We have found, from our cases studies, that the confrontation between two incommensurable theories is actually a conceptual confrontation between two scientific languages with different distributions of truth-value status over their sentences. The communication breakdown between them is often semantically correlated with the occurrence of a truth-value gap between the languages. Hence, what we should focus on, in the cases of incommensurability and conceptual schemes, is not truth or truth-functional meaning and translation, but rather truth-value status. In fact, I. Hacking put his finger on the problem about 20 years ago:
Many of the recent but already classical philosophical discussions of such topics as incommensurability, indeterminacy of translation, and conceptual schemes seem to me to discuss truth, where they ought to be considering truth-or-falsity. (1982, p. 49)
Hacking here clearly identifies the very root of both the received interpretation of incommensurability as untranslatability and the Quinean notion of conceptual schemes: neglect of truth-value status and denial of possible truth-value gaps between two incommensurable languages (or two alternative conceptual schemes). This is because both notions are rooted in bivalent semantics in which every declarative sentence is either true or false. As we have discussed in chapter 2, the translation used in the thesis of incommensurability as untranslatability is truth-preserving in nature, which requires the translation to preserve the truth-values of the sentences translated and to match the truth-values of the corresponding sentences between the target and the source languages. The alleged translation-failure between two incommensurable languages is believed to be caused by redistribution of truth-values between the sentences of the two languages due to meaning (sense and/or reference) variance of the terms involved. Therefore, the meaning relation becomes the determinate semantic relation between two incommensurable languages. The exact same line of reasoning is behind the Quinean notion of conceptual schemes and Davidson’s rejection of it. To see the link, one only needs to identify a conceptual scheme with a sentential language, construed as the totality of sentences held to be true by its speaker. Thus, redistribution of truth-values, and consequently the failure of intertranslation, would become the criterion of distinguishing two alternative conceptual schemes. If one thinks along this line of bivalent semantics, it would seem natural to deal with the notions of incommensurability and conceptual schemes in terms of truth, meaning, and translation in Tarski’s style, as Davidson does.
The furor over the translation-failure interpretation of incommensurability and conceptual schemes as well as the obsession with bivalent semantics have, I believe, obscured one of the most significant aspects of incommensurability: the obstruction of the truth-value functional relation (concerning the semantic relation between truth-value status, instead of truth-values, of the sentences between two competing languages) between the incommensurables. Too much attention paid to the meaning relation and lack of attention to the truth-value functional relation between two rival scientific languages is, to a large extent, responsible for the slow progress that has been made toward establishing the integrity and tenability of the notion of incommensurability. It is time to switch our attention from the meaning relation to the truth-value functional relation between rival scientific languages. As will become clear in the remaining chapters, it is the truth-value functional relation, instead of the meaning relation, that is the dominant semantic relation in the case of incommensurability.
Our new orientation not only emerges from the above cases studies, but also is inspired by I. Hacking’s (1982, 1983) and N. Rescher’s (1980) related works on the issue of incommensurability and the notion of conceptual schemes.
Hacking fundamentally opposes the received interpretation of incommensurability because ‘the idea of incommensurability has been so closely tied to translation rather than reasoning’ (Hacking, 1982, p. 60). He also rejects the Quinean notion of conceptual schemes for a similar reason: It exclusively focuses on truth and translation. For Hacking, even Davidson’s criticism of the Quinean notion does not fare well since ‘like Quine, he [referring to Davidson] assumes that a conceptual scheme is defined in terms of what counts as true, rather than what counts as true-or-false’ (Hacking, 1982, p. 62). Hacking believes that to think of conceptual schemes in terms of truth within the framework of bivalent semantics is misleading. ‘Bivalence is not the right concept for science’ (Hacking, 1982, p. 55). In fact, ‘once you focus on truth rather than truth-or-falsehood, you begin a chain of considerations that call in question the very idea of a conceptual scheme’ as well as the notion of incommensurability (Hacking, 1982, p. 59). Hacking thus urges us— in discussions of incommensurability, conceptual schemes, and other related issues—to switch our attention from truth and translation to truth-or-falsehood. Through the introduction of the notion of styles of scientific reasoning, Hacking starts to explore an alternative interpretation that focuses on truth-value status along the line of trivalent semantics.
Like Kuhn, Hacking (1983) found that interpreters reading an out-of-date text characteristically encounter passages that are not fully meaningful to their contemporaries. However, not all old texts cause the same level of difficulty in understanding.
An old theory may be forgotten, but still be intelligible to the modern reader who is willing to spend the time relearning it. On the other hand some theories indicate so radical a change that one requires something far harder than mere learning of a theory. (Hacking, 1982, p. 69)
When an old text involves an alien way of thinking and reasoning, even if it lies in the past of the interpreter’s own culture, it may sound so strange that it is often incomprehensible. Hacking uses the medical theory of Paracelsus, a well-known sixteenth-century medical chemist, as an illustration of such a case. Paracelsus’s works exemplify a Northern European Renaissance tradition of a bundle of Hermetic interests. He practiced many different disciplines, such as medicine, physiology, alchemy, herbals, astrology, and divination as a single art. For Paracelsus, the human body is a chemical system, similar in structure to the solar system. Each part of the human body corresponds to a celestial body: the heart to the sun, the brain to the moon, the liver to Jupiter, and so on. In this way, the human body mutually interacts with the celestial bodies and other chemical elements. This mutual interaction is exemplified by passages such as: ‘Nature works through other things, such as pictures, stones, herbs, words, or when she makes comets, similitudes, halos and other unnatural products of the heavens’. Specifically, Paracelsan medicine prescribes associations between the chemical element mercury, the planet Mercury, the marketplace, and syphilis. Syphilis was treated by a salve of mercury and by internal administration of the metal, because the metal mercury is the sign of the planet Mercury, and that in turn signs the marketplace, and syphilis is contracted in the marketplace.
Paracelsan medical theory simply appears bizarre to modern Westerners. We might conclude that Paracelsus is just a witch doctor. But Hacking argues that, as Kuhn’s experience with the reading of Aristotle’s physics, things are not that simple. Paracelsan theory was very influential in the sixteenth and seventeenth centuries and was accepted by both the populace and intellectuals. It was recorded that the students in Paris and Heidelberg protested against the proscription of Paracelsan theory at the end of the sixteenth century. It is obvious that Paracelsan theory was perfectly intelligible to his contemporaries. His theory was believed either to be true by his proponents or to be false by his opponents. Either way, for his contemporaries, Paracelsus’s assertions had definite truth-values. However, many Paracelsan sentences, such as (15), are hardly intelligible to us,
(15) Mercury salve might be good for syphilis because of associations among the metal mercury, Mercury, the marketplace, and syphilis.
Why can we not understand Paracelsan sentences? Hacking contends that, contrary to the received interpretation of incommensurability as untranslatability, it is not because we are unable to understand the words of his sentences, nor because we cannot translate his sentences into our modern language (say, English). Although his works were written in dog Latin and Proto-German, they now can be translated into modern German or English. In the Oxford English Dictionary, we can find a definition of the Renaissance word ‘Anatiferous’: producing ducks or geese, that is producing barnacles, formerly supposed to grow on trees, and, dropping off into the water below, to turn into tree-geese.1 Based on plain definitions like this, we can translate (liberal translation) Paracelsan sentences into English sentences. For instance, English sentence (15) cited above is a good translation of the original Paracelsan sentence, each word of which is plain enough for us.
The real trouble, Hacking finds, is that even when we are able to understand each word of Paracelsan sentences like (15), we are still left in a fog. What is the point of (15)? What does one argue or present by uttering (15)? Without grasping the point of (15), we cannot understand it effectively. In order to grasp the point of (15), we have to comprehend the forgotten mode of reasoning (the way of thinking and justification) that was central to Paracelsus’s thought. However, the Paracelsan mode of reasoning is alien to us. The Renaissance medical, alchemical, and astrological doctrines of resemblance and similitude, which had internalized into the Renaissance mode of reasoning, are well nigh incomprehensible to us. The Renaissance mode of reasoning is especially hard to grasp if we try to approach it from our familiar way of thinking and justification (which we have to since there is no other alternative). In this sense, ‘understanding is learning how to reason’ (Hacking, 1982, p. 60). Paracelsan sentences would not be fully intelligible to us until we have learned how to reason in his way.
More significantly, Hacking notices that there is a strong semantic correlate of our failure of understanding of Paracelsus:
The trouble is not that we think Paracelsus wrote falsely, but that we cannot attach truth or falsehood to a great many of his sentences. His style of reasoning is alien. (Hacking, 1983, p. 70; my italics)
Numerous Paracelsan sentences, when considered within the context of modern scientific theories, do not have any truth-values. We simply cannot assert or deny what was being said since there is no way to match what Paracelsus wanted to say against what we want to say. Let us put it into the terminology we have been using all along: There is a truth-value gap occurring between Paracelsus’s discourse and ours. The occurrence of a truth-value gap between the Paracelsan language and ours strongly indicates that we have experienced a communication breakdown between Paracelsus and ourselves. Such a communication breakdown could not be restored easily through normal language learning due to the involvement of an alien style of reasoning. In this sense, Paracelsus’s discourse is dissociated from or incommensurable with ours.
To explain the phenomenon of incommensurability as dissociation, Hacking introduces the notion of the styles of scientific reasoning as a substitute for the Quinean notion of conceptual schemes. Hacking finds, following the lead of A.C. Crombie, that ‘there have been different styles of scientific reasoning’ within the Western scientific tradition, such as the Euclidean style of thought in ancient Greece, and the Galilean style of reasoning in modern time, each of which has ‘specific beginnings and trajectories of development’ (Hacking, 1982, pp. 48-51). A style of reasoning is characterized by introducing novelties into scientific inquiry, including new types of objects (such as abstract mathematical objects in Platonism in mathematics), evidence, sentences, laws or at any rate modalities, possibilities, and new types of classification and explanations. Among all these novelties, the most notable feature that distinguishes Hacking’s styles of reasoning from the Quinean notion of conceptual schemes is that each new style brings with it new types of sentences, or a new way of being a candidate for truth-or-falsehood. Sentences that are meaningless and cannot be stated within one style of reasoning can be asserted to be either true or false within another style (Hacking, 1992, pp. 10-17). Thus, two scientific communities committed to different styles of scientific reasoning often find themselves experiencing a frustrated communication breakdown when one side tries an approach to the other—as with our experience with Paracelsus’s medical theory.
In contrast with the Quinean notion of conceptual schemes, Hacking’s styles of scientific reasoning have the following distinctive features.
(a) The Quinean notion of conceptual schemes is characterized by assignments of truth-values, that is, whether core sentences of a conceptual scheme are true. More precisely, ‘a conceptual scheme is a set of sentences held to be true’ and ‘two schemes differ when some substantial number of core sentences of one scheme are not held to be true in another scheme’ (Hacking, 1982, p. 58). A style of reasoning, in contrast, is concerned with truth-or-falsehood and is characterized by assignments of truth-value status. It is the styles of reasoning that ‘create the possibility for truth and falsehood’ and determine ‘what is taken to be a legitimate candidate for truth or falsity’ (Hacking, 1982, p. 57). ‘The very candidates (of sentences) for truth or falsehood have no existence independent of the styles of reasoning that settle what it is to be true or false in their domain’ (Hacking, 1982, p. 49).
(b) The styles of reasoning are not ‘sets of sentences held to be true’, but ‘would be sets of sentences that are candidates for truth or falsehood’ (Hacking, 1982, pp. 58, 64). Accordingly, the styles of reasoning are not sets of beliefs or propositions about the world, but rather the way beliefs or propositions are proposed and defended.
(c) The Quinean schemes are often characterized as a language confronting reality as in the fitting model R3. ‘A style is not a scheme that confronts reality’. Therefore, such a notion would not fall into the dogma of scheme and reality that Davidson resents (Hacking, 1982, p. 64).
(d) According to Davidson’s interpretation, cross-language understanding is a matter of designing a truth-conditional translation in Tarski’s style that preserves as much truths as possible as required by his charity principle. Since a style of reasoning is internal to any alien text, understanding is, for Hacking, not about truth-preserving translation, but ‘is learning how to reason’. ‘Understanding the sufficiently strange is a matter of recognizing new possibilities for truth-or-falsehood, and of learning how to conduct other styles of reasoning that bear on those new possibilities’ (Hacking, 1982, p. 60).
(e) Much like Kuhn’s paradigms, each style of reasoning is self-authenticating; there is no external justification as to which alternative styles of reasoning are better or worse (Hacking, 1982, p. 65; 1992, pp. 13-16).
If Hacking were forced to use the term ‘conceptual schemes’, he would define it as ‘a network of possibilities, whose linguistic formulation is a class of sentences up for grabs as true or false’ (1983, p. 71). Two schemes are distinct when the core sentences of one scheme are not held to be true or false in the other scheme.
Rescher, like Hacking, is equally unsatisfied with the Quinean notion of conceptual schemes and Davidson’s wholesale dismissal of the very notion of conceptual schemes and conceptual relativism within the framework of bivalent logic. Rescher therefore proposes a new version of conceptual schemes within ‘a three-valued framework of truth-values, one that adds the neutral truth-value (I) of interdeterminacy or indefmiteness to the classical values of truth and falsity (T and F)’ (Rescher, 1980, p. 332).
As the term suggests, a conceptual scheme is the mode of operation of concepts. ‘Concepts’ here for Rescher does not refer to mysterious Platonic entities, nor Kantian a priori mental schemes, but rather is a generic term used to denote all that is conceptual, such as ‘categorical framework (descriptive and explanatory mechanisms)’, ‘taxonomic and explanatory mechanism’, ‘fundamental concepts’, ‘modes of classification, description, explanation’. Rescher believes that, as a product of temporal evolution, our concepts of things are moving rather than fixed targets for analyses. As Quine has convinced us that there is no clear distinction between the analytical/conceptual and the synthetic/empirical, our concepts are always correlative with and embedded in a substantive view of how things work in the world. In other words, concepts are themselves loaded with substantial empirical/factual commitments. Accordingly, since all our concepts are factually committal, our conceptual schemes for operation in the factual domain (in natural sciences in particular) come to be correlative with a set of factual commitments, which constitute the essence of a conceptual scheme. In this sense, the factual commitments embedded within a scientific theory/language are its conceptual scheme. ‘Schemes differ in just this regard—in undertaking different sort of factual commitments’, or in a different way of conceptualizing the purported facts (Rescher, 1980, pp. 329-31).
The issue of scheme differentiation can also be approached from the angle of conceptual innovation. A new conceptual scheme brings with it not only new phenomena never before conceived and described, new modes of classification, description, and explanation, but also new ways of looking at old phenomena. ‘Such innovation makes it possible to say things that could not be said before—and so also to do new things’ (Rescher, 1980, p. 330). Why? Because ‘they lie beyond the reach of effective transportation exactly because they involve different factual commitments and presuppositions’ (Rescher, 1980, p. 331; my italics). Like Hacking’s styles of reasoning, factual commitments determine whether a sentence has a truth-value. In addition, only sentences that are either true or false can be used to describe presumptive facts, to make factual assertions or propositions, and are factually meaningful. Therefore,
Innovation—that availability of assertions in one scheme that is simply unavailable in the other—is one important key to their difference. One scheme will envisage assertions that have no even remote equivalents in the other framework. (Rescher, 1980, p. 331)
Consequently, Rescher reaches the same conclusion about the scheme differentiation as Hacking does: ‘The issue of scheme innovation at bottom turns out not on differences in determinate truth-values but on the having of no truth-value at all’. More precisely, ‘the key schematic changes are those from a definite (classical) truth-status to / (i.e., from T or F to I) or those in the reverse direction (i.e., from I to Tor F) ‘ (Rescher, 1980, p. 332).
Now we can see, according to Rescher, why the Quinean model of conceptual schemes cannot reveal the genuine conceptual innovation.
If the conceptual scheme C is to be thought of as an alternative to C along the lines we have in view, then one cannot think of C as involving a different assignment of truth-values to the (key) propositions of C. One must avoid any temptation to view different conceptual schemes as distributing truth-values differently across the same propositions. The fact-ladenness of our concepts precludes this and prevents us from taking the difference of schemes to lie in a disagreement as to the truth-falsity classification of one selfsame body of theses or doctrines. (Rescher, 1980, p. 331)
For this reason, the key contrast between competing conceptual schemes is not between affirmation (to be true) and counter-affirmation (to be false), but rather between saying something (to be either true or false) and saying nothing (to be neither true nor false). As such, two schemes do not dispute over the same things or facts; rather, they are about different things and facts. If so, Davidson’s charge of the third dogma is off target since there is no common content for two conceptual schemes to process.
Hacking and Rescher have proposed a very promising version of conceptual schemes different from the Quinean model, and, accordingly, a new interpretation of incommensurability within trivalent semantics. But they have offered only scattered insights here and there. Many details need to be worked out in order to make it a full case. The major issue remaining is this: If the primary function of conceptual schemes—Hacking’s styles of reasoning or Rescher’s factual commitments—is to determine the truth-value status of the sentences involved, then we need a semantic mechanism to explain why this is so. In other words, we need a workable semantic theory of truth-value conditions (not truth conditions). However, Hacking and Rescher either fail to do so or are unable to give us a satisfactory one. This is part of the reason why their insights have not gained their deserved attention up to now.
Hacking claims that truth-value status is style-of-reasoning relative, but he fails to tell us how. For example, how can the introduction of new types of objects, evidence, classifications, and explanations change the way sentences stand as candidates for truth or falsehood? Hacking does not explain it except to claim that the new style of reasoning and the new type of sentences—or a new way of being a candidate for truth or falsehood—occur together. Worse still, Hacking specifies the introduction of a new type of sentences, which have no truth-values in some earlier language but are true or false with the new style in place, as a necessary condition of a new style of reasoning. Apparently, such a clarification of the notion of the styles of reasoning suffers from circularity. Which one is logically more primitive: the occurrence of the new type of sentences or the new style of reasoning? It cannot be the new type of sentences; otherwise, we need another more primitive notion to explain their occurrence. It has to be the new style of reasoning. If so, then we need a criterion independent of the new type of sentences to define it.
Fortunately, there is one semantic theory available to explain the occurrence of truth-valuelessness, and hence truth-value status without circularity. I mean R. G. Collingwood’s and P. Strawson’s theory of semantic presupposition. To be fair to Rescher, he indeed uses the word ‘presupposition’ once in conjunction with ‘factual commitments’ as we have quoted earlier (Rescher, 1980, p.331). Nevertheless, Rescher never explores such a significant insight further except to imply that ‘factual commitments and presuppositions’ determine the truth-value status of the sentences involved (Rescher, 1980, p. 331; my italics). However, it will become clear that Rescher’s factual commitments do function as metaphysical presuppositions underlying scientific theories.
As Collingwood observes:
Whenever anybody states a thought in words, there are a great many more thoughts in his mind than are expressed in his statement. Among these there are some which stand in a peculiar relation to the thought he has stated: they are not merely its context, they are its presuppositions. (1940, p. 21)
In fact, every statement is made potentially in answer to a question; and every question involves a presupposition, such as the notorious one, ‘Do you still beat your wife?’ Thus, every statement involves a presupposition. For example, the sentence,
(16) The present king of France is bald,
could be used to answer a potential question:
(16q) Does the present King of France have hair?
The question in turn presupposes this sentence:
(16a) The present king of France exists.
Of course, a statement may have many different presuppositions. For example, sentence (16) has at least four different presuppositions, i.e., (16a), (16b), (16c), and(16d).
(16b) There is a country named France.
(16c) A person can have hair.
(16d) A subject (such as ‘a person’) can possess a property (such as ‘of having hair’).
Among those presuppositions, only one is the immediate presupposition, namely, the one from which the question immediately arises, such as (16a). The rest would be mediate presuppositions, which are indirectly presupposed by the original question.
More significantly for our later discussion is the distinction between relative and absolute presuppositions made by Collingwood (1940, p. 4). A relative presupposition is a proposition (a proposition could be true or false) that could be questioned or verified within a certain domain of inquiry, such as (16a), (16b), and (16c) for our common-sense non-metaphysical way of thinking. But some fundamental presuppositions are unquestionable and cannot be verified within a certain domain of inquiry: These are absolute presuppositions. For instance, (16d) is an absolute presupposition to be taken for granted in all of our non-metaphysical inquiry. Newtonian absolute space and time are presupposed absolutely within the Newtonian paradigm. In this sense, those absolute presuppositions are not propositions that could be false within a certain domain. Absolute presuppositions are not asserted (what is asserted could be either confirmed or negated), but made or presupposed (what is presupposed in a certain inquiry could not be denied). Any question challenging the legitimacy of an absolute presupposition, such as the question, ‘Is it true?’ or ‘What evidence is there for it?’ is a nonsense question to its believers. This is why when absolute presuppositions are challenged, their believers ‘are apt to be ticklish’.
Strawson’s notion of semantic presuppositions will be formally presented, clarified, and defended in chapter 8. Here I only want to mention Strawson’s notion of basic concepts, which function like Collingwood’s absolute presuppositions. Strawson approaches ‘the absolute components’ of our thoughts from the perspective of basic conceptual structures. For Strawson, our conceptual structure is a set of interlocking systems of concepts such that each concept, no matter which is simple or complex, could be properly understood only by grasping its connections with others, its place in the system. However, within a conceptual system, although no concepts are absolutely simple or fundamental, some concepts could be conceptually a priori to other concepts in the sense that ‘the ability to operate with one set of concepts may presuppose the ability to operate with another set, and not vice versa’. Strawson calls those presupposed concepts, such as the concepts of body, time, change, truth, identity, knowledge, etc., philosophically basic concepts.
A concept or concept-type is basic in the relevant sense if it is one of a set of general, pervasive, and ultimately irreducible concepts or concept-types which together form a structure—a structure which constitutes the framework of our ordinary thought and talk and which is presupposed by the various specialist or advanced disciplines that contribute, in their diverse ways, to our total picture of the world. (Strawson, 1992, p.24)
Notice that for both Collingwood and Strawson, absolute presuppositions or basic concepts are not absolute or basic in the Kantian sense, i.e., to be a priori, ahistorical, and universal for all conscious beings. Rather, they are absolute or basic within certain contexts. For Collingwood,
metaphysics is the attempt to find out what absolute presuppositions have been made by this or that person or group of persons, on this or that occasion or group of occasions, in the course of this or that piece of thinking. Arising out of this, it will consider ... whether different absolute presuppositions are made by different individuals or races or nations or classes. (1940, p. 47)
In this sense, ‘all metaphysical propositions are historical propositions’. Strawson’s basic concepts are basic sets of concepts that are pervasive within a particular historical, cultural, or linguistic context. What are presupposed (basic concepts) could change with what are presupposing (normal concepts and factual statements). Thus, alternative sets of absolute presuppositions or basic concepts are not just possible and desirable, but also are what really happened in the development and transformation of human intellectual history.
What happens if a presupposition of a statement does not hold? For Collingwood, if a presupposition (such as (16a)) of a question (such as (16q)) is not made, then the question simply does not arise. That implies that we cannot judge the truth-value of the corresponding initial statement that the question addresses (such as (16)) since it is not even a proposition. Strawson makes this conclusion bluntly clear. Based on his trivalent semantics (1950), a semantic presupposition of a sentence has to be held true in order for the sentence to be true or false. In other words, the truth of a presupposition of a sentence is necessary for the truth or falsity of the sentence. For example, (16) is true or false only when (16a) is true; otherwise, (16) is neither true nor false (has no classical truth-value).
In fact, a comprehensive scientific language is fully loaded with a set of absolute presuppositions (some of them are basic concepts). Its core sentences share one or more absolute presuppositions. For example, the existence of the yin and the yang as well as the five elements, as the absolute presuppositions, underlies core sentences of the language of traditional Chinese medical theory. Similarly, the existence of phlogiston is embedded within the very conceptual set-up of phlogiston theory and is presupposed absolutely by numerous core sentences of the language of phlogiston theory (say, ‘Object a is richer in phlogiston than object b’). Likewise, the assumption that there exists absolute space and time underlies the core sentences of the Newtonian language of space and time, the denial of which is unimaginable within the conceptual framework of the Newtonian language.
For this reason, I call a scientific language a presuppositional language (‘P-language’ in brief hereafter). By a P-language I mean an interpreted language whose core sentences share one or more absolute presuppositions. Even languages we use in everyday discourse (not a natural language such as English per se) are P-languages to some extent. The fact that the sun exists, rises, and sets periodically may in many everyday discourses count as an inevitable presupposition. Denial of it would play chaos with everyday communicative activity. For instance, if I promise you that I will pay back your money tomorrow morning and you understand what I mean, then both of us take it for granted that the sun exists, rises and sets periodically.
The conceptual core of a P-language consists of a set of absolute presuppositions underlying the core sentences of the language, which I call metaphysical presuppositions (‘M-presuppositions’ hereafter) of the language. M-presuppositions of a P-language are contingent factual presumptions about the world perceived by the language community. As will become clear later, Hacking’s styles of reasoning and Rescher’s factual commitments are actually different kinds of M-presuppositions of scientific languages.
The factual presumptions of a P-language could manifest themselves in different ways. First, they could be basic existential presumptions about the entities existing in the world around a language community, such as phlogiston in the language of phlogiston theory. Second, they could be basic universal principles about the existential state of the world around a language community, such as ‘Fermat’s conjecture’ in the language of classical arithmetic; the second law of motion in the Newtonian language of mechanics, etc. Third, they could function as basic categorical frameworks about the structure of the world perceived by a language community, such as the taxonomy of Copernican astronomy, or the taxonomy of the Aristotelian language of mechanics.
The traditional Chinese medical theory (CMT) is a typical P-language. All three types of M-presuppositions are identifiable within it. First, the existences of the yin and the yang as well as the five elements are existential presumptions of CMT. Both underlie numerous core sentences of the language of CMT. Second, CMT has its own unique medical category system. For example, all symptoms related to diseases are categorized as eight principal syndromes, which can be grouped further into four matched pairs: the yin versus the yang syndrome; the superficial versus the interior syndrome; the cold versus the heat syndrome; and the asthenia versus the sthenia syndrome. Third, CMT is richly embedded with a unique universal principle, the pre-established harmony and mutual influence between the human body and Heaven (as Nature or the Universe). I will analyze those and other aspects of CMT in detail in chapter 10.
As absolute presuppositions, M-presuppositions of a P-language are analytically true in the language. Denials of them signify a complete breakdown of the informative use of the language and a complete rejection of it. For example, the M-presuppositions that the sun exists and that it rises and sets periodically are presupposed in the very linguistic set-up of our natural language. Denial of them will play chaos in everyday conversation. The existence of phlogiston is analytically true for any conceptually possible interpretations of the language of phlogiston theory. Rejection of phlogiston means rejection of phlogiston theory.
It should become clear that the proper terms of incommensurability relationship are two distinct P-languages. To say that two theories, systems, or languages are incommensurable is to say that the two associated P-languages are incommensurable, or that the communication between the two language communities breaks down.
A set of M-presuppositions of a P-language is what we normally call the conceptual scheme of the language. To construe a conceptual scheme as a set of M-presuppositions of a P-language is advantageous for several reasons. First, it can catch the essence of Kant’s scheme-content dualism; that is, conceptual schemes are ‘necessary for the constitution of experience, not just necessary to control and predict experience’ (Rorty, 1982, p. 5). A conceptual scheme is not what we experience, what we believe consciously, but what makes our experience and beliefs possible. In other words, some conceptual structures are logically presupposed by all experiences and beliefs of a language community. Of course, what Kant tries to reconstruct are certain minimum conceptual structures that are essential to or universally presupposed by any conception of experience of all self-conscious beings. However, the possible existence of such a minimum limit of conceptual structure for all conscious beings does not exclude the possibility of a more localized conceptual structure embedded within a language, a tradition, or a culture that is presupposed by any specific experience and beliefs of its participants. Although those localized conceptual structures themselves, unlike Kant’s a priori categorical concepts, are factually committal (using Rescher’s terms) and could change with contexts (languages, traditions, or cultures), a certain conceptual structure could still be fundamental within a particular context in the sense that it is presupposed by other concepts, experience, and beliefs, such as Strawson’s basic concepts.
Second, as we have discussed earlier, a conceptual scheme is closely associated with a language, but is not identical to a sentential language. However, Hacking still somehow treats a conceptual scheme as a sentential language. As Hacking notes, ‘Quine’s conceptual schemes are sets of sentences held to be true. Mine would be sets of sentences that are candidates for truth or falsity’ (1982, p. 64; my italics). However, as we have argued earlier, the style of reasoning itself cannot be true or false; only the language embedded within a style of reasoning can be true or false. In fact, according to Hacking, a style of reasoning is self-authenticated and cannot be false from the perspective of its speakers. Hacking apparently confuses a style of reasoning with the language embodying the style. We need to separate a conceptual scheme (in our case, a set of M-presuppositions) from the language in which it is embedded (in our case, a P-language). As Strawson argues, a presupposition of a sentence is not a part of the sentence, and is not even logically entailed by the sentence. By the same token, an M-presupposition (such as, ‘phlogiston exists’) of a sentence (such as, ‘The element a is not richer in phlogiston than the element b) is not a part of the sentence. Thus, the M-presuppositions of a scientific language, which might be important components of the corresponding scientific theory, are not parts of the linguistic set-up of the corresponding P-language.
Third, ‘a conceptual scheme’ is a broader notion than ‘a set of concepts’ if ‘concept’ is construed either in a normal sense, namely, an abstract entity, the meaning, the interpretation, or the disposition associated with a term (‘causation’, ‘a tree’, ‘a game’), or in a functional sense, namely, whatever composes the propositional content of our assertions and beliefs. To treat conceptual schemes as a set of concepts is the approach taken by the Kantian model of conceptual schemes. It is too narrow. Instead, ‘a conceptual scheme’ should mean ‘a scheme of what is conceptual’, which includes basic concepts, categorical frameworks (lexical structures or taxonomies), modes of reasoning and justification, ways of thinking, descriptions, and explanations, and some fundamental factual commitments (universal principles, existential presumptions). To me, those are M-presuppositions of a P-language. In this sense, the presuppositional model is more comprehensive than the Kantian model.
Last but most significantly, recall the powerful insight shared by both Hacking and Rescher on scheme-transition: Scheme change and differentiation do not consist in redistribution of truth-values as the Quinean model tells us, but are semantically correlated with the redistribution of truth-value status. However, Hacking and Rescher fail to specify a badly needed truth-value condition to explain sufficiently the occurrence of truth-valuelessness and change of the truth-value status. To construe a conceptual scheme as a set of M-presuppositions promises a basic semantic framework to work out such a truth-value condition. If a conceptual scheme is a set of M-presuppositions, it is not hard to understand why the core sentences of a P-language PL1 could be truth-valueless when viewed from the perspective of another P-language PL2 that suspends the M-presuppositions of PL1 As will be presented in chapter 9, a single M-presupposition of the core sentences of a P-language is necessary for the truth or falsity of its sentences; and the conjunction of all the M-presuppositions is sufficient for the truth or falsity of its sentences. This establishes that the M-presuppositions of a language constitute the truth-value conditions of its core sentences. The truth-value status of the core sentences of a P-language is determined by its M-presuppositions, which, as absolute presuppositions, are self-evident and unquestionable for its practitioners. If the speakers of PL1 are unable to recognize and comprehend the Mpresuppositions of an alien PL2, then the core sentences of PL2, when considered within the context of PL1 will lack truth-values. Thus a truth-value gap occurs between PL1 and PL2. This is why the occurrence of a truth-value gap between two P-languages is a strong semantic indicator that the two languages embody two competing conceptual schemes.
One might wonder how our presuppositional model of conceptual schemes fares against Davidson’s criticism of the very idea of a conceptual scheme. Clearly, the model effectively sidetracks Davidson’s verificationist arguments by removing its two basic assumptions (TV) and (TF). According to the presuppositional model, a conceptual scheme is not identical with a sentential language, but is a set of M-presuppositions of a P-language. Thus, sentential-language translatability can no longer be used as a criterion of the identity of conceptual schemes. In addition, an alien conceptual scheme is not a set of sentences taken to be ‘largely true’ to ‘fit’ reality. Consequently, content-scheme dualism could not be boiled down to the claim that a conceptual scheme different from the interpreter’s is largely true as Davidson construed. In fact, the core sentences of an alien P-language have no truth-values—not to mention that they are by no means ‘largely true’—when they are considered from the viewpoint of a competing P-language. Thus, Davidson cannot derive translatability from truth (even if his truth-conditional theory of translation can still hold up) since there are no shared truths between P-languages to begin with. Therefore, Davidson’s argument from translatability against radical conceptual relativism becomes powerless in front of the presuppositional model.
As to Davidson’s argument from interpretability against modest conceptual relativism, it cannot be sustained after we remove its theoretical foundation, i.e., the truth-conditional interpretation of the charity principle, which again depends upon the shared holding-true between two conceptual schemes.
‘Wait a minute’, a baffled critic might argue, ‘if your interpretation of scheme differentiation makes sense, i.e., a truth-value gap occurs between two competing P-languages, then the result will be the same as the conclusion derived from the Quinean model: Truth-preserving translation between the two languages is impossible. If so, Davidson’s argument against untranslatability still has the teeth to tear off your very idea of conceptual schemes, right?’ Our hypothetical critic is right that the new criterion of scheme differentiation, i.e., occurrence of a truth-value gap between two P-languages, does logically lead to the failure of truth-preserving intertranslatability between them. But remember that the Quinean thesis of untranslatability between alternative conceptual schemes itself is not incoherent (although it is confusing and unproductive). It becomes incoherent only in conjunction with Davidson’s interpretation of Quinean scheme-content dualism, according to which a conceptual scheme different from the interpreter’s is largely true. But Davidson’s truth-conditional theory of translation does not allow a divorce between truth and translation. After we break the conjunction by removing the last conjunct (Davidson’s interpretation of Quinean scheme-content dualism), the thesis of untranslatability is actually harmless (and also useless). Besides, our presuppositional model does imply that two conceptual schemes are untranslatable; but it does not follow that the untranslatables are distinct conceptual schemes. It is not controversial to claim that intertranslatability is a sufficient condition for the identity of two conceptual schemes (if two languages are intertranslatable, then they embody the same conceptual scheme); so is its logical contrapositive: Untranslatability is a necessary condition for alternative conceptual schemes (two languages embedded with two different conceptual schemes are untranslatable). What is at stake in the Quine-Davidson debate is whether untranslatability is sufficient for scheme difference; or whether translatability is necessary for languagehood. The presuppositional model implies neither. This is because the failure of mutual translation between two languages does not logically lead to the occurrence of a truth-value gap between them.
Of course, our presuppositional model still faces the challenge from Davidson’s criticism of Kantian scheme-content dualism. Unfortunately, our model is not able to avoid the attack by maneuvering into a different kind of conceptual schemes as we have managed to deal with Davidson’s attack based on verifiability. We have to face Davidson’s attack head-on, as we have done in chapter 4.
1 Many writers in the Renaissance thought that geese were generated from rotting logs in the Bay of Naples and that ducks were generated from barnacles.