Find Enlightenment

The Analytic-Synthetic Distinction
A Critique And An Alternative Hypothesis
by Diana Mertz Hsieh

Date: 9 May 96
Forum: Washington University in St-Louis, Kant class
Copyright: Diana Mertz Hsieh

Through all of Kant's philosophy, his most enduring legacy has surely been the analytic-synthetic distinction. Disputes over what particular statements are analytic or synthetic aside, most philosophers today accept the distinction, in some form or other, as legitimate. Quine's influential attack on the distinction in "The Two Dogmas of Empiricism" has reportedly had little impact on Kant scholars. Regardless of present-day philosophers' acceptance of the analytic-synthetic distinction, it is clear that it lies at the heat of Kant's transcendental idealism. Without it, his attempt to ground metaphysics in synthetic, a priori principles becomes a meaningless project. But more importantly, criticism of the analytic-synthetic distinction brings Kant's basic views about the nature of concepts and relations between them into question. This examination could be far more devastating, or illuminating, to Kantianism than a simple elimination of the analytic-synthetic distinction.

In this paper, I will examine how Kant draws the distinction between the analytic and the synthetic, through the criteria of conceptual containment, identity, and the principle of contradiction. I will use those criteria as a way of critically investigating Kant's implicit view of concepts. Finally, I will briefly discuss Quine's alternative hypothesis for the analytic-synthetic distinction, and propose my own alternative view of the facts which (at least some) philosophers may misuse in erroneously accepting the distinction.

Separating the analytic and the synthetic

Kant uses at least three distinct criteria to distinguish the analytic from the synthetic: conceptual containment, identity relation, and non-contradiction. These criteria, although seemingly disparate, are related though Kant's view of concepts as constituted by one set of a priori marks and loosely related to another set of marks, with each mark being itself a concept. By examining the view of concepts that underlies Kant's account of analyticity and syntheticity, it becomes clear that the three criteria are one and the same.

In the Introduction to the Critique of Pure Reason, Kant explicates the criterion of conceptual containment, which states that a judgment is analytic if the predicate "belongs to the subject" and is synthetic if the predicate "lies outside the [subject], although it does indeed stand in connection with it." (A6/B10) An analytic judgment adds "nothing through the predicate to the concept of the subject, but merely breaking it up into those constituent concepts that have all along been thought in it, although confusedly." (A7/B11) The predicate of a synthetic judgment, on the other hand, could not be "extracted" from the subject by any form analysis. (A7/B11)

This criterion of containment relies heavily upon Kant's "notion of a concept as a set of marks (themselves concepts), which are thought together in 'analytic unity.'" (Allison 76) Conceptual containment arises when the predicate "is itself either a mark of the concept or a mark of one of its marks." (Allison 76) In other words, it is the interconnectedness of marks (i.e. concepts) that gives rise to the relations necessary for analytic judgments. In certain cases, the analyticity of a judgment is perfectly apparent using this criteria, as in the cases of "sparrows are birds" and "women are rational animals."

In other cases, however, the analyticity is uncertain, such as with the judgments "desks are types of tables" and "viruses are living creatures." (At least I am not sure whether a desk ought to be classified as a species of table or just as furniture, and the question of whether viruses are alive has not, to my knowledge, been resolved by the scientific community. Therefore, I cannot say whether these judgments are analytic or synthetic.) One solution would be to rule these examples "judgments of experience," which would preclude them from being analytic, since judgments of experience. . . are one and all synthetic." (B12) But as Kant states in the Prolegomena, concepts of experience can constitute analytic judgments, such as in the case "gold is a yellow metal," which "require[s] no experience beyond [the] concept of gold as a yellow metal." (Prolegomena 267-8) So, identifying the genus of "desk" or "virus" cannot, on Kantian terms, be considered moving outside those concepts and "appeal[ing] to the testimony of experience." (B12) And thus the problem raised by these examples of judgments of indeterminate analyticity remains.

Grice and Strawson, in their essay "In Defense of a Dogma," discount the importance (and even frequency) of disagreement over when "analytic" and "synthetic" ought to be applied. They write that those who use the terms 'analytic' and 'synthetic' do to a very considerable extent agree in the applications they make of them. They apply the term 'analytic' to more or less the same cases, withhold it from more or less the same cases, and hesitate over more or less the same cases. (341)

But this presentation of the difficulty of how to apply the distinction underestimates the seriousness of the disagreements over that application. Philosophers do more than "hesitate" over whole bodies of knowledge like mathematics, they vehemently disagree. There is not even significant agreement over Kant's own example of"7+5=12" being a synthetic statement.

Grice and Strawson might be warranted in dismissing the problem of disagreement and indeterminacy in applying the distinction if one of Kant's primary purposes in developing his transcendental idealism had not been explicating metaphysics as a science, i.e. as a system of "universal and lasting recognition." (Prolegomena 255-6) But because of that goal, disagreement over principles or application indicates a failure of the theory and thus cannot be disregarded as inconsequential, particularly when that disagreement involves so crucial a Kantian distinction as between the analytic and the synthetic.

Lewis Beck brings up another problem concerning the criterion of conceptual containment, namely that Kant both asserts that (with respect to the example of a triangle) "[the concept] is nothing more than the mere definition" and rejects a conception of analyticity based on definition, since "'definability' is a stricter condition than 'analyzability.'"(A718/B746 and Beck 6) In fact, Kant does not regard definitions of concepts as giving rise to analytical judgments, rather it is through "analytical judgments [that] we seek to approach the definition of the concepts." (Prolegomena 272-3) And so what Kant thinks is actually contained in a concept remains obscure. Beck also points out that concepts need to be sufficiently fixed, not "dependent upon variable conditions of how much the judger knows about the subject of the judgments and on his arbitrary decision of the choice and formula of his definitions," in order for the classification of judgments into analytic and synthetic not to be arbitrary. Conceptual containment, due to its problems of under-determining the classification of the analytic and the synthetic, cannot resist such arbitrariness. But perhaps the criteria of identity or non-contradiction can.

In clarifying what he means by conceptual containment, Kant introduces the criterion of identity, writing that analytic judgments are those "in which the connection of the predicate with the subject is thought through identity." (A7/B10) Synthetic judgments are defined negatively as those in which the "this connection is thought without identity." (A7/B10) This identity condition seems to only raise more questions. Surely Kant would want to regard the judgment "humans have two eyes" as synthetic, since it is an empirical proposition. And yet, in what sense is two-eyedness not part of the identity of humans?

The answer lies in Kant's explication of the example "all bodies are extended," which makes plainer the implicit view of concepts grounding Kant's analytic-synthetic distinction. In that example, extendedness (and impenetrability and figure) do not go beyond the concept of body, whereas weight does. Kant explains that one can "attach [weight] as a predicate to the concept" by "finding weight to be invariably connected with the above characters [of extendedness, figure, etc.]." That attachment is synthetic and an extension of one's knowledge. Despite that attachment however, Kant denies that we "include in the concept of a body in the general predicate 'weight.'" (B12) And so, even though "weight" is attached to the concept of body, a judgment connecting the two is still synthetic.

This example illustrates Kant's implicit view (in both the criteria of containment and identity) of a division between what I shall call "primary marks" and "secondary marks." Primary marks analytically constitute the concept, and are thus without empirical content. In the case of "body," the mark "extension" is primary. Secondary marks are synthetically tied to the concept, and can be empirical. The mark of "weight" is a secondary one in relation to "body." Beck seems to notice much the same division when he notes that there seems to be a tacit distinction between two kinds of concepts, one being a concept of highly refined analytical or abstractive unity, subject to strict definitions, and the other being a looser complex of representations, more or less held together and expandable through the accretion of new experience or subject to restriction in content through the supervention of a definition. (Beck 9)

The warrant for this implicit distinction is left, as one might expect, completely unspecified. This division of concepts into to distinct parts does, in some ways, resembles the old essence-accident distinction made in relation to substances. But that distinction is on much firmer ground, in that it depends upon the ways in which a substance could change while still being the same substance, whereas this primary-secondary mark distinction has no clear boundaries.

Those lack of distinct edges with respect to primary and secondary marks recreates the problem of the resolving disputes about the application of the analytic-synthetic distinction. Given Kant's goal of creating a science of metaphysics, to leave such an important distinction unclarified is particularly deadly to his system. If we attempt to use this primary-secondary mark distinction to determine the analytic predicates of a concept, we shall immediately run into difficulty. In analyzing the "all bodies are extended" example, Kant writes that "to meet with this predicate, I have to merely analyse the concept, that is, to become conscious to myself of the manifold which I always think in that concept." (A7/B11) But since we have attached (albeit synthetically) other predicates that "[belong] together with the concept," it is unclear how we can sort out the primary from the secondary marks, because the secondary marks have become part of the manifold which always accompanies that concept. (B12) If disagreements between philosophers emerge about which marks are secondary and which are primary, there is little more than intuition (in the non-Kantian sense) to appeal to.

So the criteria of identity seems to raise more problems than it solves. Nevertheless, it does help clarify the Kant's underlying view of concepts which informs his analytic-synthetic distinction. His reliance on the principle of contradiction as the "highest principle of all analytic judgments" in the Transcendental Analytic and the Prolegomena sheds even more light on the analytic-synthetic and primary-secondary mark distinctions. (A154/B193)

In the Transcendental Analytic, Kant states that "the principle of contradiction must therefore be recognized as being the universal and completely sufficient principle of all analytic knowledge." (A151/B191) We can determine the truth of analytic claims by appealing to nothing more than that principle, which Kant formulates as "it is impossible that something should at one and the same time both be and not be." (A152/B191) That the principle of contradiction is a "sufficient criterion of truth" for the analytic also indicates that we should be able to determine the analyticity of a judgment by ÷ testing that judgment (or rather, its negation) against the principle of contradiction. So, for example, we can conclude that "all bachelors are unmarried" is both analytic and true, since its negation asserts a contradiction. Synthetic judgments, however, "require a quite different principle from which they may be deduced, subject, of course, to the law of contradiction," namely "inner sense and its a priori form, time." (Prolegomena 267-8 and A155/B194)

This criterion for analyticity depends upon Kant's particular conception of contradiction, i.e. what it means for a judgment to assert that something both is and isn't at the same time. The distinction Kant draws between falsehood and contradiction once again illuminates the underlying separation of primary and secondary marks, the former of which constitute a concept's identity. Contradiction is "inner contradiction," whereas "if our judgment contains no contradiction, it may yet connect concepts in a manner not borne out by the object. . . and so may still, in spite of being free from all inner contradiction, be. . . false." (A150/B190) Thus even though we have added the predicate "weight" to the concept of body, denying that all bodies are heavy is a mere falsehood, not a judgment of the form "a and not-a."

With this criterion of non-contradiction for analyticity, Kant's view that it is only the primary marks of a concept form that concept's identity becomes more explicit. Although we know that all bodies have weight and although we have attached the predicate of weight to the concept of body, to deny "all bodies are heavy" does not involve a contradiction. The secondary marks do not constitute the identity of the concept, and so to deny them is not to assert that something both is and is not.

Once again, the warrant for this conception of contradiction, including the distinction between falsehood and contradiction, is absent. The origin of the law of non-contradiction in Aristotle contains no such implicit view of concepts or distinction between falsehood and contradiction. In Metaphysics, Aristotle defines truth and falsehood in the following fashion: "A falsity is a statement of that which is that it is not, or of that which is not that it is; and a truth is a statement of that which is that is is or of that which is not that it is not." (1011b) There is no distinction between contradictory and false statements; a false statement is defined as a contradictory one. Without Kant's false-contradiction distinction, there cannot be two types of classes of statements, determined by the negations of which involve a contradiction.

Aristotle's account of contradiction is at least as plausible as Kant's, thus calling into question the justification for Kant's view. On Kant's own terms, we can wonder whether the distinction between falsehood and contradiction is a priori and whether the definitions that he gives are analytic. Indeed, we can wonder whether Kant's implicit view of concepts as having both primary and secondary marks is a priori and analytic. At this point, it begins to seem as if Kant cannot even get off the ground, because if he makes the a priori-a posteriori and analytic-synthetic distinctions, he must justify those distinctions in reference to his own theories about a prioricity and analyticity. If he cannot justify those distinctions, then they cannot serve as the bedrock for the science of metaphysics that he wishes to develop. But attempting to find a warrant those basic views in a non-circular fashion would be no small feat.

Holism and Knowledge

At this point, perhaps it is best to leave the attempt to clarify and justify the analytic-synthetic distinction behind, in order to attempt an alternate account of what philosophers are distinguishing in judgments when they classify them as analytic or synthetic. In response to Quine's "Two Dogmas of Empiricism," Grice and Strawson characterize «Quine's position as indicating that he thinks that philosophers have been influenced to accept the analytic-synthetic distinction due to a bad theory. They write:

Suppose that there were [sic] a particular mistaken theory about language or knowledge, such that, seen in the light of this theory, some statements. . . appeared to have this characteristic which no statements really have. . . [and others which] did not appear to have this characteristic, and others again which presented an uncertain appearance. (Grice and Strawson 342) More precisely, Quine seems to think that there is an underlying correct view of knowledge (namely holism) which philosophers implicitly use. Because they do (on some level) accept holism, they regard some beliefs as more stable than others, and classify those beliefs as analytic. But because they do not accept holism explicitly, they do not see that all beliefs are, in principle, subject to revision, i.e. that the differences between the analytic and the synthetic are merely a matter of degree rather than one of kind. When holism is explicitly accepted, "it becomes folly to see a boundary between synthetic statements, which hold contingently on experience, and analytic statements, which hold come what may" because "any statement can be held come what may, if we make drastic enough adjustments elsewhere in the system." (Quine 211)

I think that another account of what underlies the analytic-synthetic distinction can be made, without relying upon a theory so radical as Quine's holism. As was clear in all three of the criteria used to delimit analyticity, analytic statements are conceived of as those statement whose truth value can be known without the testimony of experience. But all statements necessitate some appeal to experience, unless one is radically committed to innate concepts. If we adopt this empiricist approach, the difference between analytic and synthetic statements is one of degree, not of kind. On the analytic end of the spectrum are those statements which require, in order to determine their truth value, no more experience than was necessary to form the concepts, whereas synthetic statements are those which do require experience in addition to those required by the concepts themselves. So, for example, the statement "my father is male" tends towards the analytic, because one cannot be said to know what fathers are if one does not also know that they are male. The statement "Sarah's dog is gray" would lie on the synthetic end of the spectrum, since one could know about Sarah's dog without knowing its particular color.

By conceptualizing the grounding of the analytic-synthetic distinction in this fashion, it becomes clear that there is no longer a distinction to be made at all. Rather, there is a continuum of statements, based on the amount of information which establishing their truth value requires. The distinction/continuum is also highly relativistic, not just in the sense that statements can shift their position on the continuum through time, but that positioning will inevitably vary from person to person, based on their knowledge of (the referents of) a given concept. So, for example, due to my experience with horses, my concept "horse" is probably significantly richer than most people's, so that I would regard the statement "horses are taller than 14.2 hands" as much more analytic than most.

This explanation of the analytic-synthetic distinction, as in the case of Quine's appeal of holism, is not a new way of validating it, albeit in a highly distorted form. It is merely a suggestion as to why philosophers (and perhaps only certain philosophers) find the distinction plausible. But given the relativism of the distinction, along with the fact that it is no longer even a distinction (but rather a continuum), there seems to be little good reason µto retain it at all. And like Quine's holism, this argument about what the actual facts underlying the analytic-synthetic distinction is little more than a rough sketch of an account or a basic plausibility argument. But given all of the problems in clarifying and justifying the analytic-synthetic distinction, particularly in determining the criteria for analyticity and syntheticity, these rough sketches and plausibility arguments seem warranted.

Works Cited

Allison, Henry. Kant's Transcendental Idealism: An Interpretation and a Defense. New Haven: Yale UP, 1983. 65-80.

Aristotle. Metaphysics. trans. Hippocrates Apostle. Bloomington: Indiana UP, 1966. 1005a-1011b.

Beck, Lewis White. "Can Kant's Synthetic Judgments be Made Analytic?" Kant: a Collection of Critical Essays. Robert Paul Wolff, ed. Notre Dame: University of Notre Dame Press, 1967. 3-22.

Grice, H.P. and P. F. Strawson. "In Defense of a Dogma."Classics of Analytic Philosophy. Robert Ammerman, ed. Indianapolis: Hackett Publishing, 1990. 340-352.

Kant, Immanuel. Critique of Pure Reason. trans. Norman Kemp Smith. New York: St Martin's Press, 1929.

Kant, Immanuel. Prolegomena to Any Future Metaphysics. Revision of the Carus translation. New York: Liberal Arts Press, 1950. 255-280.

Quine, W. V. O. "Two Dogmas of Empiricism."Classics of Analytic Philosophy. Robert Ammerman, ed. Indianapolis: Hackett Publishing, 1990. 196-213.

Waskan, Jonathan. "A Kantian Model of Concepts, Analyticity, and Belief." Final paper for Kant and 19th Century Philosophy at Washington University.

Find Enlightenment at