Quine's Two Dogmas of Empiricism: Attacking the Analytic/Synthetic Distinction

Imagine that a room's temperature sensor reports an unexpected rise. You can question the sensor, the software that converts its readings, or your belief that the heating is off. The reading matters, but it does not arrive with a label identifying the mistaken assumption. Several beliefs contributed to the expectation that failed.

This invented example introduces one side of W. V. Quine's "Two Dogmas of Empiricism." The other side concerns an apparently different question: why does a statement about a bachelor seem true just because of what its words mean? Quine connects these questions by challenging a picture in which every statement has its own separable contribution from language and from experience. Understanding that connection requires more than saying that everything is interconnected.

An analytic truth is traditionally understood as true because of meaning, independently of how contingent events turn out. A synthetic statement depends for its truth on facts beyond meaning. Under an ordinary interpretation, "Every vixen is female" looks like the first kind; "A vixen crossed the garden last night" looks like the second. Learning the vocabulary seems sufficient for one, while observation is needed for the other.

These are descriptions of the distinction under examination, rather than definitions Quine accepts. His question is what gives the supposed boundary its philosophical authority. Familiar examples may produce agreement without explaining what makes the examples belong together. If a theory assigns analyticity a foundational role, the distinction needs an account adequate to that role.

The paper appeared in The Philosophical Review in 1951, volume 60, pages 20–43. Its opening identifies two assumptions of modern empiricism: a fundamental separation of analytic from synthetic truths, and reductionism about meaningful statements. Quine discusses Rudolf Carnap's work closely, while also considering older approaches to meaning and experience. He remains concerned with empirical inquiry; rejecting these assumptions is his proposed reform of empiricism. Quine, opening and §§1–6.

The linked transcription distinguishes passages from the original paper and the revised version collected in From a Logical Point of View. The discussion here uses section references because pagination differs across editions. This also prevents a later revision from being silently presented as the exact wording of the original journal article.

In §1, Quine separates two kinds of examples commonly called analytic. A logical truth remains true when its nonlogical vocabulary is reinterpreted while the logical structure is held fixed. For a simple illustration, "If a visitor is early, that visitor is early" retains its form whatever property replaces being early. The repeated predicate does the relevant work.

Other apparent analytic truths depend on vocabulary with substantive meanings. To treat the vixen example as a logical truth, we first replace "vixen" with "female fox." That replacement depends on synonymy, or sameness of meaning. The logical structure alone does not license exchanging arbitrary predicates. If the proposed account of analyticity uses synonyms, it still owes us an explanation of the relevant sameness.

The difficulty is easy to miss because competent speakers can already make many of the intended judgments. Quine is asking for an explanation of their status, not an admission that ordinary speakers are helpless with dictionaries. Knowing how to use a distinction and explaining why it supports a philosophical theory are different achievements.

Reference does not settle the question. Two descriptions can pick out the same object through different information. For an invented example, "the owner of this bicycle" and "the person who rented the upstairs flat" might identify the same person. Discovering that identity could require investigation. Substituting one description for the other would then preserve reference without making the discovery a consequence of vocabulary alone.

The natural answer is that a dictionary defines the relevant word. In §2, Quine asks what such a definition records. A lexicographer normally reports established usage. If a dictionary correctly explains a word by another expression, the explanation is supposed to fit a linguistic practice that already exists. The dictionary can be excellent evidence about that practice without creating its underlying relation of synonymy.

This is a question of explanatory direction. Imagine checking whether two measuring instruments agree by reading a report that says they agree. The report may be trustworthy, but it cannot also explain what agreement between the instruments consists in. Similarly, citing a dictionary entry does not by itself analyze the relation that makes the entry correct.

Quine recognizes an important exception: introducing a new notation expressly as an abbreviation. Suppose a policy document introduces "eligible visitor" to abbreviate "registered visitor aged eighteen or older." Within that document, the stipulated equivalence is transparent. Nothing about the argument requires us to stop making such definitions. Quine, §2.

The exception limits an overbroad reading of his criticism. His objection is not that definitions never establish synonymy. It is that this special case does not explain the whole range of relations that ordinary definitions presuppose. A successful local stipulation cannot automatically provide an account of meaning throughout a natural language.

The policy example also reveals two questions that can come apart. We can ask whether a visitor satisfies the stated condition, and we can ask whether the condition is a sensible one for the policy's purpose. The definition answers the first kind of question only after relevant facts are supplied. It does not make the policy immune to criticism.

Another proposal identifies synonymous expressions by their interchangeability without changing truth value. Section 3 examines why this is difficult. We must first specify the contexts in which substitution is permitted. Quotations, for example, concern expressions themselves: the word "vixen" has five letters, while the phrase "female fox" does not. A qualification excluding such contexts is needed before the proposal even gets started.

More substantially, predicates can apply to exactly the same objects without meaning the same thing. In a fictional collection, every red folder might be a confidential folder, and every confidential folder red. Substituting these predicates in ordinary extensional statements about the collection would preserve truth. Nevertheless, discovering that coincidence is different from learning the meanings of "red" and "confidential."

An extensional context is one in which replacement by expressions with the same reference or extension preserves truth value. An extension is the collection of things to which a predicate applies. The folder example shows why agreement in extension is weaker than the kind of synonymy sought. The agreement might disappear if somebody changes a folder's colour without changing its confidentiality.

One might require substitution to preserve necessary truth instead. Quine's concern is that an unexplained appeal to necessity can reintroduce the semantic notions that the proposed test was meant to clarify. The issue is not a ban on using modal vocabulary. It is whether that vocabulary provides an independent explanation in this argument. Quine, §3.

Section 4 considers explicit semantic rules for constructed languages. A list can identify a class of sentences with perfect precision. Quine asks what makes the selected rules rules of meaning, and the selected truths analytic, rather than simply a designated class. His challenge concerns the interpretation of the classification, not our ability to manipulate a formally specified system. That distinction matters when assessing both his objection and replies to it.

Reductionism initially appears as a proposal to translate meaningful statements into a language of immediate experience and logic. Section 5 considers Carnap's attempt in the Aufbau and distinguishes that ambitious project from a weaker assumption. Even if literal translation is abandoned, one might still suppose each statement has its own independently specifiable pattern of confirming and disconfirming experiences.

Quine challenges this weaker assumption too. A prediction about the temperature reading depends on claims about the instrument, its environment, and how the displayed number should be interpreted. The experience tests an arrangement of assumptions. Treating one sentence as possessing a complete evidential account independently of its companions leaves out those dependencies.

This is confirmation holism: the significance of evidence for a claim depends on a wider body of beliefs. It should be distinguished from the loose observation that ideas influence one another. The stronger philosophical claim concerns what encounters empirical testing. In §§5–6, Quine presents that unit in sweeping terms, extending it to science collectively. Quine, §§5–6.

The analytic/synthetic distinction enters again because an analytic truth could be treated as a limiting case: a sentence confirmed whatever experience occurs. But that characterization assumes we can assign an individual sentence its own confirmation conditions in the first place. Quine's attack on isolated confirmation therefore bears on one proposed way of identifying analyticity. This explains why the two criticisms belong in the same paper.

Return to the imaginary sensor. Suppose our prediction depends jointly on three assumptions: the sensor is calibrated, the heating is off, and no other source is warming the room. Suppose also that, given those assumptions, our model predicts a stable reading. An unexpected rise conflicts with that package. It does not deductively identify which component failed.

This conclusion has a precise logical limit. If assumptions A, B, and C together entail prediction P, and P is false, then at least one assumption in the package must be rejected. It does not follow that A specifically is false. Nor does it follow that A, B, and C are all false. The same structure can be understood without symbols: the failure of a joint guarantee does not identify the failed contributor.

Further inquiry can narrow the options. A second independently checked instrument might weaken the calibration explanation. Inspecting the heating controls might reveal an unnoticed setting. These are additional tests with their own assumptions. Nothing in this reasoning says inquiry cannot make progress or that every proposed explanation deserves equal confidence.

The example illustrates a reason to take holism seriously, but it establishes less than the whole of Quine's program. A familiar dependence on auxiliary assumptions does not by itself prove that every scientific belief participates equally in every test. The extent of the relevant system remains a substantive issue. Readers can accept the limited lesson about joint testing while questioning a more global thesis.

Section 6 describes differences between relatively peripheral claims and central commitments, including logic and mathematics. Peripheral claims have more direct connections with particular experiences. Revising central commitments can require much wider changes. Quine denies an absolute immunity from revision while retaining these differences in role and cost.

His claim that a statement can be retained through adjustments elsewhere should therefore be read with the costs in view. Preserving a claim may require rejecting an observation report or changing other commitments substantially. The availability of such a manoeuvre does not establish its wisdom. Rational assessment still concerns how well the resulting system accommodates experience and how disruptive the repair is.

Consider an invented device manufacturer that blames every failed test on interference. One interference explanation may be independently supported. A series of unsupported explanations introduced solely to protect the device adds assumptions without explaining why the failures occur where they do. Holism makes it necessary to assess the wider package; it does not confer credibility on every package that can be described.

Quine emphasizes considerations such as simplicity and conservatism in revising theory. These guide choices within inquiry. They are not a recipe that guarantees a unique repair for every surprise, and conservatism alone would not establish that an entrenched belief is true. Quine, §6.

Pragmatic criteria for choosing theories answer a different question from a definition of truth. The discussion of conservatism and simplicity does not by itself establish that truth is whatever inquiry eventually converges upon. "Two Dogmas" argues about meaning, empirical significance, and theory revision. Practical convenience alone does not guarantee that a proposition is true; we still need to assess how the theory accommodates experience.

A central objection asks whether Quine demands too much of analyticity. Failure to define a distinction in independent terms does not automatically establish that the distinction is unreal. Several concepts might be explained together. The defender can acknowledge relations among meaning, synonymy, and necessity while denying that these relations make the entire vocabulary unusable.

H. P. Grice and P. F. Strawson develop this line in "In Defense of a Dogma." They question the inference from inadequate philosophical explanations to rejection of the distinction itself. They also argue that ordinary judgments about sameness of meaning deserve weight. Their response is a substantive challenge to Quine's standards, rather than evidence that the controversy was quickly settled. Grice and Strawson, 1956, pp.141–158.

A defender might use an invented translation example. If someone systematically translates "red" as "confidential," we can identify a mistake without first possessing a complete philosophical theory of meaning. This shows that some judgments of difference are well grounded. Quine's supporter can reply that the existence of useful judgments still leaves open whether they support the strong, general boundary claimed by the theory.

This exchange separates two disputes. One concerns whether speakers can make useful semantic distinctions. The other concerns whether those distinctions divide truths into two fundamentally different kinds with the explanatory consequences traditionally claimed. Resolving the first does not automatically resolve the second. Equally, raising doubts about the second does not erase every practical difference between learning vocabulary and investigating a garden.

The sensor example suggests a practical habit: make the assumptions behind a prediction visible before declaring a single claim refuted. In a hypothetical planning meeting, a missed deadline could challenge an estimate of task duration, a staffing assumption, or the expectation that a dependency would arrive. Each explanation suggests different evidence to collect. The philosophical lesson helps formulate the investigation; it does not select the answer in advance.

A second habit is to distinguish a stipulated category from an established fact about its usefulness. A team may define an "active customer" as somebody who logged in during the previous month. Counting people under that rule is one task. Deciding whether the count captures meaningful engagement is another. Changing the definition may improve a report while making comparisons with earlier reports difficult.

These contemporary examples are author-created illustrations, not predictions made by Quine or empirical studies confirming his philosophy. Their value lies in the questions they expose: which assumptions contributed to a judgment, what a definition settles, and what remains open to investigation. They offer a way to use the argument without claiming that the paper supplies a complete theory of management or psychology.

The connection between the two dogmas remains the chapter's difficult point. If evidence bears on connected claims, and linguistic commitments operate within the same inquiry, separating an individual sentence into a purely linguistic part and a purely factual part needs justification. Quine presses that demand. His opponents can dispute whether the demand has been met without denying the familiar usefulness of either definitions or observations.

• W. V. Quine, "Two Dogmas of Empiricism". Read §§1–4 for the semantic arguments and §§5–6 for reductionism and holism; note the transcription's edition markers.
• H. P. Grice and P. F. Strawson, "In Defense of a Dogma", The Philosophical Review 65 (1956), pp.141–158. Read alongside Quine to compare their standards for explaining a distinction.
• W. V. Quine, "Two Dogmas in Retrospect", Canadian Journal of Philosophy 21 (1991), pp.265–274. A later primary reflection; the linked publisher page supplies an extract and publication information.