Curry's Paradox: When Sentences Talk About Themselves

A Strange Thing Your Friend Could Say

Imagine your friend looks at you and says: “If what I’m saying right now is true, then time is infinite.”

That seems like a perfectly normal sentence, right? It’s a conditional—an if-then statement. But something weird is happening here. The sentence refers to itself. It says: if this very sentence is true, then time is infinite.

Here’s the problem. There’s a short, simple argument that seems to prove that your friend’s sentence alone forces time to be infinite. Not that time is infinite, but that the existence of the sentence somehow proves it. And if that’s true, then you could replace “time is infinite” with anything—“all numbers are prime” or “pigs can fly”—and the same argument would prove that too.

That can’t be right. Sentences about themselves shouldn’t be able to prove anything about the world. But figuring out exactly where the argument goes wrong turns out to be surprisingly difficult.

This is Curry’s paradox, named after the logician Haskell Curry, who wrote about it in 1942. And it forces anyone who thinks seriously about truth, sets, or logic to make some hard choices.

How the Argument Works

Let’s walk through it slowly. Call the sentence your friend uttered “K.” What does K say?

K = “If K is true, then time is infinite.”

Now, let’s suppose K is true. If K is true, then what K says must be the case. And what K says is “If K is true, then time is infinite.” So, under the supposition that K is true, we know two things:

If K is true, then time is infinite.
K is true (that’s our supposition).

From (1) and (2), using a rule called modus ponens (which says: if you have “if A then B,” and you have A, you can conclude B), we get:

Under the supposition that K is true, time is infinite.

Now, we’ve shown that if K is true, then time is infinite. But that’s exactly what K says! So we’ve proven K itself.

K is true.

Now apply modus ponens again. We have “if K is true then time is infinite” (that’s K) and we have “K is true.” So:

Time is infinite.

We started with nothing but the existence of a self-referential sentence. We used only what seem like obvious rules about truth and logic. And we ended up proving that time is infinite.

Replace “time is infinite” with any claim you like—something you know is false—and the same argument “proves” it. That’s the paradox.

Where Do Curry Sentences Come From?

The sentence K is what philosophers call a Curry sentence. It’s a sentence that’s equivalent (by the rules of some theory) to a conditional with itself as the “if” part. You can create Curry sentences in several different areas of logic.

Truth and Language

The simplest way is using truth. If you have a theory that treats truth in a “transparent” way—meaning that saying “sentence X is true” means the same thing as sentence X itself—then any sentence that says “If I am true, then P” will be a Curry sentence. That’s what K does.

Sets

You can also get Curry sentences in set theory. In the “naive” version of set theory (the one that seems most natural but leads to paradoxes), for any condition you can state, there’s a set of things that satisfy it. So consider the set of all things that are members of themselves only if time is infinite. Call this set C. Now ask: is C a member of itself?

If C is a member of itself, then C satisfies the condition for membership—which requires that if C is a member of itself, then time is infinite. So from “C is in C” you get “if C is in C then time is infinite.” Using modus ponens, you get “time is infinite.” And from there you can prove that C is in C—and then that time is infinite. The same structure appears.

Properties

The same trick works for properties (the property of being such that you have that property only if time is infinite) and for many other things. Wherever you have a way for sentences to refer to themselves or for things to apply to themselves, Curry sentences can pop up.

What Makes This Different from Other Paradoxes

You might have heard of the Liar paradox: “This sentence is false.” If it’s true, it’s false; if it’s false, it’s true. Curry’s paradox is different in an important way: it doesn’t use the word “false” or “not” at all. It works entirely with “if-then.”

This matters because some philosophers thought they could solve paradoxes by changing how negation works—by saying that sentences can be both true and false, or neither true nor false. But Curry’s paradox shows that these solutions aren’t enough. Even if you fix negation, the conditional can still cause trouble.

As one philosopher put it: Curry’s paradox dashes hopes for easy solutions to the Liar and Russell’s paradox. You can’t just mess with negation and call it a day. You have to say something about “if-then” too.

How Philosophers Try to Escape

Nobody thinks we should actually accept that any sentence can prove anything. So something in the argument must be wrong. The question is what.

Option 1: Reject the Self-Reference

The simplest response is to say that sentences like K aren’t really possible. Maybe self-reference is banned, or truth can’t be applied to sentences that talk about their own truth. This is the approach taken by theories like Tarski’s hierarchical view of truth (where truth is always truth about a lower level) or by set theories that restrict which sets can exist.

These are called Curry-incomplete responses. They accept that any theory that did have Curry sentences would be trivial, but they deny that good theories need to have them. This approach can keep classical logic intact.

Option 2: Change the Logic

But many philosophers think that truth and sets should be able to talk about themselves. They want theories that are Curry-complete—theories where Curry sentences exist but don’t cause triviality. To make that work, you need to reject some logical principle used in the argument. Here are the main possibilities.

Reject Contraction

The argument used a principle that you can call contraction: if from A you can prove “if A then B,” then you can prove “if A then B” without the assumption. (In the argument above, this was the step where we concluded K itself from “under the supposition that K is true, time is infinite.”)

Some philosophers say this contraction is the problem. They develop logics where using an assumption twice is different from using it once. If you need the assumption “K is true” twice in the proof—once to get the conditional and once to apply it—but the logic doesn’t let you reuse it that way, the paradox doesn’t go through.

Reject Modus Ponens

Others say the problem is with modus ponens itself—the rule that says from “if A then B” and A you can conclude B. In some logics, this rule fails for certain conditionals. The sentence “if K is true then time is infinite” might be true, and “K is true” might be true, but “time is infinite” doesn’t follow.

This sounds strange, but there are careful ways to develop this idea. The key insight is that you might accept a conditional as true without accepting that you can detach its consequent from its antecedent in all cases.

Option 3: Reject Conditional Proof

The argument used something called conditional proof: if from assuming A you can prove B, then you can assert “if A then B.” Some philosophers reject this rule, at least for the kind of conditional involved in Curry sentences. They might accept that from “K is true” you can derive “time is infinite” without agreeing that “if K is true then time is infinite” follows.

The Bigger Picture

Curry’s paradox is more than a curiosity. It shows something deep about the structure of logical and semantic paradoxes. Many paradoxes—the Liar, Russell’s paradox, Grelling’s paradox—can be seen as having the same underlying form, which Curry’s paradox reveals in its purest form, without any help from negation.

This has led to arguments about what counts as a “uniform solution” to paradoxes. If two paradoxes have the same structure, should they get the same kind of solution? Or can you solve them differently based on their details? There’s no settled answer.

More recently, philosophers have discovered versions of Curry’s paradox that involve a theory’s own notion of “validity” or “entailment.” These validity Curry paradoxes seem even harder to escape, because they target the very idea of what follows from what. Some philosophers now think these push us toward extremely revisionary logics, where the most basic rules of reasoning have to change.

What Nobody Knows

Nobody has a solution to Curry’s paradox that everyone accepts. Each response comes with costs. Restricting self-reference seems arbitrary to some. Changing logic feels too extreme to others. And every non-classical logic has to confront further challenges, including the validity Curry versions.

The debate continues. But the very fact that a few simple lines of reasoning about a self-referential sentence can lead to such profound questions about truth, logic, and the nature of reasoning—that’s part of what makes Curry’s paradox worth thinking about.

Appendix: Key Terms

Term	What It Does in This Debate
Curry sentence	A sentence that is equivalent to a conditional with itself as the “if” part; it’s the engine that drives the paradox
Modus ponens	The rule that says from “if A then B” and A, you can conclude B; it’s used in the paradoxical argument
Contraction	A principle saying that using an assumption twice is the same as using it once; rejecting it blocks the paradox
Conditional proof	A rule saying that if you can prove B from assuming A, you can assert “if A then B”; rejecting it blocks the paradox
Curry-complete theory	A theory that contains Curry sentences (so it faces the paradox) but claims not to be trivial
Curry-incomplete theory	A theory that avoids the paradox by not allowing the sentences that cause it

Appendix: Key People

Haskell Curry (1900–1982): An American logician who first identified this paradox in formal systems during the 1940s; he was interested in how systems of logic can have surprising consequences.
Peter Geach (1916–2013): A British philosopher who emphasized that Curry’s paradox doesn’t use negation, making it a harder problem for people who tried to solve paradoxes by changing how negation works.
Arthur Prior (1914–1969): A New Zealand-born logician who showed that Russell’s paradox and Curry’s paradox share the same underlying structure.

Appendix: Things to Think About

The sentence K says “If K is true, then time is infinite.” But couldn’t we just say K is false and be done with it? Why doesn’t that work? (Try assuming K is false and see what happens.)
If you had to give up one logical rule—modus ponens, conditional proof, or the idea that sentences can talk about their own truth—which would you choose? What would you lose by giving it up?
The argument seems to prove that time is infinite. But obviously it can’t really do that. So where exactly does the argument go wrong? Try writing it out step by step and see if you can spot the moment it stops being convincing.
Suppose someone says: “We just shouldn’t allow self-referential sentences.” Is that a satisfying solution? What if a sentence doesn’t directly refer to itself but refers to something that refers back to it? How do you draw the line?

Appendix: Where This Shows Up

Computer science: Programming languages that allow self-reference (like when a function calls itself) need careful rules to avoid infinite loops or logical contradictions. Curry’s paradox is related to problems that arise in these systems.
Mathematics: The foundations of set theory—the basic assumptions about what sets exist—were shaped partly by the need to avoid Curry-like paradoxes. Different set theories make different choices about this.
Language and meaning: Advertisers, politicians, and anyone who says “What I’m telling you is the truth” is using a self-referential claim. Curry’s paradox shows how tricky these can be.
Law: Legal documents sometimes contain self-referential clauses (like “this contract is valid only if it is valid”). Philosophers of law have debated whether these clauses create similar paradoxes.