Recipe
Disprove a universal claim with a counterexample
Use when: you suspect \forall x\, P(x) is false and need to produce one object that breaks it
Goal: one verified counterexample: an object the claim is actually about, where the claim's conclusion fails
Taught in Counterexamples and disproof
Quick recipe
-
1
Break the claim into its domain (what kind of object x ranges over) and its condition (what must hold for that object). A counterexample must fit the domain, then fail the condition.
Example
For "every prime number is odd," the domain is prime numbers, and the condition to fail is "is odd." A counterexample must actually be prime, then turn out even.
Why: misreading the domain or the condition is the single most common way a proposed counterexample turns out invalid.
-
2
Try the smallest, largest, or otherwise unusual values in the domain before working through it systematically.
Example
Testing primes in increasing order: 2, the smallest prime, is worth checking first.
Why: edge cases are disproportionately likely to break a claim that was only checked against typical values.
-
3
Check both halves separately: confirm the candidate genuinely belongs to the domain the claim is about, then confirm it fails what the claim asserts.
Example
2 is prime (fits the domain) and 2 is even, not odd (fails the conclusion), so 2 is a genuine counterexample.
- the candidate fits the domain but fails the conclusion — it is a genuine counterexample.
- the candidate fails to fit the domain itself — it says nothing about the claim; discard it and keep searching.
Why: a candidate that never fit the domain in the first place was never a real test of the claim at all.
-
4
Report the specific object found, not just the fact that one exists. No general argument is needed beyond this one checked case.
Example
n=2 disproves "every prime number is odd," since 2 is prime and 2 is not odd.
Why: a disproof by counterexample is only as convincing as the one concrete object it names, so the object itself must be stated plainly.
Common mistakes
- Proposing an object that fails the domain itself, like using n=4 against a claim about primes.
- Stopping the search after checking only a few values with no failure, then declaring the claim true.
- Giving a general argument for why the claim seems false instead of naming one concrete object.
- Testing values in a fixed order without ever trying the small or extreme cases most likely to break the claim.