Why Gamete Production Does Not Define Sex

Or: What We Can’t Induce From How We Reproduce

Kim Hipwell
22 min readJul 27, 2023


Sex As Gamete Production

The idea that gamete production clearly defines two sexes is one of the rhetorical lynchpins of transphobia.

The basic idea is simple. The two types of gamete (sperm, egg) are taken to map to two sexes: the type of gamete that a person produces then determines which of those two sexes they belong to.

This notion is useful to transphobes because from this premise they can construct an argument that trans existence is nonsensical (“if you can’t change what kind of gametes you produce, then you can’t change sex”), and from there it’s a very short step indeed to crafting prohibitions on being trans (“science says trans people don’t exist, so we’re gonna make it so that you actually can’t change legal sex”).

This is the kind of reasoning directly being used in support of the current wave of legislative assaults on trans people.

Sex And Society

There is, however, a fundamental flaw with the theory of sex-as-gamete-production (SAG), which is this: the processes that we actually use to classify people into sexes aren’t predicated on gamete production, legislation notwithstanding.

  • When legal sex is formally declared at birth — which is of course prior to any gamete production — it is done so primarily on the basis of genital anatomy.¹
  • There are many people who don’t produce gametes — for a variety of reasons — and they aren’t regarded as being “sexless”.
  • When legal sex is changed later in life, it is clearly done without belief that “transitioning” leads to a change in the type of gamete produced.

So, sex-as-gamete-production (SAG) is completely adrift from social reality. Everyday notions of what defines a person’s sex are much broader.

SAG is used to focus us away from these everyday notions and onto one specific aspect of sex — “reproductive sex” — as a tactic of rhetoric used in support of repression. Primarily employed as a justification for making trans lives unlivable, it secondarily operates to reinforce patriarchal gender beliefs, enforcing presumed “reproductive roles” on people.²

Lost In The Woulds

Do Or Did Or Will Or Would Is Not Actual Science

The existence of people who don’t produce gametes is obviously a severe embarrassment for the principle of SAG.

To mitigate this “problem”, the theory of ubiquitous gamete production is elaborated with an epicycle of virtual entities: gametes that are not actually produced but (somehow) could potentially be or “ought” to be — what we might call “subjunctive gametes”.

Here is the general approach as described by Heather E Heying:

Heying characterizes this as “actual science”, but there’s nothing scientific about the all-encompassing hedging she’s using here. Rather than shore her definition up, it instead raises obvious questions, such as: how does one determine if an individual would or will produce eggs in the absence of them so doing?³

Possible Worlds And Modal Logic

To reason about counterfactuality, we can turn to modal logic, an extension of classical logic that invokes the notion of a set of possible worlds, in which things are different from our own. This gives us a formal system that allows for discussing notions of necessity and possibility. (Reasoning about past and future states through temporal logic uses fundamentally the same machinery).

An immediate problem with Heying’s definition when thinking about “will or would, but for” in terms of modal logic is that there are, intuitively, possible worlds in which we produce either gamete. Straightforwardly so, because the processes that brings us into being are subject to chance fluctuations: factors at play include the randomness of genetic crossover, the arbitrariness of which of the available gametes combine, and the intervention of environmental factors causing developmental variations (such as mutation or mosaicism).⁴

Such happenstance means that, logically, every individual “would, but for developmental … anomalies” be both male and female according to Heying’s formulation; and that’s not quite what she had in mind.⁵

Image of the Rebis, licensed CC BY-NC 4.0, from University of Manchester Digital Collections https://www.digitalcollections.manchester.ac.uk/view/MS-GERMAN-00001/15

An Accessibility Relation Based On The Point Of Conception

To “whittle down” the set of possible worlds to deliver one that meets Heying’s intuition, we would need to define an accessibility relation that connects only possible worlds in which an individual can (possibly) produce only one gamete type.

A seemingly reasonable way to approach this would be to define accessibility as being limited to the set of worlds in which an individual has the same genome, taking the existence of an individual to begin at the point of conception, and so considering only possible life histories post-conception.⁶

But this can’t work for Heying’s purposes either.

Some individuals who don’t produce gametes simply won’t produce them in any such accessible possible world: no matter how many times you rewind a life history to an earlier point in life and “replay the tape”, you can only average out chance developmental fluctuations.

Photo by Gritt Zheng on Unsplash

If an individual has the same genome in each possible world, then the same “genetic anomalies” (taken as any characteristic of the genome that prevents gamete production) will naturally be present in every world in the set. That would mean some people simply don’t have a sex by Heying’s own definition; again, not the conclusion she had in mind.⁷

Genomes allow for various outcomes depending on environmental conditions — but they cannot be considered to universally be containers of an inherent “gamete producing” purpose or program that is contingently thwarted.

An Accessibility Relation Based On Counterpart Similarity

We might abandon the idea that we are talking about transworld individuals with a specific life history, and instead focus on an accessibility relation that picks out only worlds that contain counterparts of the real-world individual that are sufficiently similar to them.

But there’s a very big problem with this, a problem that “subjunctive gametes” are supposed to distract from: bodies don’t parse out into a straightforward binary sex classification. How then could appropriate counterparts be picked out by an accessibility relation?

Consider this definition of “female” in the Heying mould:

Note that it’s not the simple presence of ovaries that is deemed important, rather the idea that those ovaries “would” produce gametes “if they did”.

This may seem rather an odd tack away from “simple biology“ into the metaphysical, but it stems from the fact that the anatomical characteristics associated with sex don’t work directly as a basis for defining binary sex because they vary beyond a binary pattern.⁸

For a variety of reasons, many women don’t have ovaries. To account for that, we would need a similar notion of who “would” have ovaries “if they did” — and if we try and pursue that idea we soon discover that there is no clearcut answer to that question.

Consider, for example, Androgen Insensitivity Syndrome (AIS) which takes various forms, because insensitivity to androgenic hormones can range on a continuum from mild to complete. That means that, as per the article linked below:

The phenotypic spectrum of AIS … encompasses individuals with a completely female phenotype to male phenotype with infertility/undervirilization

More generally, intersex traits create a “phenotypic spectrum” that includes not just people with ovaries or testes, but also people with gonadal dysgenesis who may have “undifferentiated” gonads, in forms such as ovotestes or streak gonads. Likewise, the form of external genitalia may be “ambiguous”:

The existence of such a continuity of morphology ranging from “typical male” to “typical female” means that any definition of sex based on physical characteristics immediately opens up the “risk” of having to concede that there isn’t a clear cut sex binary after all (and the consequent notion that there may be paths that allow moving between sexes).⁹

Simply put, there is no principled way to universally pick out “similar” counterparts of every individual that consist only of a single “gamete production” type. There are many situations in which any adjudication of counterpart similarity is liable to challenge — not least by the individual affected by that adjudication.

This is a consequence of the real-world complexity that SAG and “subjunctive” gametes are trying to distract from. A clear-cut sex binary is not a physical reality, so the anti-trans movement argues for its social agenda to be built on the basis of an invocation of virtuality that is not tied to any material state; an intangible confection that does not necessarily relate to detectable physical characteristics.

Heying’s statement is an expression of her feelings about what “ought” to be the case (but stubbornly is not). That’s not “actual science” — just vibes.

Creationist Thinking About Gametes

Logical Breakdowns

The conviction that gamete production “ought” to be somehow inherent to the human condition — immanent even if not present — is defended in various ways wherever the idea of “subjunctive” gametes is challenged.

For example, SAG is often defended by comparing bodies to manufactured objects such as cars or toasters, where cessation of function does not imply loss of purpose:

So bodies are here compared to machines, as undergoing “failures in production,” or as being “organised to” or “set up to” produce gametes.

But bodies aren’t like machines. The teleological argument that relies on this false analogy has long been discredited.¹⁰

There is nothing that bodies “ought” to do in the absence of doing it, because bodily functionality is evolved, not designed. Evolution has no foresight, no goals that it is planning to achieve — and nor do its products. They are not “set up to” do anything.¹¹

The apparent “design” that is embodied in organisms is a mirage: to argue otherwise is to engage in creationism.¹²

Of Gametes And Gas

Objects that are manufactured are designed for specific purposes. Knowing the purpose of a manufactured object helps to explain why it exists and why it is constituted as it is: an explanation in terms of what the object is for.

A toaster is created in order to toast, a blender exists to blend, and so on. Even when these objects can’t fulfill their purpose, knowing what the purpose is gives us adequate justification for saying that we know what that object ought to do. The toaster’s element should heat up, the car’s engine should provide motive power, and so on.

Photo by Daniel Salgado on Unsplash

Conversely, there is no inherent purpose to naturally-occurring objects like an ocean, or sunlight, or radioactive decay. One might talk of the uses that these things are put to — as a means to generate electricity, for example — but those uses do not supply the reason why those objects come into being. They are simply a set of things that an object can do.

Evolution creates naturally-occurring structures that do complex things, not structures that are for a purpose. Eyes, for example, did not evolve in order to contribute to vision — this is not what they are for.

All we can say is that contributing to vision is what eyes do, generically (we will consider genericity shortly).

Photo by David Clode on Unsplash

Likewise, bodies are not for producing gametes; bodies are no more “gamete producing machines” than they are “fart producing machines”, “headache having machines” or “eye rolling machines” — and that’s the case whether the body produces those things or not.

Bodies are not for any of these things; subjunctive gametes no more characterize the body than do subjunctive farts.

Mirages Of Bodily Function

Function In Biology

Teleological thinking about biology is pervasive, and encouraged by the ways in which we describe bodies.

For example, the word “function” is pervasively used in describing what an organ generically does: we say that the function of the heart is to pump blood, that the function of lungs is to oxygenate blood, the function of kidneys is to filter blood, and so on.

This is not mistaken in itself, but unfortunately the word “function” elides between the for/do distinction.

We can say both that a hairdryer has the function of drying hair (that’s what it’s for), and that radioactive decay in a nuclear reactor has the function of generating electricity (that’s what it does).

So while talking about an organ’s “function” is technically correct, it can also lead to some very fundamental conceptual confusions.

By A. Konby (?) — Internet Archive, Public Domain, https://commons.wikimedia.org/w/index.php?curid=1493624

Functions, in the biological sense, are not expressions of purpose; so if a naturally-occuring object doesn’t do something that happens to be a generic function of other objects of its type, then it simply doesn’t have that function. It doesn’t somehow inherently have that function qua “purpose”; there’s nothing it comes into being for, no teleological ghost attached to it.

Generic Functions (Aren’t Universal Functions)

The assumption that function is something present even in its absence really makes no sense, but (you will be unsurprised to learn) that doesn’t prevent it being used in advocating for SAG.

This is done by claiming that objects have an implicit functionality, that stems from the fact that they are included in a class of objects that have that function generically.

Often this is done by drawing an analogy between gamete production and bipedality.

But generic generalizations (aka generics) do not allow us to conclude anything about individuals — they are simply statements that describe salient characteristics of a class of objects. For example:

  • humans are bipedal
  • ducks lay eggs
  • ticks carry Lyme disease

Although we take “humans are bipedal” to be a true statement, it’s only true as a statement about humans as a class; it is not a statement about individual humans, who are not universally bipedal.

This distinction is important to recognize: for example, we (should) take that non-universality carefully into account in the design and operation of accessible spaces and services.

Photo by Yomex Owo on Unsplash

In fact, to be salient, a generic characteristic not only doesn’t have to be universal, it doesn’t even have to necessarily be true for a majority of members of the class. Less than 50% of ducks lay eggs, for example, while less than 1% of ticks carry Lyme disease.

Therefore, the fact that a class of objects generically has a specific characteristic or “function” does not imply that members of the class even generally do, let alone that they implicitly should.

Bullshit About Bipedality

This makes it clear that the generic statement that “humans are anisogamous” doesn’t license us to make attribute an implicit function of gamete production universally to every individual human.

Nevertheless, attempts to defend the universality of SAG are often made by asserting that generic bipedality implies universal bipedality:

Teleological explanations for biological forms are straightforwardly assumed here — we are “supposed” to be bipedal, and nature “isn’t perfect, so disorders and defects happen”. We’ve already seen this kind of teleological reasoning isn’t applicable to evolved organisms.

It’s no less than absurd to say that bipedality is “literally in our DNA” when talking about a situation where bipedality does not develop. As we’ve already discussed, no purpose inheres in DNA and re-running it across possible worlds will not reveal a potentiality that is not present. The genome does not strive towards specific outcomes.

To insist that generic statements indicate an implicit universality in this way is a fundamental error of thought, like assuming that all ducks have or should have an implicit “function” of laying eggs.

A person with one leg does indeed have a “monopedal phenotype”, by definition. That’s because a phenotype is simply the set of observeable physical properties of an organism.

Such a person is not a “biped with a leg missing”, just as ducks that don’t lay eggs are not “ducks with an egg laying function missing”, and just as humans that don’t produce gametes aren’t “gamete producers with a gamete producing function missing”.

Historical Impacts

A variant of this fundamental error is to talk about the historical impact of the fact that “humans are bipedal”, such as claiming that skeletal changes during human evolution means that every part of every human somehow reflects an intrinsic bipedality.

Photo by Eugene Zhyvchik on Unsplash

But this is simply the same fallacy dressed slightly differently. People who are “monopedal” can have skeletons that reflect an evolutionary history of bipedality. That doesn’t imply that those people who don’t walk on two legs are, somehow, bipedal. (No more than the existence of glacial features in a landscape imply the contemporary presence of a glacier).

This is no different to saying bipedality is “literally in our DNA” or the equivalently erroneous statement that our chromosones tell us that “sex is in every cell of our body” — these are all ideas that make the bumbling assumption that genotype is a more accurate representation of phenotype than phenotype itself.

The “Proper” Function Of What?

SAG is occasionally also defened using one other conception of function in biology: the idea of “proper” function as advanced by Ruth Millikan, as part of a wider philosophical approach concerned with mental representation, which is called “biosemantics”.

The notion of “proper” function aims to to be the basis of a principled account of “malfunction” in organisms, and as such aligns with naturalistic accounts of disease: both are grounded in the notion that reproduction itself is the central determinant of what is “normal” and therefore healthy functioning.

Christopher Boorse’s explanation given in this SEP article captures this idea:

when a process or a part (such as an organ) functions in a normal way, it makes a contribution that is statistically typical to the survival and reproduction of the individual whose body contains that process or part

However, “proper” function cannot be used to anchor SAG because it is subject to the same “counterpart” problem that we saw for possible worlds semantics. There aren’t always clear analogues of process or part between bodies, and therefore it’s not always possible to identify which “statistically typical” pattern would be applicable counterfactually.

Gonadal agenesis or dysgenesis, for example, does not allow for the identification of an unimplemented “proper” function of gamete production: an ovotestis cannot be said to clearly have either the “proper” function of an ovary, or that of a testis.

Rebecca Mason discusses this objection to SAG as part of a broader rejection of essentialist definitions of “woman”:

A definition rooted in reproductive function cannot impose “subjunctive” reproductive function by fiat on bodies that don’t conform to what is “statistically typical” for reproductive function.

Just as genericity does not license an assumption of universal applicability across a population, nor does typicality.¹³


The Fatuity Of SAG

The fatuous synecdoche of SAG insists that a specific sense of “sex” adequately represents the entirety of the concept.

As a justification for actions that immiserate trans and intersex lives this is risible. As others have observed, it is senseless to suggest that complex social patterns could be aptly mapped out in terms of gametes.

It seems like some of SAG’s advocates are just suckers for its specious simplicity; others hawk it because it neatly aligns with broader ideologies such as complementarianism or the just-so gender mythologies of evolutionary psychology.

Whatever their motivations, what they are advocating for is a naive essentialism that bootstraps itself into existence by conjuring up phantom essences. SAG is an apparition assembled from a vision of ghostly gonads.

The Precarity Of SAG

The precarity of SAG as a definition is clear from the fact it leans on intuitions about what bodies “ought” to do instead of being based on observable phenomena. As we’ve seen, SAG is:

  • not aligned with the social reality of how we define “sex”
  • reliant on assumptions about possibility (subjunctivity) that are not supported by consideration of counterfactual conditions
  • aligned with invalid “creationist” thinking about intrinsic purpose in biology (teleological arguments)
  • aligned with invalid extrapolations from the generic or typical function of the body or body parts (normative claims)

We’ve also seen that various concessions to diagnosing sex through characteristics other than “gamete production” have to be made to prop up SAG. This hints at the outlines of a “family resemblance” model of sex that cloudily defines it:

  • either by gamete production
  • or by the possession of gonads of a specific type (due to similarity to people who produce gametes)
  • or by other morphological characteristics (due to similarity to people who possess gonads of a specific type)
  • or by some vaguely intuited potential to possess gonads of a specific type (extended SAG here requiring a notion doubly disjoint from empirical reality, categorizing people in terms of a function that “ought” to be implemented by a part of the body that they “ought” to have)

Note that, when expressed in this way, it’s obvious that the connections to “gamete production” are not rooted in “function” at all — merely in patterns such as anatomical similarity; nor does this mandate a conformance to the metaphysical binary that SAG is so desperate to enforce.

What’s more, family resemblance implies extensibility. It’s easy to see that additional links to these chains of similarity could additionally or alternatively encompass other patterns of resemblance that underpin our social attributions of sex; this would give us a definition that is not so precariously rigid and adrift from social reality.

The Ideology Of SAG

Sustaining SAG beliefs are deep-rooted heteronormative assumptions about reproduction as the purpose of life.

But attributing such purpose to human bodies, universally, is hand waving at best and perilously close to fascist thought at worst: this notion is the basis of eugenicist thought, pronatalist positions, and the queerphobic mindset.

Rudolf Hess once stated that “national socialism is nothing but applied biology”, and SAG is certainly “applied biology” in his sense. That is, it is a pseudo-scientific concoction that wraps ideology in the mantle of “biology” for the purpose of inflicting further damage on marginalized groups.

SAG is mythologizing motivated by an agenda of social conformity, nothing more profound.



A practice which, of course, long precedes Karl Ernst von Baer’s discovery of the mammalian ovum in 1827.

Note too, that the initiation of oogenesis prior to birth is irrelevant to this observation, because a) there is no actual check that oogenesis has happened before sex is assigned on the basis of genitalia and b) female sex is still assigned in situations where oogenesis doesn’t occur.


A familiar, deeply embedded anti-feminist “biology is destiny” view of the world. Note that the surgical enforcement of the “sex binary” via non-consensual surgery is specifically enabled through exceptions in the current wave of anti-trans legislation.


There’s a deep circularity here: the premise of SAG is that gamete production determines sex, but in the absence of gamete production we have to determine what gametes would have been produced by reference to the sex of the individual. Uh oh.

So, to prop up SAG we need a notion of how to define sex that is based on characteristics other than gamete production. However, if we had that we would no longer need to refer back to gamete production at all, because our broader conception — whether anatomical or societal or otherwise — would already encompass sex in all cases.

However, this approach doesn’t suit the political purpose that SAG serves; to deliver a facade of theory to justify the binary which is to be enforced.


Views on what characterizes the same individual across different possible worlds vary, but note that there’s no intrinsic reason why transworld identity can’t range as widely as this — it is often taken to.

Some analyses of transworld identity encompass, for example, the idea that an individual might have had a different set of parents, which would provide an even wider basis for genetic variation than we are considering.

The argument for “bare identities” outlined in the SEP article linked above takes identity to be an even broader notion, something not requiring any specific properties that “follow” an individual across the set of possible worlds, just a “haeccity” or “thisness” which serves to mark the identity.


Advances in technology also make it harder to insist that this definition is watertight.

For example, documented here are some mice that are clearly both male and female according to Heying’s definition:

Technological advances also illustrate that sex differentiation possibilities are perhaps much closer together than Heying imagines:


This captures the notion that the individual’s genome (rather than gamete production itself) defines sex. I’ve previously laid out issues with this idea in “If It Ain’t One Thing, It’s Another” (see footnote 5 of that article).

Note too that mutation and mosaicism can affect development after conception. Which means that there is a case to be made that even when replaying the tape only from the point of conception, we might end back up at the idea of individuals plausibly being both male and female.

For example, different people with 45,X/46,XY mosaicism are known to have produced both types of gamete:

The latter paper additionally notes:

Of course, it should be noted that the incidence of normal fertile females who have a 46,XY karyotype is not known because it is not routine to check the karyotype in fertile women


Note that considering individual identity to be tied to a specific genome would make the notion of what exists in a set of “possible worlds” much narrower than what we intuitively accept as within the range of possibility.

For example, using this approach, one could not say there is a possible world in which I naturally have green eyes, because in fact I have blue eyes.


Indeed, the fact that physical characteristics of the body may also be altered by the process of “transitioning” lends validity to the concept of sex as a mutable condition, rather than an abstraction rooted in an unalterable provenance or potentiality.

This would obviously be uncomfortable for transphobes to acknowledge; so they don’t, sometimes at great length.


The metaphor that bodies are machines is pervasive in how we talk about the body in English:

  • The heart is like a pump
  • The lungs are like bellows
  • The eyes are like cameras
  • The brain is like a computer

This encourages us to reason as though parts of the body have purposes like those of machines, and to construct arguments that are intended to justify doing so. This is a pervasive type of thinking that can mislead us into making some very wrong assumptions.


Sometimes, evolution itself is wrongly attributed with having or conferring the purpose of reproduction:

This is a fundamental misunderstanding. Reproduction is simply the mechanism that enables evolution to occur.

Of course reproductive capabilities are selected for in evolution, but reproduction itself is not a goal or purpose of evolution any more than phosphoresence or bipedality are.


Many self-declared skeptics/rationalists are incredibly credulous when it comes to religious fundamentalist/creationist patterns of argument, so long as those arguments are being used to affirm their own preconceptions or prejudices about gender or sexuality.

Skeptical Inquirer recently published a piece co-authored by Jerry Coyne, an arch-critic of intelligent design creationism, in which he quite explicitly puts forward the creationist position that bodies are designed for specific purposes:

A skeptic may have the ability to point out the fundamental errors in reasoning when talking about the specific, narrow domain of evolution denial; but fail spectacularly in other contexts to achieve the same clarity.

Alongside talk of design, that article contains a veritable litany of anti-science talking points taken from the religious right — workshopped rhetoric full of polished misdirection, elision, and carefully masked fallacies.

It might not be intuitive that “skeptics” will uncritically and shamefully defend nonsense such as SAG or even complementarian doctrine; but this will be unsurprising to anyone conscious of the inherent sexism and heteronormativity endemic in — for example — the field of evolutionary psychology.


Note also that naturalistic accounts of disease are subject to objections made on the basis that what we treat as being “typical” is contingent.

It doesn’t seem plausible that a naturalistic approach can define how we draw a non-arbitrary bounding line between a) natural variation and b) forms of “deviation” from a standard of normality that require medical intervention.

Consider that historically we have treated as medical conditions perfectly innocuous states of being such as left-handedness and homosexuality. (The latter was not fully depathologized until as recently as 2013 and is still subject to quack “treatment” in the form of abusive “conversion” practices, increasingly widely outlawed as forms of torture).

Basing an account of “disease” on what is considered to be “normal” is hostage to the assumptions or motives of the parties making the call about normality — there is no objective way to make that call.



Kim Hipwell

PhD in Cognitive Science, interested in the structures of natural and artificial languages. Thrives on atonal music and trans rights. She/her.