Arthur Sullivan

W.V. Quine on the Analytic-Synthetic Distinction

W.V. Quine is arguably the most accomplished metaphysician still living today, and the most innovative and important empiricist of the twentieth century. Quine has, in effect, re-established the boundaries of metaphysics with his unwavering criterion of unambiguous ontological commitment, and his insistence that philosophy is continuous with science, but with a broader scope. In addition, Quine has brought empiricism to an unprecedented peak by strictly adhering to Ockham's razor, the principle of parsimony, and a rigorous self-imposed limit to extensional entities.

Much of the progress and insight with which Quine is accredited rests on his collapse of the distinction between analytic and synthetic truths in the essay "Two Dogmas of Empiricism." Quine held that the distinction, that had for centuries been considered both logically and epistemologically necessary, had never been very well made or conclu- sively defined, and what is more, that there was no good reason to make it. In the opening paragraph of "Two Dogmas," Quine asserts that a collapse of the elusive and problematic distinction will blur the boundary between philosophy and natural science, as well as shift the focus of the empiricist movement toward pragmatism.

In order to understand why Quine holds that the analytic-synthetic distinction was never conclusively drawn, let us trace its development through the empiricist tradition. The philosophy of David Hume is an appropriate starting point, for, besides being the apex of classical empiricism, it is where the behaviourist epistemological doctrine of pragmatism finds its origin. Pragmatism, the shift of emphasis instigated by Hume, is the insistence that, in terms of coming to understand human experience, the observable consequences of that which we claim to know are more important and more illuminating than speculation over the origin and essence of our ideas.1

A fundamental empiricist premise is that ideas, and subsequently knowledge, come from experience. Thus, truths such as "all bodies are heavy," long considered by most philosophers to be analytic a priori, meaning that the subject necessarily contains the predicate, were for Hume synthetic a posteriori. What this means is that for Hume, the observing a constant conjunction of two concepts the predicate of heaviness with every body encountered the mind synthesizes them, customarily connecting them by a psychological habit. In contrast to these matter of fact propositions, Hume does allow that some truths are analytic a priori, such as "2+2=4," but that propositions of this status are purely relations of ideas. They are never to be met with in the world, and have no observable consequences.2

The proposition "all bodies are heavy" may be used to illustrate just how fuzzy and inconclusive the analytic-synthetic distinction really is. Let us take the thought of Gottfried Leibniz, for example. In his view, all that is must accord with a pre-established harmony, and therefore every subject necessarily contains all of its predicates. All truths are thus analytic.3 Hume chooses a different starting point and reaches a conclusion that is virtually opposite. It seems Quine is most justified in asserting that the distinction has never been well made; each philosopher's view on the subject simply reflects his or her metaphysical standpoint. This standpoint tells us something about their particular predispositions and approaches to logic, but nothing about the world.

Upon fully unpacking the repercussions of his rigid distinction, Hume finds himself in a somewhat nihilistic metaphysical position. Essentially, all that we claim to know or hold as necessary is explained in terms of the structure of our minds - a necessary connection is fallaciously attributed wherever a constant conjunction is perceived, by a psychological habit. John Stuart Mill, the last of the classical empiricists, went one step beyond Hume, and asserted that all knowledge, even the supposed eternal truths of mathematics, is synthetic a posteriori, simply generalization from experience.4

The idealist movement which dominated the nineteenth century, though, zealously pointed out that strict adherence to the tenets of classical empiricism not only leads one to conclude that metaphysics is, in a sense, illusory. That position is also painfully unable to account for the many complex relations which underlie and compose reality. For example, exclusive existential commitment to particulars led the classical empiricists to a resemblance theory of universals. This theory is criticized for, among other reasons, turning to an infinite regress - we call an object "chair" because it resembles, or shares a significant number of attributes with, another object which we call by the same name; the reason we call the second object chair is because it resembles another similar object to which we refer with the same name, and so on ad infinitum. Hence idealists argued that empiricism is an untenable position.

A related idealist criticism is very useful in illustrating the movement from classical empiricism to the analytical school of the twentieth century. Locke, Berkeley, and Hume were confined to an Adamic, building-block view of language, wherein names refer directly to objects, and thereby commit existentially to that which is named. Jeremy Bentham was the first empiricist to point out flaws in this position - for example, the word "two" would commit one to abstract entities, to numbers out there in the world, which is clearly anti- empiricist. The unavoidable ambiguity inherent in this view of language was made clear by Frege. More than one name frequently refers to the same object, or one name to several objects.5

Enter Bertrand Russell, who, in effect, ushered out classical empiricism and set the stage for its more sophisticated modern descendant. Russell recognized the need for a non-empirical mode of cognition, for at least some ideas to come from other than experience, in order to vault by the impasse where Hume's corpse still lay decomposing. Russell did accomplish a startling amount in his lifetime, but he was forever inconsistent and often confusedly undecided. His insistence on abstract entities, such as universals, runs contrary to Ockham's razor, which was to become the fundamental epistemological principle of Quine's analytical philosophy.

Quine picked up on Russell's efforts. He, too, was determined to benefit from idealist criticisms of empiricism and reshape that school into a most tenable position. Quine wished that his account be more consistent than Russell's, though, and that it be established with a staunchly minimalist vocabulary. Russell grasped the full import of Frege's distinction between sine and bedeutung, between the sense of a word and the object$gt;s< to which it is employed to refer. He instigated a shift of emphasis which Quine drew even further. This shift of emphasis was not only from mental entities to verbal, but from words to propositions. Russell considered the sentence, not the name, to be the unit of meaning. Quine, however, realized that there were two basic reasons why the primacy of sentences does not satisfactorily solve the problem of names. First, sentences such as "Pegasus is a winged horse" still commit one to abstract or fictional entities. Second, if names are absolutely not referential, then sentences are meaningless because they refer to, denote, and commit one to nothing.6

"Predication over designation" is Quine's famous formula for dealing with the problem of names. He held that one does unambiguously commit oneself to an object by predicating it in a quantified sentence. The problem of abstract or fictitious entities does not arise because all statements are phenomenologically translated. "This is blue" becomes {Ex}{x is blue sense data}. Although names are not referential, the quantified sentence is still meaningful because to be is defined by Quine as to be the value of a variable.7 The existential reference, as well as the requisite ontological commitment, are accomplished by predication. The long-supposed necessary bond between name and object is relinquished to an outdated and unworkable theory. Quine's thought here resembles Russell's theory of indefinite descriptions. This again illustrates how he picked up and fine-tuned many notions which Russell left incomplete. Russell just quantified things known only by description - such as "x is the author of Waverly, or "x is the first line of Gray's elegy" - whereas Quine saw the need to quantify all that we can know.

In light of his unwavering ontological commitment, Quine was redefining not simply empiricism, but reality itself. It is in this context, and toward this metaphysical end, that Quine sought to collapse the time-honoured distinction between analytic and synthetic propositions. This task is most forcibly demonstrated in his critique of Kant's views on the subject.

In terms of the analytic-synthetic distinction, Immanuel Kant was the most influential idealist philosopher. Kant admitted to being woken from his dogmatic slumber, in which he was under the influence of Wolff, a Leiboizian rationalist, by Hume and the problem of induction. However, he held that the issue was effectively addressed by recognizing the validity of the synthetic a priori proposition. A synthetic truth is defined by Kant as one whose contradiction is logically possible, while analytic truths are deemed true in virtue of their meaning, independently of facts.

. . . either the predicate B belongs to the subject A, as something which is contained in this concept A; or B lies outside the con cept A, although it does indeed stand in connection with it.8

The reason which was glossed upon earlier why the distinction has never been conclusively drawn - that it is somewhat subjective and arbitrary: whether one holds that "all bodies are heavy" is analytic or synthetic merely reflects personal predisposition and metaphysical standpoint - is taken a step further here by Quine. When he asserts that Kant's distinction is limited to the subject-predicate form, he is ontologically rephrasing this objection, and in so doing, he reaches to the very heart of the issue. What this means is that the way in which one categorizes propositions affects only linguistic usage, and not anything in the world, and thus there is no good reason for making the distinction.

There do not exist two distinct types of reality in the world which require two distinct modes of expression. This leads Quine to conclude that the analytic-synthetic distinction is a purely logical convention that is ontologically unnecessary and empirically superfluous. In this respect, Quine agrees with the radical empiricism of Mill, with its claim that there is no a priori knowledge. The fact that something is the case, or even the fact that something seems to be necessarily the case, does not imply the reality of a priori truths. Quine goes so far a to refer to the notion of a priori knowledge as a "metaphysical article of faith."10

Quine's assertion that the distinction rests precariously on an ambiguous notion of meaning serves a likewise function. On the surface, it is a reason why the distinction has never been well made, but in unpacking this assertion Quine reaches to the metaphysical core of the issue, and it turns upon the reason why there is no essential need to draw the distinction. One can recognize something as true, according to Kant, independently of experience, in virtue of its meaning, if the subject contains its predicate. Quine puts this notion of analycity itself under his microscope. Much like he exposed the flaws inherent in the bond between name and existence, he investigates and shatters the privileged link supposed to exist between certain privileged subjects and their predicates.

When Quine points out that Kant leaves the concept of "contain" at a metaphorical level, he has found the fulcrum on which the entire distinction rests. Kant never explicated exactly what it means for one word or concept to contain another. Quine reasons that synonymy of some sort is required, that the specific subject-concept must at some point overlap the predicate-concept. The first prospective answer to the problem of synonymy is meaning.

Quine holds that there are only three plausible theories of meaning - reference, mentalism, and intentional objects.11 The reference theory is the view that the meaning of the word refers to that for which the word stands. In addition to Quine's demonstration of the problem of words, which renders untenable the reference theory, Frege's point about the essential need to distinguish between meaning and reference effectively demonstrates that synonymy cannot be based upon this reference theory. The mentalist theory holds that meaning is a mental entity, an abstract thought chord that is somehow struck by the word. Quine points out that if this were the case, communication would be impossible, because meaning would be subjective, private, and arbitrary. Meaning as an intentional entity, as an essential part of an idea or object, is the most difficult to define, to locate, or to know. It is therefore of little worth to a minimalist empirical approach; Quine scalpels this last notion of meaning away with Ockham's razor.

Hence, Quine's thorough analysis of the notion of meaning has yielded nothing on which to base synonymy. No privileged link exists to allow the assertion that any subject contains any predicate, and therefore there is no way that any proposition may be deemed true in virtue of its meaning. In "Two Dogmas," Quine then turns to definition, interchangeability, and verification to see if either notion contains a ground for the synonymy that analycity seems to presuppose.12 Definition does not clarify synonymy - if "bodies" are defined as "that which is heavy," that is a mere convention; "heavy" is either presupposed or created to suit an end, neither of which are grounds for a necessary connection.

Interchangeability is simply too broad to demonstrate synonymy. The fact that two terms have the same truth value and may, in some situations, be substituted for one another is insufficient, because the two terms have different significance in terms of Quine's holistic account of experience, different relevance to his web of belief, or interrelated network of revisable propositions which form the basis for all intellectual activity.

Finally, synonymy cannot be deduced from verifiability, because Quine reinterprets the traditional empiricist appeal to experience holistically, to apply to his web of belief. The entire system as one body must have significance. Thus, isolated sentences are not affirmed or denied but ascertained pragmatically. Their always revisable significance and status vary according to their relation to the entire web.

Truth is neither a matter of degree, as it is from the idealist's holistic viewpoint, nor of correspondence, a position to which classical empiricists were led. It is rather judged in terms of the ability of the statement to resolve and generate questions. Truth becomes a relational matter, concerned with how well a statement fits into the web. The maxim of minimation is Quine's criterion for ascertaining the merit of an unfamiliar proposition, and it may be understood as composed of two aspects.13

Not surprisingly, the first is simplicity - like his empirically-minded medieval forerunner William of Ockham, Quine holds that a simple explanation with a minimalist vocabulary is more valuable than a more complex one which appeals to notions which we cannot precisely define. Second, the principle of minimal destructiveness guides our judgement. If we encounter something that does seem to be the case, but runs contrary to that which we previously held to be true, we revise our web of belief in such a way that the new proposition is agreeably incorporated in the way that has the least effect on the entire significance of the web.

Quine has studied the analytic-synthetic relation from all plausible angles, and all conceivable reasons for making the distinction have dissolved under his scrutiny. The notions of analycity and a priori, he concludes, are unnecessary fictions. There is no need to postulate a special kind of truth in order to account for reality as purposively and regularly structured. All sentences are corrigible and revisable, mere parts of the web of belief which get shifted in respect of its entire significance, and any attempt to classify or differentiate propositions is demonstrably unfounded.

The terms of Kant's transcendentalist model, intended to explain the conditions of possible experience rather than ontologically commit, are at best metaphorical when viewed through Quine's microscope. Philosophy textbooks often refer to Kant's first Critique as an answer to Hume and the problem of induction. However, Quine's analysis has demonstrated that Kant was dealing with essentially different questions than Hume. Hume was concerned with a rigorous description of what our subjective experience entails, whereas Kant waded through the conditions necessary for a rational agent to experience at all.

The natural sciences were next to come under Quine's microscope, as he picked up on the work of French physicist and philosopher Pierre Duhem. Duhem pointed out that, because of the inevitability of multiple uncontrollable variables, it is not possible conclusively to prove any hypothesis experimentally.14 Duhem's objection fits harmoniously with Quine's demonstration that all sentences are corrigible and revisable. Quine zealously developed it to criticize both reductionism and the logical principles behind hypotheses.

Reductionism is the breaking down of sentences to their consequences. It grew out of Humean pragmatism, but reductionists, such as Rudolf Carnap, went a step further and asserted that a proposition is identifiable with, or logically equivalent to, its consequences in the world.15 In addition to Duhem's point, that no consequences can ever be conclusively proven or relied upon, Quine wished to shift the focus of empirical evidence from these isolated and specific but unsure hypotheses to the entire system of sentences, to the whole web of belief. In terms of the logical principles supposed to support hypotheses, the demonstration that there is an inherent indeterminacy, even in things held to be eternal and unalterable, supports Quine's tenet that all sentences are corrigible.

This point illustrates Quine's tendency toward pragmatism, his commitment to natural science, and his solution of Hume's problem of induction. Synonymy is pragmatically accounted for, and with it is accounted for our ability to learn and to perceive experience as regularly structured, in terms of the observable consequences of our words. Quine's notion is behavioristically comparative, analogous to the resemblance theory of universals. Words or expressions that are similarly reacted to, that are employed in the same way, must have similar status, and in this way, strangely, pragmatism accounts for synonymy of meaning without postulating the reality of meanings.16

Natural science is shown to be continuous with philosophy, because it is no longer concerned with ultimate truths, but with data that is corrigible and revisable that fits agreeably into the web of belief. The scientist is now governed by the maxim of minimation - he or she is seeking to find and retain the data which accords with the web. Philosophy has a broader scope than natural science because it also deals with the meta-questions; for example, what exactly it means to be anything, or what, precisely, one is doing when one asserts a predicate of a subject.

Hume's problem is, in one sense, left intact, because all is revisable, and one could never be ultimately or conclusively certain of anything. On the other hand, though, the scientist and the philosopher do get valuably close to absolute knowledge, and that insight provides the empirical criteria for viewing the world as it is. Science retains its present form and present utility; he has merely shifted the logical characterization of theorumhood.17 Quine treads a mediating path between retiring to backgammon with Hume, because metaphysics is fundamentally flawed, and postulating virtual realities, which do not unambiguously ontologically commit, to account for experience, as idealist philosophers tend to do.

Exactly what kind of model is Quine's, and to what type of world is he committed? His position is comprehensive and pragmatic - his desert landscape is illuminated by a holistic account drained of its idealist flavour. Pretentious and dogmatic aspects which empirically flawed preceding holistic models, such as Hegel's Absolute, are trimmed away by Ockham's razor, newly honed to a lethal edge by the shattering of long-supposed-to-be necessary distinctions and connections. Quine's web of belief is influenced by, and encompasses, the entire scope of reality. It is established with a minimalist vocabulary, and is an efficient and integral vehicle toward his metaphysical end unambiguous ontological commitment, which leads to a somewhat bleak but rigorous membership in the world. Quine agrees with Franz Brentano that universals have no existential import, so particulars are all that are the case, all that inhabit this desert landscaper.18 Physical objects exhaust the domain of substance, and man becomes a mere four dimensional physical object. All states of mind are psychologized, or reduced to their impact on behaviour. Effectively, idealist criticisms have not simply been taken note of, but idealism has been hijacked, and the result is a new kind of empiricism and an original view of the world.

Notes

1. W.V. Quine, Two Dogmas of Empiricism. From a Logical Point of View (Cambridge: Harvard University Press. 1953), 37.
2. David Hume. A Treatise of Human Nature. ed. Ernest C. Mossner (London: Penguin, 1969), 117-230. This is a very superficial paraphrase of Book 1, Part 3.
3. Stephen M. Cahn, Classics of Western Philosophy (Indianapolis: Hackett Publishing Co., 1977), 436.
4. Cahn, 978.
5. Bertrand Russell. "On Denoting." Logic and Knowledge, ed. Robert Charles Marsh (London: Unwin Hyman Limited, 1956), 45.
6. W.V. Quine, "A Logistical Approach to the Ontological Problem." The Ways of Paradox (Cambridge: Harvard University Press. 1976), 198.
7. Quine. "Logistical Approach," 201.
8. Immanuel Kant. Critique of Pure Reason. trans. Norman Kemp Smith (London: Macmillan Education Limited, 1929), 47.
9. Quine. "Two Dogmas." 21.
10. Quine, "Two Dogmas." 37.
11. Quine, "Two Dogmas," 22 24.
12. Quine, "Two Dogmas." 24-42.
13. Quine, "Two Dogmas." 44-45.
14. "Duhem, Pierre." Encyclopedia of Philosopy, ed. Paul Edwards (New York: Macmillan Company, 1967), Vol. 2, 423.
15. Quine. "Two Dogmas," 41.
16. Quine, "Two Dogmas," 45.
17. W.V. Quine, "Meaning and Inferemce." From a logical Point of View, 162.
18. W.V. Quine. "On Carnap's Views of Ontology." The Ways of Paradox, 204.

Bibliography

Cahn, Stephen M., ed. Classics of Western Philosophy. Indianapolis:
    Hackett Publishing Company, 1977.

Edwards, Paul, ed. Encyclopedia of Philosophy. Vol.2. New York:
     Macmillan Company, 1967.

Hume. David.  A Treatise of Human Nature. ed. Ernest C. Mossner.
     London: Penguin Books, 1969.

Kant, Immanuel. Critique of Pure Reason, trans. Norman Kemp Smith.
     London: Macmillan Education Limited, 1929.

Quine, W,V. The Ways of Paradox. Cambridge: Harvard University Press.
     1976.

-. From A Logical Point of View. Cambridge: Harvard University Press,
     1953.

Russell, Bertrand. Logic and Knowledge. ed. Robert Charles Marsh.
     London: Unwin Hyman Limited, 1956.