What Is a Category Mistake? Asking a Question That Cannot Have an Answer

The philosopher Gilbert Ryle introduced the term in 1949 with an example that has been repeated ever since. A visitor is shown around a university, sees the colleges, the libraries, the laboratories, the playing fields and the administrative offices, and then asks where the university is. The visitor has mistakenly assumed that the university is another institution of the same kind as the colleges and libraries, to be found alongside them, when it is the way in which those things are organised. The question is not difficult to answer, it is confused, and the correct response is to explain the error rather than to search for a missing building. Ryle offered further examples, including a spectator at a parade asking when the division will appear after watching the battalions, batteries and squadrons pass, and the point in each is that words can belong to different logical categories while looking grammatically parallel, so a perfectly well-formed sentence can express an incoherent thought.

The concept was not introduced for its own sake but as a weapon against a particular philosophical position:

• •The Cartesian view that mind and body are two distinct substances, one physical and one not, interacting somehow
• •Ryle's claim that this treats the mind as a thing of the same logical type as the body, which is exactly the visitor's error transferred to a person
• •His alternative that mental terms describe dispositions to behave rather than inner episodes in a private theatre, so being intelligent is a way of doing things rather than a hidden process accompanying them
• •The phrase he coined for the position he rejected, describing it as the ghost in the machine, which passed into general use largely detached from his argument
• •The behaviourist reading of his view, which he denied and which critics pressed anyway, and which has been the main line of objection since
• •The lasting influence of the diagnosis regardless of whether his positive account succeeded, since the idea that some philosophical problems are conceptual confusions rather than open questions became widespread

The practical test is whether the predicate can apply to the subject at all, rather than whether it happens to be true. Numbers do not have weight, so asking whether seven is heavy is not false but nonsensical. Legal persons and natural persons share a word and not every property, which is why arguments about corporate responsibility frequently founder. A common source of trouble is the part and the whole, where properties of components are wrongly attributed to the assembly or the reverse, and the fallacies of composition and division are close relatives of the category mistake. Another is the treatment of processes as objects, so asking where the mind is located, where a habit is stored or what the economy wants can generate confusion of exactly this kind. The diagnosis is contestable and frequently contested, since claiming that an opponent has made a category mistake is a strong move that dismisses their question rather than answering it, and it can be used to avoid a genuine problem by declaring it meaningless.

Several live debates turn partly on whether one is being made. Asking whether a computer can think is treated by some as a straightforward empirical question and by others as a category mistake resting on a misuse of a word whose home is in describing whole living creatures. Questions about whether a brain scan shows a decision or merely correlates with one raise similar issues about attributing to parts what belongs to persons, and several philosophers have argued at length that neuroscience routinely commits this error in its language. Debates about consciousness frequently divide between those who see a hard problem requiring explanation and those who see a conceptual confusion generating a pseudo-problem. Asking what caused the universe, if time began with it, may be malformed in the same way. In each case the disagreement is not about facts but about whether the question is well posed, which is uncomfortable because there is no agreed procedure for settling that, and one side's dissolution of a problem is the other side's evasion of it.

In 1957 the linguist Noam Chomsky offered a sentence that has become famous: colourless green ideas sleep furiously. Every word is ordinary and the grammar is perfect, yet the sentence says nothing, because ideas cannot be green, green things cannot be colourless and sleeping cannot be done furiously. Chomsky's point was that grammar and meaning are separate systems, and the sentence is a string of category mistakes packaged in a well-formed shell. It shows why grammatical correctness gives no guarantee that a question or claim makes sense.

Philosophers still disagree about what a category mistake actually is. Some hold that a sentence such as seven is heavy is meaningless, neither true nor false. Others argue that it is simply false, since seven is not heavy. There is also the question of poetry. When a writer says that time is a thief or that a voice is velvet, the category crossing is deliberate, and the reader understands it as metaphor rather than confusion. The difference seems to lie in intention and uptake: a metaphor invites you to notice a resemblance, while a category mistake takes the crossing literally and expects an answer.

A close cousin of Ryle's idea appeared earlier in logic. Around 1901 Bertrand Russell found a paradox in the foundations of mathematics: consider the collection of all collections that are not members of themselves, and ask whether it is a member of itself. Either answer contradicts itself. Russell's remedy, his theory of types, worked out over the following years, sorted every object into a level and required that statements about a collection concern things of a lower level. Under this scheme the troublesome question is not false but cannot even be properly asked, which is the logician's version of a category mistake.

Programming languages build the same safeguard into everyday tools. A type system records what kind of thing each value is, a number, a piece of text, a date, and refuses operations that make no sense for it. Ask many languages to add a number to a word and they stop with a type error, some before the program even runs, much as a careful reader halts at what colour Thursday is. Logicians such as Alonzo Church developed typed systems in the twentieth century, and their ideas passed into the design of languages, so a code editor underlining a mismatch is a mechanical cousin of the check Ryle urged on philosophers.

A question can be malformed rather than hard, treating something as the kind of thing it is not, which is why asking where the university is after seeing every college gets an explanation rather than an answer. Ryle introduced it in 1949 to attack the idea of mind as a separate substance. Claiming an opponent has made one is a strong move, since it dismisses the question rather than answering it, and can be used to dodge real problems.