The Analytic/Synthetic Distinction

Here are two sets of sentences. Read them and see if you notice a difference between them.

Set I:

All doctors that specialize on children are rich.
All pediatricians are rich.
Everyone who runs damages their body.
If Holmes killed Sikes, then Watson must be dead.

Set II:

All doctors that specialize on children are doctors.
All pediatricians are doctors.
Everyone who runs moves.
If Holmes killed Sikes, then Sikes must be dead.

When you read Set II, something feels different. You don’t wonder whether those sentences are true or false in the way you do with Set I. They seem obviously, unavoidably true. “Well, of course all pediatricians are doctors,” you might think, “that’s just what the word means.” And if someone tried to deny one—“Not all pediatricians are doctors, some aren’t at all!”—you’d probably think they were confused about words.

Philosophers call sentences like those in Set II analytic. Sentences like those in Set I are synthetic.

Here’s the basic idea: an analytic statement is true just by virtue of what its words mean. A synthetic statement is true (or false) depending on how the world actually is. “All bachelors are unmarried” is analytic. “All bachelors are lonely” is synthetic—you’d have to check.

This distinction might seem simple, even boring. But philosophers have argued about it fiercely for over two hundred years. Because if you think about it carefully, the line between “true by meaning” and “true by fact” starts to blur.

Why Philosophers Got Excited

Here’s what tempted so many philosophers about analyticity. Think about mathematics. How do we know that 7+5=12? We don’t go around checking pairs of groups to see if they always make the right total. Mathematicians don’t run experiments. They sit in chairs and think. But somehow they seem to discover real truths about the world—truths that help us build bridges and launch rockets.

How is that possible? How can pure thinking tell us anything about reality? That’s one of the oldest puzzles in philosophy.

Some philosophers thought analyticity could solve it. Maybe mathematical truths are analytic—true just because of what the concepts mean. If “7+5=12” is analytic, then we don’t need experience to know it. We just need to understand the concepts involved. That would explain how we can know mathematical truths from an armchair.

Immanuel Kant, the philosopher who first made this distinction famous, actually disagreed. He thought “7+5=12” was synthetic—the concept of 12 isn’t contained in the concepts of 7, 5, and plus. But he also thought it was knowable without experience. This led him to the strange category of “synthetic a priori” knowledge, which became one of the most debated ideas in philosophy.

Later, Gottlob Frege developed a much more precise theory. He created modern logic—the system of symbols and rules you might see in a philosophy or math class. He defined a logical truth as a sentence that stays true no matter what you substitute for its non-logical words. For example, “All doctors that specialize on children are doctors” stays true even if you replace “doctors” with “cats” and “specialize on children” with “chase mice,” getting “All cats that chase mice are cats.” That’s a logical truth.

But “All pediatricians are doctors” isn’t a logical truth in that strict sense. Replace “pediatricians” with “mice” and “doctors” with “cats,” and you get “All mice are cats”—false. So to capture this sentence as analytic, Frege said you need to swap in definitions. If “pediatrician” is defined as “doctor that specializes on children,” then the sentence becomes the logical truth above.

This made the analytic-synthetic distinction seem precise and useful. And it led to an ambitious program: maybe all of mathematics could be reduced to logic through definitions. Maybe even our knowledge of the physical world could be analyzed into logic plus experience. This was the dream of a movement called Logical Positivism.

The Trouble Begins

But problems started showing up.

First, there was the “paradox of analysis.” If an analysis gives you the definition of a term—like “brother” means “male sibling”—then it should be uninformative. “Brothers are male siblings” should be the same as “Brothers are brothers.” But real philosophical analyses are supposed to be informative. They’re supposed to teach us something. How can a statement be both analytic and informative? Nobody had a satisfying answer.

More seriously, Willard Van Orman Quine launched a full-scale attack on the whole distinction in the 1950s. His argument went something like this: What does it even mean to say a sentence is “true by virtue of meaning”? How would you test that? You’d need to distinguish between facts about meanings and facts about the world. But Quine argued that there’s no non-circular way to do this. Every attempt to define “analytic” appeals to notions like “synonymy” (same meaning) or “definition,” which themselves need explaining. And those explanations appeal back to “analytic” or something equally unclear. It’s a small circle of concepts that explain each other but never really connect to anything solid.

Quine also argued that even our beliefs about logic and mathematics might be revisable if experience pushed us hard enough. Scientists sometimes change their theories in big ways. Maybe, in principle, we could change even our logic. If that’s possible, then no statement is absolutely immune to revision based on experience. And if no statement is absolutely immune, then nothing is really analytic.

This view is called confirmation holism—the idea that our beliefs face the test of experience not one by one, but as a whole system. You can always protect any particular belief by adjusting others around it. So the distinction between beliefs that are “true by meaning” and beliefs that are “true by fact” doesn’t hold up.

What About Obvious Cases?

But hold on. “All pediatricians are doctors” seems genuinely different from “All pediatricians are rich.” Any ordinary person can tell the difference. Are we really supposed to believe there’s no distinction at all?

Quine had an explanation. Some beliefs are more “central” to our web of beliefs—harder to give up, more protected from experience. The analytic-sounding ones are just the most central, most stubborn beliefs. But they’re not a fundamentally different kind of belief.

This explanation has problems too. “The earth has existed for more than five minutes” is extremely central—hardly anyone would give it up. But it doesn’t feel analytic. Meanwhile, “Bachelors are unmarried” is pretty trivial and could be changed easily if we wanted (just change the definition). But it feels analytic. So centrality doesn’t match the feeling.

What about the fact that you can’t even imagine a married bachelor? That seems special. But Quine would say: that just shows how deeply you hold the belief, not that it’s a different kind of truth. People once couldn’t imagine curved space either.

Where Things Stand Now

Philosophers still disagree about whether the analytic-synthetic distinction is real. Some, following Quine, think it was a mistake from the start. Others think Quine was too harsh—there really is something special about statements that are true by definition.

One interesting recent approach comes from linguistics. Noam Chomsky and his followers argue that human language has deep, innate structures. Maybe analytic truths reflect facts about this language faculty—about how our minds naturally organize concepts. This wouldn’t give us knowledge of the external world, but it would explain why some statements feel so certain.

Another approach tries to ground meaning in how words actually connect to the world. On this view, a word means what it does because of causal relationships between our brains and things out there. “Horse” means horse because we use it when horses are around, and other uses depend on that basic connection. This might give us a natural, non-mysterious way to talk about meaning without relying on introspection or armchair intuition.

But even if we rescue the notion of analyticity, it’s not clear it does the philosophical work people wanted. An analytic truth might tell you about your concepts—about how your mind works—but not necessarily about the world outside your mind. That “bachelors are unmarried” can’t be imagined false might just be a fact about your concept of bachelor, not a deep truth about the universe.

So What?

The debate over analyticity matters because it’s really about whether philosophy can discover truths just by thinking. If there are analytic truths, then armchair reflection makes sense. Philosophers can analyze concepts and learn something. If there aren’t, or if analytic truths only tell us about our own minds, then philosophy might need to be more like science—connected to evidence, humble about its methods.

Lots of people find this unsettling. Philosophy has traditionally claimed to discover important truths about knowledge, justice, consciousness, and morality—without running experiments. If analyticity fails, that whole approach might be in trouble.

Other people think that’s fine. Maybe philosophy should be continuous with science. Maybe thinking carefully is valuable, but it doesn’t give you a special pipeline to truth.

The debate isn’t settled. Philosophers still argue about it. What’s remarkable is that such a seemingly simple distinction—between sentences that are true by meaning and sentences that are true by fact—turns out to be so hard to pin down. It’s a good reminder that even our most basic ways of dividing up the world can become mysterious when you look at them closely.

Appendices

Key Terms

Term	What it does in this debate
Analytic	A sentence that is true (or false) just by virtue of what its words mean, like “All bachelors are unmarried”
Synthetic	A sentence whose truth depends on how the world actually is, like “All bachelors are lonely”
Logical truth	A sentence that stays true no matter what you substitute for its non-logical words, like “All doctors that specialize on children are doctors”
Confirmation holism	The view that our beliefs face the test of experience as a whole system, not one by one
A priori	Knowable without relying on experience (often linked to analyticity, but not the same thing)

Key People

Immanuel Kant (1724–1804): A German philosopher who first made the analytic-synthetic distinction famous. He thought “7+5=12” was synthetic but still knowable without experience.
Gottlob Frege (1848–1925): A German logician and philosopher who created modern logic and tried to show that mathematics could be reduced to logic through definitions.
W.V.O. Quine (1908–2000): An American philosopher who launched the most famous attack on the analytic-synthetic distinction, arguing it couldn’t be made sense of.
Noam Chomsky (born 1928): An American linguist who revolutionized the study of language and whose work offers a way to potentially rescue the analytic-synthetic distinction within a scientific framework.

Things to Think About

If a computer were programmed to use the word “bachelor” correctly, would “All bachelors are unmarried” be analytic for it? What if it could never learn the definition—just used the word correctly in every situation?
“Cats are animals” seems analytic. But what if we discovered that the things we call cats were actually clever robots from Mars? Would “cats are animals” turn out to have been false all along? Or would we just have discovered that the things aren’t really cats?
Quine’s criticism suggests that meaning and belief can’t be neatly separated. But if they can’t, how do you explain the fact that you can think “Jane is a pediatrician” without instantly thinking “Jane is a doctor”? Doesn’t that prove the meaning of “pediatrician” is separate from your belief that all pediatricians are doctors?
Some philosophers think the analytic-synthetic debate is really about whether philosophy can discover truths from the armchair. Do you think there are truths you can discover just by thinking carefully about concepts? Or does all knowledge ultimately come from experience?

Where This Shows Up

Legal arguments often turn on the “meaning” of words. Courts have debated whether “marriage” by definition involves a man and a woman—a question about analyticity.
Computer programming and artificial intelligence face similar issues. Can a computer understand meaning, or does it just manipulate symbols according to rules? That’s a version of the analyticity question.
Science education sometimes involves this debate. When a teacher says “by definition, water is H2O,” are they stating a fact discovered through experiment or a definition we’ve chosen?
Everyday arguments like “That’s not really a sandwich” or “By definition, that’s not art” appeal to analyticity—they’re trying to settle disputes by appealing to meanings rather than facts.