⁂ Concepts · Logic · No. 73
Category

The census form: a classification that participates in constituting what it counts.

Category

A fundamental class used to organise thought or reality - and the question of whether the divisions are found or imposed.

Theme
Logic, language and interpretation
Tradition
Cross-tradition
Term
κατηγορία · katēgoria
Term fixed
c. 330 BC

In plain terms

Every classification has to start somewhere. Underneath the particular kinds - dogs, chairs, Tuesdays - there are broader ones, and underneath those, a few that everything falls under: things, properties, relations, quantities, times.

Those are categories, and the question about them is whether they carve nature or organise us. Aristotle thought the first: the highest kinds are features of what there is, and language reflects them. Kant thought something closer to the second: the categories are the forms our understanding imposes, which is why experience is lawlike.

The practical version of the question is easier to see and harder to answer. When a state decides which categories it counts people in, it is not merely recording. The categories shape what can be noticed, funded and argued about - and the groups they name have a way of becoming real by being counted.

As a philosopher would put it

Aristotle's ten categories - substance, quantity, quality, relation, place, time, position, state, action, passion - are the highest predicates, exhaustive by his account and derived from the ways something can be said of a subject.

Kant's transformation is decisive. His twelve categories are pure concepts of the understanding, derived from the forms of judgement, and they apply to appearances rather than to things in themselves. Categories stop being an inventory of being and become the conditions under which anything can be experienced as an object at all.

The Vaiśeṣika padārthas are an independent parallel - substance, quality, action, universal, particularity, inherence and later absence - reached by systematic argument rather than borrowed. Ryle's category mistake supplies the twentieth-century use: an error consisting not in a false claim but in placing something in the wrong logical type, which is his charge against Descartes. Hacking's looping effects add the social dimension: categories applied to people change the people classified, which makes human kinds unstable in a way natural kinds are not.

Where you meet it

The England and Wales census has asked about ethnicity since 1991, and the categories have changed with every round. A mixed or multiple category was added in 2001; Gypsy or Irish Traveller in 2011; Roma in 2021.

Each change altered what could be seen. Before 2011 there were no national figures for Gypsy and Traveller populations, so their outcomes in health, housing and education could not be compared with anyone's - not because the data was bad but because the category did not exist. After 2011 the disparities were measurable, and became arguable.

The categories also do not simply record. People sort themselves into the boxes offered, the boxes shape how they describe themselves elsewhere, and organisations adopt the classification for their own monitoring, so the census scheme propagates. That is Hacking's looping effect at national scale: a classification designed to count a population participates in constituting it, and the officials revising the form are making decisions about what will be visible for a decade.

What is actually disputed

Whether categories are discovered or imposed.

The realist position is that some divisions carve at joints - that the difference between substances and properties, or between kinds and individuals, is in the world and any adequate scheme will find it. The Kantian position makes categories conditions of experience, so the question of whether they match things in themselves cannot arise. The conventionalist position treats them as instruments, assessable by usefulness and not by correspondence.

Human kinds make the dispute practical. If classifying people changes them, then the categories used by states, clinics and researchers are partly constitutive of what they measure, and the usual criteria for a good classification - stability, exhaustiveness, mutual exclusivity - conflict with accuracy about a moving target. Nobody has a general account of how to classify well under those conditions.

Commonly confused with

  • Concept 1781 A concept is any general unit of thought. A category is meant to be one of the few most fundamental ones, and the whole interest lies in the claim to fundamentality.
  • category in mathematics Category theory is an unrelated formal subject about objects and morphisms, and the collision of names is entirely accidental.
  • categorising in the pejorative sense Putting people in boxes is a complaint about a practice. Categories in this sense include substance and relation, which nobody objects to.

Defined against

  • Particulars A category is what individuals are sorted into, and the question of whether the sorting is real is the universals dispute in another vocabulary

Philosophers

Works

Schools

Related concepts

Concept 1781 · Definition c. 330 BC · Universals c. 380 BC · Form c. 380 BC · Substance c. 330 BC

Aristotle's ten categories are meant as the highest kinds of what can be said of a thing; Kant's twelve are forms of judgement rather than kinds of being, which is a fundamental shift in what a category is.