Simple, right? It's true by necessity. It can't be otherwise, at least when given the assumptions that are needed for ordinary arithmetic.
Welcome now to the lumber yard. Cut two feet from this log. Cut two feet from that one. Put them together. It makes four, yes?
Not. Quite.
Immanuel Kant referred to busted assumptions like this one as instances of the "transcendental illusion": A logically necessary truth seems to imply facts about the real world, but the real world is stubbornly messy and even illogical: 2+2 does not equal 3.97, but in the lumber yard, it absolutely does.
The problem gets still more serious in metaphysics: Buddhism teaches that suffering is the result of clinging emotionally to the transient phenomena of the world; when these phenomena change or disappear, the one who clings is fated to suffer. A Buddha, however, would be perfectly free from clinging in this way--and therefore perfectly free from suffering.
How can one know this to be the case, not just in a given moment, but prospectively? "If I am perfectly free from improper emotional attachment, then I will be perfectly free from suffering" may be necessarily true, but it may suddenly cease to apply to me, much as the formula 2+2=4 ceases to apply to the imperfectly cut logs in the lumber yard. The unflagging serenity of the Buddha turns out to be just another transcendental illusion.
Obligatory libertarian content: Axiomatic accounts of human economic behaviour, such as praxeology, are identically situated, and they face identical epistemological limits.
(For both these systems of thought, there may or may not be useful ways around this difficultly.)
