Two Ontologies of Gender

Why the debate is stuck

Watch Richard Dawkins and Helen Joyce reason about biological sex¹ and you see something close to surgical. The chromosomes, the gametes, the developmental cascade — the categories are clean, the evidence is real, and the case that sex is binary and mind-independent is made without a wasted word. Then watch the same clarity fail to land. Their opponents are not moved, not because the biology is wrong but because the argument keeps sliding off, as if aimed at a target standing somewhere the biologists cannot see. The two camps have been talking past each other for a generation now, and the usual explanation — one side is lying, or stupid, or evil — explains nothing, because it is offered symmetrically by both. Something structural is jamming the exchange.

The jam has a precise location, and I want to isolate it before anyone takes a side, because until it is isolated the sides cannot even state their disagreement. Consider the sentence gender is performative. When a gender-critical writer such as Joyce uses it, she means it as a debunking: gender is a performance in the sense that a séance is a performance, a display staged to pass off something unreal as real. When Judith Butler wrote it in Gender Trouble, she meant nearly the opposite: gender is performative in the sense that a promise or a christening is performative — an utterance that brings into being the very thing it names. The words are identical. The metaphysics is inverted. Two people can affirm the same sentence and be asserting contradictory claims about what kind of thing gender is, and neither notices, because each hears the sentence in his own ontology and assumes the other means what he would mean.

Two Senses of Performance

The word doing the damage is performance, and it has two senses that the debate never separates.

In the first sense, performance is imitation. To perform is to enact something you are not — the actor performs Lear without being a king, the impressionist performs a voice that is not his. On this reading, a male person who identifies as a woman is acting: the underlying fact is fixed by biology, and the gestures, the name, the pronoun are a costume laid over a body that the costume does not change. Performance here is a falsehood enacted through language. The truth is prior and physical; the performance can only conceal it or fail to.

In the second sense, performance is creation. To perform is to bring a thing into existence by doing it — the umpire performs an out by calling it, the sovereign performs a law by signing it, and there is no out and no law antecedent to the performance for the call to correspond to. On this reading gender is not a mask over a pre-existing fact; it is a social status that exists only in and through the repeated acts that constitute it. To perform womanhood is to occupy the category of woman within a field of expectations, recognitions, and responses that the performance itself sustains. The performance does not hide the truth. It makes the truth.

Same verb; opposite causal direction. One camp reads the arrow as pointing from a fixed reality to a possibly-false description of it. The other reads the arrow as pointing from an act to the reality it institutes. Every subsequent disagreement inherits this inversion, and none of it can be resolved downstream, because the two are not disagreeing about the facts within a shared ontology — they are disagreeing about which ontology the word gender belongs to.

Two Realities Colliding

This is why the exchange never converges. Each side accuses the other of denying reality, and — this is the part that makes the debate genuinely hard rather than merely heated — each is right within its own ontology. Laid side by side, the two frames are internally coherent and mutually unintelligible.

Read down the left column and you have a consistent position: there is a physical fact of the matter, our words either match it or they do not, and gender-talk that departs from the biology is fiction dressed as description. Read down the right column and you have an equally consistent position: our very sorting of bodies into “male” and “female” is a classificatory act, gender is a social kind that exists because we enact it, and the standard by which to judge a social kind is not whether it matches biology but whether it is in fact enacted and recognized. The two columns are not two answers to one question. They are answers to two different questions wearing the same words.

The decisive line is the bottom one. The camps do not merely disagree about the evidence; they disagree about what counts as evidence — about the truth condition that a claim like “she is a woman” is supposed to satisfy. One camp holds language to correspondence: a term is used correctly when it matches a mind-independent fact. The other holds it to enactment: a term is used correctly when it successfully does the social work the community recognizes it as doing. A dispute about evidence can be settled by looking. A dispute about what would even count as looking cannot.

Descriptive Categories and Constitutive Ones

Strip the heat away and the disagreement is an instance of something this volume has been building toward: the difference between two kinds of category. Some categories are descriptive — they carve up a world that is already there, and they answer to correspondence. Electron, kidney, mammal, and sex are of this kind; they are attempts to name joints the world has independently of us, and a descriptive category is corrupted precisely when it stops sorting the world the same way twice. Other categories are constitutive — they do not name a pre-existing joint, they create the thing they name by being collectively enacted. Money, citizen, touchdown, checkmate, married: none of these exists in the physics; each exists because a community of agents behaves as though it does, and would cease to exist the moment the behavior stopped.

Constitutive categories are exactly what earlier chapters called patterns: configurations of shared information, stable enough to coordinate behavior, that become real by structuring the expectations of the agents who carry them. A cultural schema that enough people run is not a fiction merely because you cannot find it in a body. It is a social fact — as real as a border, and as invisible to a microscope. To ask whether money is “really” real, or whether a border “really” exists, is to make a category error: you are demanding that a constitutive category satisfy the truth condition of a descriptive one. It will always fail that test, and its failing it proves nothing, because it was never answering to correspondence in the first place.

Volume 2 supplies the general principle underneath this. Truth is not a single relation; a statement is true or false only relative to the truth conditions its terms invoke, and those conditions are conditional on the frame in which the terms live. “There are two sexes” and “gender is real because enacted” can both be true because they are checked against different conditions — one against gametes and chromosomes, the other against recognition and uptake. Neither truth refutes the other, any more than “the piece is a bishop” refutes “the object is carved boxwood.” They are true at different layers.

The Mirror Illusion

Now the tragedy, and it is a genuine one. Because the public debate collapses these two layers into one, each side hears the other making the error it most despises. When the realist hears gender is performative, he cannot hear the constitutive sense; he hears the imitative sense, and so he hears a license for delusion — permission to call a falsehood true by fiat. When the constructivist hears sex is real, she cannot hear it as a modest claim about gametes; she hears it as the opening move of a politics that has always used “nature” to deny people standing — permission to dress oppression as biology. Neither translation is faithful. Each is the most damning thing the sentence could mean read in the wrong ontology.

The result is a mirror. Each camp looks at the other and sees, with real conviction, the very sin it condemns: the substitution of ideology for reality. The realist sees the constructivist bending the world to fit a doctrine. The constructivist sees the realist bending a doctrine to fit as the world. And because each is measuring against its own truth condition, each is not wrong to see it — a constitutive claim does look like ideology-over-reality if you demand correspondence of it, and a descriptive claim does look like reality-over-people if you have decided that all classification is politics. The confusion is ontological, not moral. But it is experienced as moral, on both sides, which is why the argument runs so hot and settles so little. Each mistakes the other’s level of reality for bad faith.

Both camps, seen clearly, are defending something legitimate. The biologist is defending the integrity of empirical categories — the discipline that lets a word keep meaning the same thing across cases, without which no science is possible. The theorist is defending the autonomy of social categories — the recognition that human beings genuinely constitute large parts of their world through language, and that a demand to reduce every such category to physics would erase money, marriage, and nations along with gender. Neither defense is contemptible. The failure is that each absolutizes its own layer and denies the other one exists. The realist talks as though the only real things are the physical ones; the constructivist talks as though even the physical ones are really social. Each is guarding a real border while denying there is any territory on the far side of it.

The Corrective

The stalemate dissolves the moment the two layers are named as two. The corrective is one sentence:

Sex is empirically real; gender is socially real. They occupy different ontological layers and answer to different truth conditions.

Sex is a descriptive category. It carves a real biological joint — bimodal, developmentally deep, consequential — and claims about it answer to correspondence. No amount of enactment changes a gamete type, and pretending otherwise is a category error in the constructivist’s direction: forcing a descriptive claim to satisfy a constitutive truth condition, demanding that biology bend to recognition. Gender is a constitutive category. It is a social status enacted into being, and claims about it answer to uptake, not to chromosomes. Insisting that gender is “just biology” is the mirror-image error: forcing a constitutive claim to satisfy a descriptive truth condition, demanding that a social kind justify itself at the bench of the laboratory. Both errors are the same error — a collapse of two layers into one — committed from opposite sides.

This does not make the hard questions easy, and I want to be exact about what the frame does and does not deliver. It does not tell you which layer governs in a given room — the sports league, the prison ward, the clinic, the dating pool, the census — because those are questions about which real category should control a particular decision, and they cannot be answered by ontology alone. What the frame delivers is the precondition for arguing about them without talking past each other: an agreement that both categories are real, on their own terms, so that the fight becomes a tractable dispute over jurisdiction rather than a shouting match in which each side denies the other’s reality exists at all. That is the difference between a war and a negotiation.

So this chapter takes no side in the object-level dispute, and that restraint is deliberate. Its work is to show why the war persists — not because one camp loves truth and the other hates it, but because the word gender straddles descriptive and constitutive uses, and until that is made explicit every argument circles the same invisible axis. The parity case that follows asks what occupational distributions can and cannot establish. This chapter only builds the frame, and the frame is a proposal rather than an exhaustive map: sex and gender can be real in different senses of real, without that distinction deciding every boundary case or policy jurisdiction.