ in Aristotelian physics), could come to be accepted as false within a new theory (Newtonian dynamics). Davidson concludes:
We may now seem to have a formula for generating distinct conceptual schemes. We get a new out of an old scheme when the speakers of a language come to accept as true an important range of sentences they previously took to be false (and, of course, vice versa). (1984, p. 188)
Quine clearly thinks along the same line in terms of his holistic picture of scientific development. For Quine, any scientific theory is like a field of force with experience as its boundary conditions, and the total field is underdetermined by the totality of all possible experience. A new theory could emerge from an old one by making sufficient adjustments in the interior of the field. After readjustments ‘truth values have to be redistributed over some of our statements’ (Quine, 1980, p. 42). It is conceivable that if an old theory is undergoing a radical readjustment to the extent that truth-values over sentences have to be redistributed in a systematic way, then the new theory that emerges is radically distinct from the old one. In this case, we can say that two theories embody two different conceptual schemes. Thus we have reached a truth-value criterion of identity of conceptual schemes independent of the translatability criterion, implicit in the Quinean conceptual relativism:
(TV) Two conceptual schemes or languages differ when some substantial sentences of one language are not held to be true in the other in a systematic manner.
In other words, the difference between two conceptual schemes is semantically signified by the redistribution of truth-values over the sentences of two languages that embody the conceptual schemes.
The above truth-value criterion of conceptual schemes is not just different and independent from, but is actually more fundamental than, the translatability criterion. The latter is in fact a logical consequence of the former. Because of a systematic redistribution of truth-values from one language to another, the truth-preserving translation between them can hardly be carried out, and is even impossible in principle. This is why Davidson insists that we cannot get around the translatability criterion and that untranslatability is a necessary condition for distinguishing one conceptual scheme from another.
We can dig still deeper into a tacit assumption behind the Quinean truth-value criterion of conceptual schemes. Redistribution of truth-values over the sentences between two competing languages presupposes that although the speaker and the interpreter in discourse may assign opposite truth-values to some sentences in the other’s language, they agree that all sentences in the alien language are either true or false. That means that they agree on the truth-value status of these sentences. Thus, there is no truth-value gap between two languages:
(TF) Most sentences in the speaker’s language are either true or false from the point of view of the interpreter’s language no matter how disparate the two languages are.
Davidson strengthens the grip of the truth-conditional claim of conceptual schemes further to require that, according to his truth-functional interpretation of the principle of charity, most sentences of an alternative conceptual scheme not only have truth-values, but also have to be true, from both the viewpoints of the speaker and the interpreter. That means the speaker and the interpreter must and actually do share the same notion of truth and the same truth conditions if understanding is possible. This in turn presupposes the above assumption that the speaker and the interpreter have a shared belief of the truth-value status of the sentences in each other’s language. All the sentences in an alien language are either true or false from the interpreter’s point of view. Consequently, the possibility of the occurrence of a truth-value gap between two alien languages is excluded a priori as long as we stick to the principle of charity in interpretation. ‘If we cannot find a way to interpret the utterances and other behavior of a creature as revealing a set of beliefs largely consistent and true by our own standard, we have no reason to count that creature as rational, as having beliefs, or as saying anything’ (Davidson, 1984, p. 137).
As I will argue extensively in the rest of the book, my major reservation with the Quinean notion of conceptual schemes is not just the many theoretical difficulties it faces, but rather with its basic assumption (TF); for it does not square with observations of many celebrated conceptual confrontations between opposing conceptual schemes revealed in the history of natural sciences and cultural studies, especially those under the name of incommensurability. These familiar conceptual confrontations are, to me, not confrontations between two conceptual schemes with different distributions of truth-values over their assertions, but rather confrontations between two scientific languages with different distributions of truth-value status over their sentences due to incompatible metaphysical presuppositions. The advocate of an alien conceptual scheme not only does not hold the same notion of truth as ours, but also does not agree with us on the truth-value status of the sentences in question. These scheme innovations, at bottom, turn not on differences in truth-values, but on whether or not the sentences in the alternative conceptual scheme have truth-values. We simply cannot find the agreement taking ‘the form of widespread sharing of sentences held true’ between two incommensurable languages.
Conceptual schemes and contents come in a pair in scheme-content dualism but could be treated separately. In his arguments of inverifiability, Davidson focuses on the scheme side of the dualism. Such a strategy turns out not to be successful for at least two reasons. First, it is verificationist in nature. The conclusion Davidson is trying to reach is not that alternative conceptual schemes absolutely do not exist, but rather that it is meaningless or unintelligible to assert that they do in that we cannot semantically verify them. The critics, of course, could challenge such verificationist tactics: From the fact that there is no reliable way of telling whether something is the case, it does not follow that it is not the case, nor it is unreasonable to believe that it is the case. In any case, why do we have to hang our hat on the post of verificationism? Second, it only applies to one kind of conceptual scheme, namely, Quinean sentential language, but leaves other possible kinds such as S1 untouched.
Fortunately, both sides do not need to remain deadlocked on the issue of verificationism since there is another way in which conceptual relativism can be made intelligible without appealing to any verifiable identity criteria of conceptual schemes, that is, scheme-content dualism. As Davidson himself acknowledges, ‘If we could conceive of the function of conceptual schemes in this way, relativism would appear to be an abstract possibility despite doubts about how an alien scheme might be deciphered’ (2001a, p. 41). Besides, a conceptual scheme, no matter whether it is SI, S2, or any other possible bearer, can make sense only if there is some content to schematize. A dismantlement of scheme-content dualism would render any form of conceptual scheme baseless.
Even before Kant, we have been accustomed in empiricist tradition to ‘split the organism up into a receptive wax tablet on the one hand and an “active” interpreter of what nature has there imprinted on the other’ (Rorty, 1982, p. 4). ‘Since Kant, we find it almost impossible not to think of the mind as divided into active and passive faculty, the former using concepts to “interpret” what “the world” imposes on the latter’ (Rorty, 1982, p. 3). In reference to scheme-content dualism, the empirical contents schematized by our mental conceptual schemes in Kant’s model would be either the world-as-it-is or what the world imposes upon us as ‘given’, i.e., unconceptualized, ineffable raw physical thrust of stimuli upon our organs, passively received through our two sensible intuitions. Both Davidson and Rorty are convinced that such a sharp Kantian ‘given-interpretation’ distinction must be the foundation of any workable scheme-content dualism. That means that the empirical contents of a conceptual scheme have to be either uninterpreted ‘reality (the universe, the world, nature)’ outside of the mind, or ‘experience (the passing show, surface irritations, sensory promptings, sense data, the given)’ within.6 For Davidson, the notion of ‘experience’ E in the scheme-experience distinction has to be totally and purely experiential, scheme-free, and universal for all human beings, as we have defined as El in Quinean scheme-content dualism. The notion of ‘the world’ W used in the scheme-world distinction cannot be, as Rorty tries to convince us, the world as we have been experiencing it, full of familiar objects. Instead, it ‘must be the notion of something completely unspecified and unspecifiable—the thing-in-itself (1982, pp. 14-17). The world or experience (as empirical contents of schemes) is, putting it metaphorically, organized or interpreted by different conceptual schemes to produce different ideologies, theories, or beliefs about the world (as cognitive contents of schemes). To describe them fully here:
(El) thin experience (experience narrowly construed): sensory experience independent of all conceptual or linguistic schemes, theories, and interpretations, or ‘the sensuous given’ imposed from the world upon our receptive minds, i.e., sense data or ‘uninterpreted sensations’, including Kant’s ‘passively received intuitions’, Locke’s ‘simple ideas’, Hume’s ‘impressions’, and Quine’s ‘surface irritations’ or ‘sensory promptings’.
(Wl) the world-as-it-is: the world outside of any possible conceptualization, like the Kantian ‘thing in itself, which is unnamable, unspecifiable, and thereby an ungraspable physical thrust of stimuli or neutral materials.
These two possible empirical contents constitute the core of Kantian scheme-content dualism. Kantian scheme-content dualism, Davidson and Rorty argue, is presupposed by any notion of conceptual schemes—or more precisely, what Rorty calls ‘the Kantian notion of “conceptual framework”—the notion of “concepts necessary for the constitution of experience, as opposed to concepts whose application is necessary to control or predict experience’” (Rorty, 1982, p. 5). Clearly, Quinean scheme-content dualism is a version of Kantian scheme-content dualism.
Davidson urges us that the above Kantian scheme-content dualism cannot be made intelligible and defensible, which is actually one more instance of the various harmful dualisms that are part of the Cartesian and empirical philosophical legacy, i.e., the third dogma of empiricism.
In what sense is scheme-content dualism the third dogma of empiricism? To answer this, we need to identify what Davidson has in mind in referring to ‘empiricism’.7 Empiricism can be construed as a theory about the formation of beliefs/knowledge, and especially, how the mind acquires its cognitive content about the world. This version of empiricism, as a theory of the contents of the mind, has been traditionally associated with the representational theory of mind/knowledge that claims that our mental states, including ideas and beliefs, represent aspects of the world. On the other hand, empiricism can be understood as a theory about the justification of belief/knowledge. Empirical foundationalism, which appeals to some sort of universal sensory experience (usually sense data or the given) as the ultimate evidence or final foundation of our knowledge or warranted beliefs, has been a dominant theory of epistemic justification. The dominant empiricist theory of truth, namely, the correspondence theory, is a hybrid of the representational theory and foundationalism.
Kantian scheme-content dualism is seen as the third dogma of empiricism by Davidson because it is either a logical consequence or a presupposition of the two versions of empiricism. The scheme-world distinction, usually associated with the fitting model R3, is nothing but another version of the representational theory of mind/knowledge; for both presuppose ‘a theory-neutral reality’ on the one hand and a mental entity (ideas or conceptual schemes), which represents or corresponds to the reality, on the other. For many conceptual relativists, a conceptual scheme is not just a convenient way of ‘saving the phenomena’, but rather a mental framework for facing, fitting, or representing reality. J. Searle, for example, regards conceptual schemes as a subspecies or central conceptual core of ‘system of representation’ of a language-independent reality. By comparison, sense data or the given in the scheme-given distinction, usually associated with the organizer model Rl, is commonly thought of as the neutral raw materials from which our beliefs about the world are formed and justified. Such an ultimate source of beliefs is carved out and their foundation is secured by the scheme-given distinction. In this sense, both empiricism as a theory of justification and that as a theory about mental contents presuppose the scheme-given distinction. Furthermore, as Davidson notes, a common feature shared by both empiricism and Kantian scheme-content dualism is that they impose epistemological intermediaries between the mind and the world, that is, our beliefs and the world consisting of familiar objects our beliefs are about. More precisely, the scheme-world distinction turns schemes into intermediaries between the mind and the world, while the scheme-given distinction introduces two layers of intermediaries, sense data/the given from the content side as one layer and schemes from the scheme side of the distinction as the other. If one thinks these intermediaries are necessary or inevitable for us to understand the mind-world relationship, as Davidson believes that conceptual relativists do, one falls prey to the third dogma.8
Why is such a third dogma so harmful and in need of rejection? A simple answer from Davidson is, those imposed intermediaries between the mind and the world, no matter whether it is sense data/the given to be organized or schemes doing the organizing, simply do not play any epistemological role in either determining or justifying the contents of our beliefs about the world.
Let us start with the scheme-given distinction presupposed by both empiricism of belief formation and that of belief justification. Both Davidson and Rorty discover that the scheme-sense-data distinction is deeply rooted in the Kantian ‘picture of the mind as a passive but critical spectator of inner show’, the active faculty of the mind using schemes to ‘interpret’ what the world imposes on the passive faculty, i.e., the given. What are imposed onto the mind via its receptive faculty, i.e., sense data/the given, are purely private, subjective ‘objects of the mind’, which not only supply neutral raw materials in formation of our mental contents, but also constitute the ultimate evidence for our beliefs about the world.9 Such a Kantian model of a picture of the mind, according to which the subjective (‘experience’) is the foundation of objective empirical knowledge, sanctions the search for an empirical foundation for justifying our beliefs outside the scope of the totality of beliefs, by grounding them in one way or another on the testimony of the senses, i.e., El. Davidson believes that such a foundationalist view of justification is essentially incoherent. Confronting the totality of one’s beliefs with the tribunal of experience simply does not make sense, ‘for of course we cannot get outside our skins to find out what is causing the internal happening of which we are aware’ (Davidson, 2001b, p. 144). ‘Our beliefs purport to represent something objective, but the character of their subjectivity prevents us from taking the first step in determining whether they correspond to what they pretend to represent’ (Davidson, 2001a, p. 43). In fact, ‘nothing can count as a reason for holding a belief except another belief. Davidson argues that we should abandon the search for a basis for knowledge outside the scope of our beliefs, because ‘empirical knowledge has no epistemological foundation, and needs none’.
On the other hand, Davidson reminds us, the existence of the given and its role in belief formation has been challenged by a revised view of the relation of mind and the world. For instance, according to the causal theory of meaning, meanings cannot be purely subjective or mental; instead, words derive their meanings from the objects and circumstances in whose presence they were learned. Similarly, the truth-conditional theory of meaning tells us that our sentences are given their meanings by the situations that generally cause us to hold them as true or false. In general, all states of mind, like beliefs, thoughts, wishes, and desires, are identified in part by the social and historical contexts in which they are acquired. The fact that the states of mind are identified by causal relations with external objects and events convinces us that sense data or the given plays no epistemological role10 in determining the contents of the mind, no matter whether it is belief, meaning, or knowledge.
Therefore, Davidson concludes, the scheme-given distinction is a deep mistake because ‘an adequate account of knowledge makes no appeal to such epistemological intermediaries as sense data, qualia, or raw feels’. ‘Epistemology has no need for purely private, subjective “objects of the mind’”. Since scheme and content come as a pair, the notion of scheme interpreting and organizing the given goes by the wayside with the given (Davidson, 2001a; 2001b, pp. 141-6).
What about the scheme-world distinction associated with representationalism and the correspondence theory of truth? Could it remain intact? Do we still need schemes to form our beliefs about the world in order to have true beliefs? Davidson’s answer is negative. If sense data or the given (El) cannot justify whether a belief is true, neither can the world (Wl) since only beliefs can justify beliefs. Fortunately, although no epistemological justification can be given, nor is it needed for truth, we have a good reason, not a form of evidence, for supposing that a vast majority of our beliefs are true according to Davidson’s theory of radical interpretation based on the principle of charity. In this sense, ‘beliefs are in nature veridical’ (Davidson, 2001b, pp. 146-53). Davidson is convinced that beliefs are true or false, but they not only do not need any epistemological justification, they also do not represent anything. ‘Nothing, however, no thing, makes sentences and theories true: not experience, not surface irritations, not the world, can make a sentence true’ (Davidson, 1984, p. 194). There is no need at all to introduce the world (Wl) and schemes that face or fit the world since truth does not consist in a confrontation between what we believe and reality. Without the scheme-world distinction, the representational theory of perception and its twin sister, the correspondence theory of truth, are undermined.
In addition, Davidson warns us that the epistemological intermediaries imposed by the third dogma are not just impotent, without any significant epistemological role to play in the formation and justification of beliefs and knowledge, but are harmful. They cause more problems than they provide the answers that they are supposed to supply. The major motivation behind introducing sense data or the given into epistemology starts with a genuine concern about the relationship between the mind and the world: How can we be sure that our beliefs faithfully represent the world or the reality out there? Kantian scheme-content dualism has been frequently presented as a response to such an epistemological quandary by securing our grip on the world through introducing some unassailable ‘given’ as ultimate evidence of justification. However, Davidson argues that instead of finding any solace for our epistemological anxieties, ‘introducing intermediate steps or entities into the causal chain, like sensations or observations, serves only to make the epistemological problem more obvious ... since we cannot swear intermediaries to truthfulness’ (Davidson, 2001b, p. 144). ‘The disconnection creates a gap no reasoning or construction can plausibly bridge’, which prevents us from holding directly unto the world. Consequently, ‘idealism, reductionist forms of empiricism, and skepticism loom’ (Davidson, 2001a, p. 43).
The conclusion is obvious: ‘We should allow no intermediaries between our beliefs and their objects in the world’ (Davidson, 2001b, p. 144) by getting rid of Kantian scheme-content dualism that imposes them.
I do not intend to defend Kantian scheme-content dualism. It is only too clear today that a pre-schematic raw material, the given or sense data, is indeed objectionable. It is a well-worn topic argued repeatedly by many, including many conceptual relativists criticized by Davidson, such as Kuhn and Feyerabend. What Davidson said above is nothing new except that he unfairly attributes such a myth to conceptual relativism. I doubt that any contemporary conceptual relativists would continue to back up the scheme-El distinction. Kantian thing-in-itself W1 is not a viable option either. Let all this be granted, as it should be.
Of course, Davidson and Rorty have much bigger fish to fry than simply rejecting Kantian scheme-content dualism. They want to undermine conceptual relativism by removing its foundation, namely, any form of scheme-content dualism. To claim a victory, Davidson and Rorty have to show us that all attempts to disjoin scheme and reality are doomed to failure. In other words, they have to convince conceptual relativists that no scheme-content distinction of a certain sort can be drawn, which is innocent enough to be immune from the charge of the third dogma of empiricism, but solid enough to supply a foundation for conceptual relativism. Obviously, Davidson and Rorty have not done so. What they have done is to identify and criticize scheme-content dualism of a certain sort, i.e., Kantian scheme-content dualism, and jump to a wholesale dismissal of scheme-content dualism in general.
Following suggestions from J. McDowell (1994) and others (M. Baghramian 1998), I believe there is at least one kind of non-Kantian scheme-content dualism that can sustain conceptual relativism. The central question is: Are there other kinds of empirical contents, neither El nor Wl, available for scheme-content dualism? Truth is always nearby if you know where to look. How about our common-sense experience and the world as it is experienced by us?
Let us begin with experience. Sense data such as a patch of color, an indescribable sound, or a fleeting sensation are not our lived-experience like the perception of a yellow ball, loud music, or a feeling of love. Sense data is what W. James (1909) and C.I. Lewis (1929) called ‘thin experience of immediate sensation’, which should be distinguished from the so-called ‘thick-experience of everyday life’, such as our reflection of happiness and sadness, our perceptions of trees, rivers, and other people, as well as our experience of the world full of common objects, horses, buildings, etc. Davidson has never seriously entertained the possibility of thick-experience as the content of schemes. He always speaks of ‘experience’ in the sense of ‘thin experience’. Even when he does mention experience with plurality, which is thick-experience in nature—’events like losing a button or stubbing a toe’—he dismisses it right away by reducing those events into thin experience (Davidson, 1984, p. 92). However, Davidson does touch, I think, the notion of thick-experience under a different name, i.e., the world as it is experienced by us. Remember that Davidson’s mission is to ‘restore the unmediated touch’ with the world. ‘In giving up the dualism of scheme and world, we do not give up the world, but re-establish unmediated touch with the familiar objects whose antics make our sentences and opinions true or false’ (Davidson, 1984, p. 198). Clearly, the world that Davidson wants to keep is not the Kantian world-as-it-is, i.e., the world Rorty (1982) commends Davidson for bidding farewell. Instead, it is our common-sense phenomenal world with familiar objects like trees, tables, people, rocks, and ‘knives and forks, railroads and mountains, cabbages and kingdoms’; that is, the robust common-sense world defended by many common-sense realists. In this sense, Davidson’s notion of the world is much like the pragmatist notion of ‘funded experience’—’those beliefs which are not at the moment being challenged, because they present no problems and no one has bothered to think of alternatives to them.’ As Rorty notes, Davidson seems to perform the conjuring trick of substituting the notion of ‘the unquestioned vast majority of our beliefs’ for ‘the notion of the world’. From this view, since the vast majority of our common beliefs must be true, the vast majority of the objects that our common beliefs are about must also exist. It turns out that we can start with ‘the unquestioned vast majority of our beliefs’, which is just ‘our funded experience’ or ‘thick-experience’, but end with the world as it is experienced by us. ‘The thick-experience of everyday life’ or ‘the thick-experience of the world of things’ is in fact the world as it is experienced by us. For this reason, I will place both ‘thick-experience’ and ‘the world as it is experienced by us’ under one roof:
(E2/W2) either our thick-experience or our common-sense world, namely, the world which the unquestioned vast majority of our beliefs are thought to be about, or the world full of familiar objects, the stars, the trees, the grass, the animals, the rivers, the people, and so on.
Could E2/W2 be the proper content of a scheme-content dualism that can void the pernicious Kantian scheme-content dualism under attack? It is certainly innocent enough to rebut the charge of the third dogma of empiricism. Obviously, such a dualism does not prevent us from having direct contact with the world. The remaining question is whether E2/W2 is so innocent as to not be pure or solid enough to qualify as a proper empirical content of scheme-content dualism. Based on his reading of the upshot of Davidson’s arguments against the third dogma of empiricism, Rorty thinks so:
The notion of ‘the world’ as used in a phrase like ‘different conceptual schemes carve up the world differently’ must be the notion of something completely unspecified and unspecifiable—the thing-in-itself, in fact. As soon as we start thinking of ‘the world’ as atoms and the void, or sense data and awareness of them, or ‘stimuli’ of a certain sort brought to bear upon organs of a certain sort, we have changed the name of the game. For we are now well within some particular theory about how the world is. (1982, p. 14)
Davidson certainly agrees with Rorty on his dismissal of W2.
To understand fully Davidson’s and Rorty’s rejection of E2/W2 as a possible content of scheme-content dualism, we need to dig deeper into some basic assumptions which they associate with the dualism. Davidson believes that conceptual relativism apparently presupposes a commonality underlying alternative schemes: If there are alternative conceptual schemes, there must be one common element for them to conceptualize (Davidson, 2001a, p. 39; 1984, p. 195). I do not think this commonality requirement of conceptual relativism is objectionable as long as it is modest. What is controversial is how to explain it. Does ‘to be common’ mean ‘to be neutral to any schemes or be free of any interpretations or conceptualizations’? Davidson and Rorty think so. For Davidson, for the notion of alternative conceptual schemes to make sense at all, there has to be some common element shared by all possible conceptual schemes, and such a common element has to be empirically pure or theory/concept neutral in the sense that ‘there be something neutral and common that lies outside all schemes ... The neutral content waiting to be organized is supplied by nature ...’ either as El or as Wl (Davidson, 1984, pp. 190-91; my italics). Similarly, Rorty believes that conceptual relativism commits to the existence of some ‘neutral material’ shaped by our concepts. For Rorty, the dismissal of the notion of ‘neutral material’ marks the downfall of the notion of the conceptual. We thus have reached Davidson and Rorty’s first assumption of scheme-content dualism:
(D1) The neutrality of the content: Any empirical content of scheme-content dualism must be neutral to and beyond all schemes, theories, languages, and ideologies.
If the content as ‘the scheme-neutral input’ or ‘the theory-neutral reality’ is ‘untouched by conceptual interpretation’, then it cannot be contaminated by any concepts or interpretation. Then the scheme-content distinction has to be fixed and sharp; not any overlapping or intertwine between scheme and content is possible. This is Davidson’s and Rorty’s second assumption of scheme-content dualism:
(D2) The scheme-content distinction has to be rigid and sharp, without any overlapping, intertwining, or shifting between scheme and content.
Now we are able to understand fully why E2/W2, according to Davidson and Rorty, could not qualify as the empirical content of scheme-content dualism of the sort we are looking for. It has become common wisdom since Kant that there cannot be a purely unconceptualized content to our experience. Despite the decline of Kant’s transcendental philosophy, the Kantian doctrine of concept-ladenness of experience is still very much alive today. During the last century, we have witnessed various attacks on the contrast between the observed and the theoretical, which leads to the thesis of the theory-ladenness of observation, and the dismantlement of the very notion of the given (in e.g., Kuhn, Feyerabend, and Sellars). Of course, our thick-experience or the common-sense notion of the world as we experience it is not exempt from the invasion of concepts either. As McDowell argues, those experiences are already equipped with conceptual content. This is why Rorty insists that E2/W2 cannot be the content of scheme-content dualism since it is not neutral and universal to concepts and theories. When conceptual relativists introduce E2/W2 into the scene, the rules of ‘the game’ set up by D1 and D2 are violated.
Davidson and Rorty present a dilemma to conceptual relativism: To make sense of scheme-content dualism, we need to clarify its empirical content, which is either concept-neutral (as with Wl or El) or concept-laden (as with E2/W2). If the former, it has been proved to be a metaphysical myth—it simply does not exist—or a third dogma of empiricism; if the latter, it does not qualify as the content required by scheme-content dualism. Conceptual relativism’s response is simply this: Conceptual relativism faces such an inescapable dilemma only if we accept Dl and D2 as Davidson and Rorty try to seduce us to do. However, clearly conceptual relativists do not have to take this route, which would only lead to its own peril. Although Dl and D2 are indeed essential for Kantian scheme-content dualism, they are by no means essential for the conceptual relativist notion of a conceptual scheme.
In fact, many conceptual relativists reject the requirement of neutrality out-of-hand, or they even build their relativism upon such rejections. As Rorty has noticed, it is conceptual relativists, such as Quine, Kuhn, and Feyerabend, who take the lead in the battles against ‘givenness’ and ‘analyticity’. Rescher has put his finger right on the issue:
The idea of a pre-existing ‘thought-independent’ and scheme-invariant reality that is seen differently from different perceptual perspectives just is not a presupposition of the idea of different conceptual schemes. The Kantian model of a potentially differential schematic processing of uniform preschematic epistemic raw material is nowise essential to the idea of different conceptual schemes. (1980, p. 337)
One can abandon entirely the myth of neutral content without giving up the idea of alternative conceptual schemes. On the contrary, as I have argued earlier, it is exactly the denial of a fixed and absolute analytic-synthetic distinction and the abandonment of the given and a priori that turns Kantian conceptual absolutism upside down and thus makes conceptual relativism possible.
However, if there is no preschematic ‘given’ for all conceptual schemes to process, what is the empirical content of our schemes? My answer is: the thick-experience. The answer may face some quick rebuttals such as this: Since our experience is already scheme-laden, how can we separate a scheme from its content if we still want to make sense of the scheme-content distinction at all? Especially, how can two competing schemes share a common content if the content itself could be the very making of the schemes involved?
To address those questions, we can think of a conceptual scheme as a central point along a rope—or fishing net if you wish—with one side connected to our experience of the world and the other side to our beliefs about the world. ‘Our conceptual mechanisms evolve in a historical dialectic of feedback dialectic between cognitive projection on the one hand and experiential interaction with nature on the other’ (Rescher, 1980, p 340). On the one hand, with the abandonment of the sharp, rigid analytic-synthetic distinction, our concepts are no longer thought of as being a priori, but rather being a posteriori in the sense that meaning and concepts are both the bearers and the products of our factual beliefs. Concepts are not only the tools of inquiry but also its products. On the other hand, a similar dialectic interaction exists between scheme and experience. Our concepts are products of our past experience in the form of an unquestioned vast majority of our beliefs about nature. But our experience itself is richly endowed with conceptual inputs. We will never encounter situations with a conceptual tabula rasa. Clearly, there can be no rigid, sharp distinction between a scheme and its empirical contents.
However, it does not mean for one moment that no meaningful distinction between certain schemes and experience can be drawn at all. To make the point vivid, imagine that our conceptual framework consists of multiple layers of schemes. At the bottom is our most fundamental set of concepts, dispositions, prejudgments, or absolute presuppositions, which I will call basic experiential concepts. Like Kantian concepts, these concepts are highly general and pervasive, permeate every facet of our sensory experience, and are presupposed by our experience in general. In this sense, the basic experiential concepts can be plausibly said to ‘structure’ or ‘schematize’ our everyday experience. Unlike Kantian concepts, which are a priori, independent of any experience, the scheme of our basic experiential concepts is globally a posteriori as a product of past experiences. They evolve through human interaction with the natural and social environment during millions of years of human evolution. Unlike Kantian concepts, which are absolutely basic, i.e., having a fixed and invariant structure, our basic experiential concepts are not absolutely basic in a Kantian sense. Instead, they are hypothetically or historically basic in the sense that on the basis of our past evolutionary history and current structure of environment, those concepts are foundational—that is, universally presupposed by our experience—but we acknowledge that changes in the nature of those concepts could occur over time— for example, if our evolutionary path is altered in the future due to some unforeseeable dramatic environment change. Although we cannot give an exhaustive list of all basic experiential concepts here (I doubt we could ever do so), they are concepts mostly related to our sense perception, individuation, duration, and identity of objects within space and time. For example, to enjoy and experience the lovely, beautiful flower in front of me on the table, I need to have a concept of differentiation (such that I can distinguish the flower from its background), a concept of relative stability of the object (I know that it will not melt into thin air in the next moment), a concept of identity (I know it is the same flower sent to me by my lover yesterday), a concept of myself (I am the subject who is enjoying the flower), a concept of space and time, and so on. We can safely assume, based on Darwinian evolution theory, that there are some basic experiential concepts shared by human cultures and societies.11 In this sense, they are global or universal.
Accordingly, our thick-experience is the product of the dialectic interaction between our basic experiential concepts and experiential input from nature, whatever it may be, such that there is no way to separate which is form and which is content. Some may notice that I continue to use a suspicious notion, ‘experiential input from nature, whatever it may be’. But the notion used here is, borrowing Reseller’s comment on the notion of ‘scheme-independent reality’, ‘not constitutive, not as a substantial constituent of the world—in contrast with “mere appearance”—but a purely regulative idea whose function is to block the pretensions of any one single scheme to a monopoly on correctness or finality’ (1980, p. 337). In other words, I intend to use the notion to emphasize the empirical root of our basic experiential concepts. Our thick-experiences thus can be thought to be a common content shared by other higher-level schemes. Therefore, we can have commonality without neutrality.
Besides our basic experiential concepts as the foundation of our conceptual and experiential life, there are some more advanced sets of concepts or metaphysical presuppositions associated with cultures, intellectual traditions, or languages. Some of those conceptual schemes are radically distinct and schematize our rudimentary thick-experience in different ways to form different worldviews, cosmologies, or ways of life. The presence of different conceptual schemes manifests itself most dramatically when we come across some different ways of categorizing, a way of conceptualization, what seems to be the same experience. The best-known case would be classification of color in different cultures. As the literature on color amply demonstrates, apparently individually identifiable color samples, which two cultural groups presumably experience in the same way (assuming they all have normal vision), are often categorized into totally different color systems using different concepts of color. Similar examples are plentiful in anthropological and historical literature. If not only what one experiences determines what one believes, but also what one believes shapes what one experiences (the thesis of the theory-ladenness of observation), communities that adopt two radically distinct conceptual schemes, and thus different worldviews, would have different ‘more conceptually enriched experiences’, so to speak. This is the reason why Kuhn makes a seemingly absurd claim that such communities ‘live’ in two different ‘worlds’.
Despite Davidson’s very influential criticism of the very notion of conceptual schemes, the notion continues to enjoy its popularity and remains ubiquitous not – only in contemporary philosophy, but also in many other interpretative disciplines such as cultural studies, historical and classical studies, anthropology, comparative linguistics, cognitive science, and the history of science. Accordingly, conceptual relativism is still very much alive and remains in power. The sustained resistance to Davidson’s criticism is, I think, not simply due to its limited scope and some internal flaws of Davidson’s arguments, but mainly due to the explanatory power and intuitive appeal of the notion itself. For many philosophers and theorists in related fields, the notion serves as a conceptual foundation for their pet theories, such as social constructionism, multiculturalism, feminism, and postmodernism. For others, the notion plays a crucial role in many significant philosophical investigations, such as the issues of realism vs. anti-realism, incommensurability, inter-language translation, cross-language and cross-cultural understanding/ communication, etc. The wholesale dismissal of the notion would leave us with the problems of how to explain different ways in which the mind can mediate reality in general, and how to explain the variability of the ways the world is understood and conceptualized by different cultures, traditions, and languages in particular.
However, there is still another major reason responsible for Davidson’s failure that has not been widely recognized:12 Davidson’s attacks miss the very target that he wanted to hit. What Davidson attacks fiercely is not the very notion of conceptual schemes as it identifies in his ‘war declaration’, but a notion of conceptual schemes or, more precisely, the Quinean notion of conceptual schemes and its underlying Kantian scheme-content dualism. This is because, as we have argued above, Davidson’s two lines of arguments against conceptual relativism have a very limited scope: His arguments of inverifiability can only apply to the Quinean notion of conceptual schemes, which mistakenly construes conceptual schemes as sentential languages; his arguments against scheme-content dualism are targeted at Kantian scheme-content dualism only. However, both Quinean notion of conceptual schemes and Kantian scheme-content dualism are unsatisfactory for many conceptual relativists also, not just for the reason that they indeed face many conceptual difficulties as Davidson points out, but mainly because they simply cannot carry the weight of conceptual relativism. They cannot catch the essence of conceptual relativism. As such, Davidson is successful insofar as the Quinean notion of conceptual schemes and Kantian scheme-content dualism are concerned, but ‘the victory’ cannot be validly extended to a more robust notion of conceptual schemes. The three essential components of conceptual relativism are undamaged by Davidson’s attacks.
I have argued that conceptual schemes are not identical to sentential languages, and that scheme-content dualism neither entails nor presupposes Kantian scheme-content dualism—in giving up Kantian scheme-content dualism, we do not give up scheme-content dualism. Conceptual relativism can sustain a viable version of scheme-content dualism that is not subject to the charge of the third dogma. Conceptual relativism does not need to detach itself from the world as it is experienced by us through introducing any epistemological intermediary between the world and us. We stand together with Davidson and Rorty to applaud the loss of the Kantian world-as-it-is and the dissolution of the given, but we do not lose our world as it is experienced by us on the way. On the contrary, in terms of our conceptual schemes, we are connected to the world as close as it could be. And only through conceptual schemes can we be connected to the world. As a nondogmatic empiricism, conceptual relativism, as it is properly construed, acknowledges our common experiential root and celebrates our conceptual diversity at the same time.
This, I think, is a better reply to Davidson.
1 Davidson, 1984, pp. 184, 185f; 2001a, pp. 39-40.
2 It is not controversial as to whether translatability is a sufficient condition of languagehood.
3 For example, see Rescher, 1980, pp. 326-9.
4 For a few good comprehensive criticisms of Davidson’s arguments from inverifiability, please refer to N. Rescher, 1980, M. Forster, 1998, S. Hacker, 1996, and D. Henderson, 1994.
5 Please refer to the list of readings given in the previous note.
6 Davidson, 1984, p. 192; 2001a, p. 41; 2001 b, pp. 140-44.
7 In response to Davidson’s early criticism, Quine (1981) identifies two versions of empiricism, one as a theory of truth and the other as a theory of evidence. Based on her reading of Quine’s distinction, M. Baghramian (1998) explores the connection of these two versions of empiricism with the scheme-content in detail. Many of my comments on the topic draw inspiration from Baghramian (1998).
8 Davidson, 1984, p. 198; 2001a, p. 52; 2001b, pp.143-4.
9 Rorty, 1982, pp. 3-40; Davidson, 2001a, p. 52.
10 Davidson does not deny that senses do play a causal role in knowledge and acquisition of language. He admits that senses are crucial in the causal process that connects beliefs with the world (2001a, pp. 45-6).
11 Many wild thought-experiments about possible creatures endowed with very different thick-experience from humans’ (such as the visitors from Mars with different ranges of sensory perceptions from humans) assume a radically different environment from ours.
12 M. Lynch (1998) is a notable exception. I have drawn many inspirations from his essay.