Non-circular definitions of gender

Gender as a cluster-concept

A while back I read an excellent piece by

Lane Taylor

on the circularity problem for a particular definition of womanhood. I highly recommend giving that piece a read, but so that we are all caught up, I’ll recapitulate the objection in miniature.

When defining a term, we have two objects of interest: The definiendum and the definiens. The definiendum is the term being defined and the definiens is the definition provided. As an example, say someone wanted to know what the meaning of ‘book’ was, and I provided the answer: ‘a collection of paper enclosed in a binding’. The definiendum would be the term ‘book’ and the definiens would be ‘a collection of paper enclosed in a binding’.

Now, typically, when we construct a definition, the goal is to inform our audience of the meaning of the term defined. In other words: we want an illuminating definition such that, if someone was unfamiliar with the term in question, upon learning the definition they would come to understand what the term means. Given this constraint, a non-circularity condition imposed on our attempts at definitions. In particular, the definiendum cannot appear in the definiens. To see why lets modify my toy example from earlier: A ‘book’ is a ‘book-like object’. Typically, if A is defined as B then A means B, and if A means B we can substitute A for B in sentences in which they occur. So now we get: A ‘book’ is a ‘book-like-like object’. But ‘book’ is still there, so we get: A ‘book’ is a ‘book-like-like-like object’. You see how this goes, we have entered a sequence of substitutions that will never terminate. If someone is looking for an informative definition of a term, this certainly won’t do.

So what does this have to do with gender? There is a definition of woman-hood commonly called the Self-ID view that goes as follows: A woman is anyone who identifies as a woman. So far so good, so what’s our definiendum? ‘Woman’. And our definiens? ‘Anyone who identifies as a woman’. Alright, now lets run our test: ‘Anyone who identifies as a anyone who identifies as a woman’. Oh no. ‘Anyone who identifies as a anyone who identifies as a anyone who identifies as a woman’. Welp, we are in quite a pickle aren’t we? In his article, Lane clearly articulates a couple ‘ways to go’ for someone sympathetic to the Self-ID view in response to this circularity problem, but I want to propose a different tact that transitions nicely into the way I like to think about this topic.

Generally, I don’t think we should view ‘identification’ as a defining trait of gender. Here’s one reason to think this: it seems like people could still be male or female if they were restrained from birth such that they could never utter the phrase ‘I am a man!’ Yet ‘identification’ plays a crucial role in our day to day experience of ascertaining other peoples genders. If John says to me ‘I’m a man’, all else equal I’ll come to believe (on the basis of that statement) that John is a man. So, what gives? I think the best explanation here is that identification is an epistemic criteria for determining gender. In other words, one is not the gender to which they identify in virtue of their identification, but finding out how someone identifies is one of the better ways to find out what someones gender is. In general, I think this is the right way to conceptualize gendered behavior. Not as the thing in virtue of which someone is the gender that they are, but rather as an indicator that we use to infer their gender. So far so good, in fact, nothing I’ve said thus far need be objectionable to the essentialist.

But why are these often good indicators, and why does what we rely on to determine gender seem to shift over time (I.e. which behaviors we appeal too to more strongly infer peoples genders)? I think the answer ties into my preferred view on the matter, so lets outline that briefly then connect it back to our foregoing comments on identification.

My preferred view of gender is as a sort of ‘cluster concept’. Famed substacker and defender of shrimp rights

Bentham's Bulldog

has written a nice piece that encapsulates something quite like my view (see here) but there are some meaningful distinctions. This excellent article by

Defending Feminism

and the subsequent podcast between Bentham and herself were largely the impetus for some recent revisions I made to this view in my own mind, and finally deciding to write it up. With credit paid where it’s due, lets take an example of a gendered concept and work through how to understand it on a cluster concept view. I define woman as ‘anyone with a sufficient quantity and magnitude of intrinsic and extrinsic feminine traits’. There’s a lot here so let’s move from the end of the definition to the start, as I think it will make more sense. I just take ‘feminine traits’ as those traits often associated with those of the female sex (this avoids circularity worries!). Intrinsic and extrinsic here refer to (broadly) mental and physical features respectively. For instance people often talk about a gendered experience as a sort of phenomenological state, the corresponding feminine version would be an intrinsic feminine trait (I’m speaking vaguely here because there may be, say, a ‘non-binary phenomenal state’ and I am attempting to leave room here for my view to schematically accomodate a definition for those folks). Having long hair, or wide hips, might be examples of feminine extrinsic characteristics. What about quantity and magnitude? I think there are a wide range of these traits (intrinsic and extrinsic) and that some of them ‘count more’ in various points in history. If someone were to ask me ‘in what sense is gender a social construct?’ this is where I’d point them on my view. Say we assign a ‘gender point value’ to each trait. The phenomenal state I mentioned is worth 10, having long hair worth 100, having an inclination (mental state) towards having children may be worth 50. Say our threshhold of gender points was 100, then even if someone had both the first and the last of these traits, they wouldn’t have enough ‘points’ to clear the threshold. But our views of womanhood may shift over time and the weighting of these traits can shift in tandem. Continuing with our toy example, the phenomenal state mentioned might become worth 150, having long hair worth 10, and having an inclination towards having children worth 5. Now, you can only be a woman if you have the first trait. As a brief aside, I’m not saying necessarily these traits, or these weightings, are in fact what I or anyone else have in mind when they use the term ‘woman’ (indeed I would be shocked if they were!) This is merely intended as an illustrative exercise of how I understand the trait weighting. This specific feature of my view, and the fact that I explicitly exclude behavioral traits like wearing lipstick, are going to be the major differentiators between my view, and a view like Bentham’s. One final important point to note is that I essentially understand this as a schema you can apply to other gender concepts, though it gets trickier to specify what the traits are supposed to initially cluster around for genders other than male or female (i.e. nonbinary folks), but I’ll leave that worry to the said for the time being, maybe to be revisited on another occasion.

One virtue of this analysis, is that it has a nice explanation of how those in the category of ‘woman’ can drift over time. As society alters the weighting or list of the traits, the people who meet the standard will change. But the meaning of the term in so far as its related to traits associated with adult females doesn’t change. This lets us account for why, say, a peasant in the 1800s might make different classification of trans people then we would: because the traits and their weighting in the peasants social context are different from our own. But we can understand the meaning of the term in the same way (i.e. as a cluster concept).

How does this connect back to the point about epistemology I raised earlier? Well, this view also nicely accommodates why different epistemic tools shift in and out of favor for determining gender. If our weighting is heavily skewed towards intrinsic characteristics, then indicators of those intrinsic characteristics will more consistently give good answers to who is which gender. Someone reporting that they “identify as a man” is reporting something about their intrinsic features. So as we more heavily weight intrinsic vs extrinsic characteristics, different identifiers become more useful. This allows us to ‘subsume’ the Self-ID view quite naturally as self identification acting on an evidentiary basis, not as the ‘essence’ of gender, while still explaining some of its initial plausibility vis a vis its usefulness at identifying internal characteristics.

So, this cluster concept view has a satisfying way of assimilating the Self-ID view, and understanding the interaction between patterns of behavior and gender more specifically, that doesn’t essentialize gender in virtue of behavior which could lead to worries of stereotyping. On a closing note of humility: I’m sure there are some glaring issues with this view, I admit I’m not particularly well read on this area of literature but the posts I’ve mentioned as well as some tertiary reading left me feeling like this was an interesting problem to tinker with, and the result seems interesting enough that I felt it was worth typing up.