On the traditional view, a word's meaning is the concept in the speaker's mind: what I mean by 'water' is my mental image of water, and that image fixes what I refer to. Hilary Putnam unsettled this in 1975 with a simple fiction.

Twin Earth is a planet identical to ours in every detail but one: the clear, drinkable liquid found in its lakes is not H₂O but another compound, XYZ. In 1750, before chemical composition was known, a speaker on Earth and their twin on Twin Earth are in exactly the same mental state when they say 'water'. Yet the Earth speaker's 'water' refers to H₂O and the twin's to XYZ. Reference therefore depends on something beyond the mental state — the actual structure of the substance in the world. Putnam's conclusion is famous: meanings just ain't in the head.

His second claim is the division of linguistic labour: most of us cannot tell an elm from a beech, yet we use the word 'elm' correctly, because the community contains experts who can make the distinction and reference leans on their knowledge. Meaning is a social achievement, not an individual one.

The two theses founded externalism in philosophy of mind and, with Kripke, the 'new theory of reference'. They are used in debates running from cognitive science to artificial intelligence: what connection to the world must a system have for its words to be meaningful? Putnam himself changed position several times in later years, moving away from scientific realism — a frequently cited example that changing one's mind in philosophy is not a failing but the consequence of following the argument.