Steelmanning Gender

Steelmanning: the practice of articulating an opponent's argument in its strongest, most persuasive form, to ensure a fair and thorough understanding before responding.

As I’ve argued in previous blogs, I think the term ‘gender’ has questionable definitional utility. Recently, a chap named Gerardo on an Evolutionary Psychology forum I occasionally visit made a fair attempt to defend the concept of gender. I came away just as convinced in my own arguments, but Gerardo made such a reasonable attempt to ‘steelman’ gender that I thought the conversation worth sharing here.  

Gerardo
Your article argues that we do not need the term “gender” for two reasons: (1) sex is sufficient to define males, females, and the intersex minority, and (2) everything else falls under the category of maleness/femaleness. I agree with the first claim, but the second one merely shifts the problems to the category “maleness/femaleness,” and it is not clear how gender-related terms could be translated into that new terminology. For example, does "maleness/femaleness" include the bodily and psychological variations among males, females, and the intersex minority, and also the norms and categories that each society selectively imposes on males, females, and the intersex minority? In what way can each gender-related term be reformulated as referring to “maleness/femaleness”? Without answers to these issues, it remains only a promise of clarification, not a concrete alternative to current terminology.

Then your article responds to six objections: (1) discrimination against people with gender dysphoria, (2) denial of the right to have a gender identity, (3) forcing a binary categorization, (4) denial of happiness/satisfaction of transgender people, (5) discrimination against transgender people, (6) denial of the freedom to have a gender identity.

Your article responds to objections 1 and 5 by saying that the interlocutor begs the question by assuming gender, but the replies also beg the question by assuming the denial of gender. I think there are reasonably clear definitions of gender dysphoria and transgender, and raising the conceptual bar to an unreachable level is not a reasonable move.

Your article responds to objections 2 and 6 by using a strawman (the “40 genders”) and an alarmist description of problems supposedly caused by transgender people (all of which can be solved without gender eliminativism). I think this doesn’t qualify as a solid response to the objection.

Your article responds to objection 3 by appealing to the sufficiency of maleness/femaleness, but as I said above, this does not offer clear answers to the relevant questions. If we accept (as your article states) that ‘in some traits, females can appear more male than males, and males can appear more female than females,’ it is understandable that some gender-atypical individuals may naturally prefer to follow the social norms of the other sex or receive the social treatment typically afforded to that sex.

Your article responds to objection 4 by claiming that ‘testimony is not reliable’ and that ‘people who left Christianity shouldn’t be trusted when they say they are happier.’ However, it is false that all testimony is untrustworthy, and selectively distrusting Christian apostates reflects Christian in-group bias rather than the actual reliability of the testimony provided by these apostates.

James
First, I’d like to commend your reply - it’s rare that someone offers such a thoughtful reply of disagreement to one of my posts, and I appreciate your effort. I will ty to summarise what I think are your issues with my article, to confirm if I understand your position aright. Your issues fundamentally boil down to: 

1} You think I am overcomplicating the question, as in your view gender is already well defined in psychology and sociology.

2} You contend that gender is understood as the social and cultural dimensions of sex (roles, expectations, norms.

3} You believe that gender identity is a person’s internal sense of belonging to one of those gender categories.

4} You insist that people clearly experience gender dysphoria and benefit from transition, and therefore that this makes the concepts practically meaningful.

5} And if I could be so bold, you’d go so far as to say that the fact that societies treat people differently based on sex shows that “gender” refers to a real social phenomenon, and that what you call ‘eliminativism’ about gender is misguided, since the term captures socially real and psychologically useful aspects of human life.

If I have done your reply justice, it still fails to convince for several reasons. The first is you are doing the very thing I say causes some of the problems to begin with - you are conflating descriptive usefulness with ontological existence, when I am separating the two by saying that the descriptive usefulness is misjudged because the category does not track anything real. As I still maintain, personality traits, cultural roles, preferences, behaviours, aesthetics, etc, can be described directly, without creating an abstract umbrella term that falsely suggests these traits cluster in a way that forms a definable category. The term “gender” is simply an unnecessary and misleading abstraction, where ‘maleness’ and femaleness’ as I described them simply points to characteristics found within the two sexes.

Moreover, your suggested use of clinical definitions does not answer the question of what gender is, or how it escapes the circularity problem - that definitions of gender identity boil down to “Gender identity = one’s internal sense of gender.” My ‘fairydust’ analogy demonstrated that adding an abstract umbrella term is pointless if the underlying traits can be described individually. You ignored this, and did not justify (to my satisfaction, though barely at all) why the abstraction is needed rather than the traits themselves.

You then conflated DSM/clinical definitions with metaphysical definitions. “Gender dysphoria” describes distress to do with identity - not the nature of gender. I argued that there is no objective method that determines gender identity - none at all. I’d also say my “hundreds of genders” concern is not a strawman - it is a direct consequence of recklessly defining gender as “internal identity” - but it’s hard to say I’m strawmanning when I think even the single use of gender is extraneous.

Gender lacks a coherent ontology - and to function as a real category, it would require clear definitional boundaries, measurable criteria, externally verifiable properties, and non-circular definitions - and it has none of those. And everything gender is supposed to explain can already be described via better categories, like sex, temperament, personality traits, preferences, socialisation, societal norms, hormones, sexuality, aesthetic choices, psychological distress and cultural narratives - the term “gender” adds no new explanatory power. 

Gerardo
I do not think that the concept of gender is always well defined in psychology or sociology; I think there are multiple definitions, some defensible and others not. I also do not think that all definitions refer to the sociocultural dimensions of sex, nor do I accept a rigid nature–culture dichotomy. I do not think that gender identity is limited to an internal feeling, and I do not consider transition to be universally beneficial: it can help in some cases and not in others. My rejection of eliminativism is not based on claiming that “gender is real,” but on the fact that eliminativist arguments are methodologically deficient. I do not confuse descriptive usefulness with ontology: I consider that what exists are human beings with their properties and processes, and that gender-related terms are useful for describing some of their regularities. You have not shown that it is possible to describe the same regularities using an eliminativist terminology. To have a serious discussion, you need to respond to the arguments I’ve presented and avoid attributing to me a position that I do not hold.

James
I tried to carefully consider your arguments and respond to each of them. I do not believe I have done your reply a disservice. Perhaps the best recourse, then, is this. You tell me the condition and context under which you believe the term "gender" is most aptly used and most definitionally valid - just give me your single best argument - and I will respond to that.

Gerardo
If you want to respond to my arguments, here’s a list of 8 objections, so you can address them one by one: (1) my objection that you don't explain how “maleness/femaleness” can replace every gender-related term; (2) my objection that you beg the question in the debate about discrimination by assuming the denial of gender; (3) my objection that there are reasonably clear definitions and it's unreasonable to raise the conceptual bar to an unreachable level; (4) my objection that you created a strawman about the “40 genders”; (5) my objection that you made an alarmist description of problems supposedly caused by transgender people; (6) my objection that all the problems you mentioned can be solved without gender eliminativism; (7) my objection that you adopt the false assumption that testimonies of trans people and apostates of Christianity are untrustworthy; ( 8 ) my objection that you adopt a Christian in-group bias when you selectively distrust the testimonies of trans people and apostates of Christianity.

James
You made a good start with your first reply, but since you've been disappointing in not absorbing my replies. I believe gender is an extraneous construct and that we don't need the word at all. So it's up to you to state why you think we need it. And therefore I repeat my request: You tell me the condition and context under which you believe the term "gender" is most aptly used and most definitionally valid - just give me your single best argument - and I will respond to that.

Gerardo
I expected you to address my eight arguments in detail and to avoid speculating about my position, but that didn’t happen. I’ll answer your last question as soon as I have time, but please note that dialogue must be reciprocal, and that I have as much right to expect answers to my arguments as you have to expect answers to your questions.

James
I've simplified it so we don't talk past each other. And by offering to address what you think is the single, strongest argument for the existence of gender, I'm showing willingness to address your best.

Gerardo
I understand that your intention was to simplify so that we wouldn’t talk past each other, but I think your choice was not optimal, because by attributing to me a position I do not hold, your reply forced me to clarify that I do not think that. Furthermore, by not responding in detail to my eight arguments, your reply did not allow me to understand how you would address each point. In my initial message, I first tried to faithfully represent your position and then respond to it. I think it would be good if you tried to do the same and respond in detail to my eight arguments.

You asked me which elucidation of the term “gender” I consider most appropriate and most definitionally sound. I hold that the concept of gender is a hyperonym that groups together a set of related hyponyms. Both the hyperonym and the hyponyms it comprises can be improved, but this implies a reformist project rather than an eliminativist one. The need for the hyperonym is based on its use in fruitful and naturalistic research programs (please note that I am not saying that every use of a gender-related term is of this type; I am strongly critical of postmodern and anti-biological views about gender).

The hyperonym could be defined as follows:

G(I)=(S(I),T(I,S),B(G,T(I,S)),B(I,B(G,T(I,S)))), which reads:

The gender of the individual I is an ordered collection of interrelated elements, and its components are:

• S(I): the individual’s membership in a population defined by sex, either male, female, or the intersex minority.

• T(I,S): the typicality or atypicality of the individual's bodily and behavioral traits relative to the sexual population to which the individual belongs.

• B(G,T(I,S)): the group’s behaviors toward the individual’s typical or atypical bodily and behavioral traits.

• B(I,B(G,T(I,S))): the individual’s behaviors in response to the group’s behaviors toward the individual's typical or atypical bodily and behavioral traits.

This approach rejects the nature-nurture dichotomy found in postmodern approaches to gender, and is compatible with naturalistic approaches as proposed by the biological and social sciences.

James
Thank you for elaborating – but your definition still does not show that these need an umbrella term at all. I remain convinced that the hyperonym is redundant. You can describe all the parts adequately without the umbrella. Moreover, your model still cannot answer the key definitional challenge: What is gender?

You say “Gender is sex + trait typicality + group response + individual response.”

But that is not a natural kind or a unified construct. It is a suite of unrelated variables stapled together.

I appreciate you’re making a better attempt than most, but it still leaves the term ‘gender’ redundant, in my view.

You say "My point is simply that current eliminativist proposals do not offer a better alternative." But my assertion is we do not need a better alternative to a term that was, in my estimation, constructed erroneously and superfluously. I believe you're trying to find justification for a term that humanity just does not need.

Gerardo
Your arguments don’t seem convincing to me.

Your first argument says that the hyperonym is redundant and that we can describe the parts without the hyperonym. But this is a problematic reductionist assumption, because the description of isolated parts ignores their interactions, and the interactive phenomenon deserves a name. It would be equivalent to saying that society is nothing more than a sum of individuals, and that an individual is nothing more than a sum of atoms. We need terms that name systems and patterns of interaction, and this is an example of a term with that function. It would be absurd to claim that science should eliminate all names that refer to such kinds of patterns.

Your second argument says that it is not a unified construct but “a suite of unrelated variables stapled together.” This is false, because each variable depends on the previous on: the second cannot exist without the first, the third cannot exist without the first two, and the fourth cannot exist without the first three. Therefore, it is false that they are “unrelated.” Defining the fourth requires all variables: an individual’s response to the group response to the typicality of the individual’s traits relative to the distribution of the trait in the sex to which the individual belongs. This is the opposite of an example of “a suite of unrelated variables stapled together.”

Your third argument is that it is not a “natural kind.” But this is problematic and depends on how you define “natural kind” and which category you contrast it with. And if you exclude this from “natural kinds” because it involves social variables, you are adopting a nature–nurture dichotomy that is objectionable, because society and culture are also natural phenomena.

Your fourth argument says that, in your view, the term is erroneous, superfluous, and unnecessary, and therefore does not require a better alternative. But this is more a case of begging the question than a solid argument, and it does not address the problems I mentioned in the three previous objections: the need for names that refer to interaction patterns, the falsity of the claim that the variables are “unrelated,” and the unjustified assumption that it is not a “natural kind,” presumably because it involves social variables.

James
I’m feeling mixed emotions here – a combination of very mild frustration that we can’t seem to converge on what is the fundamental issue with gender, and positive regard for your efforts that you are trying your best to give it a good whack. Usually at this stage I start to suspect that the interlocuter doesn’t really wish to believe anything different, but I don’t sense you’re guilty of this. But to get to this point and be where we are is problematic, because I maintain that you do keep talking past the central issue.

Your analogy to “society” or “systems” doesn’t work here, because you still have not shown that this particular grouping of variables constitutes a coherent category rather than an arbitrary bundle. The fact that variables interact does not automatically generate a new ontological category. Lots of things interact. Not all of them deserve new names.

My remaining issues are: 

1. You still haven’t shown that these need an umbrella term at all. Interaction alone does not justify a hyperonym. You need a principled reason for grouping these variables together rather than hundreds of other interacting social and psychological variables. So far you haven’t provided one that satisfies, in my view.

2. You assert dependency, but dependency ≠ unity. Variables can depend on each other and still fail to form a natural kind. Your model is still – sex, trait typicality, group response, individual response – and those are ontologically different kinds of things - biological, statistical, sociological, psychological. Interaction does not magically turn them into a single construct. This is why I said it is a “suite of unrelated variables stapled together.” – yes we can try to relate things tangentially, but in the case of an invented word like gender, interaction does not erase heterogeneity.

3. The natural-kind objection stands. You keep reframing my point as though I am denying that social phenomena are natural. That is not what I said. The question is whether your specific cluster meets the criteria of a natural kind - cohesive, explanatorily unified, and projectible. Your model doesn’t. It is heterogeneous, overinclusive, and its boundaries are arbitrary. Calling something involving social variables “natural” is not the same as demonstrating that it is a natural kind.

4. You still haven’t answered the core definitional question. I ask: What is gender? Your definition is: Gender = sex + trait typicality + group response + individual response. But nothing about that list is principled or unified. It is a verbal wrapper around multiple independent domains of explanation. You haven’t shown why this combination deserves a name like gender, why the boundaries are drawn here, or what explanatory work “gender” performs that cannot be performed more clearly by discussing each part directly as I did in my original article.

Until you do that, I maintain that your hyperonym remains unnecessary and conceptually inadequate.

Gerardo
You say that you appreciate my effort, but you think that I don’t address the central issue, and that you are frustrated because we do not converge in our conclusions. I think I do address each point you raise, and that the lack of convergence does not stem from a lack of goodwill, but from assumptions and criteria we don’t share—and that we should make explicit in order to avoid relying on them as question-begging premises.

You say that I haven’t shown that the grouping forms a coherent category, or that the variables and their interaction justify a hyperonym. I think I have justified the notion according to the standards used in the social, cognitive, and behavioral sciences. What I don’t know is what criterion you use to decide whether a grouping is coherent and whether a pattern deserves a name. That evaluation depends on the criteria we adopt. I don’t consider "gender" to be as firmly established as an atom, nor as dispensable as the ether. I place it alongside notions such as depressive disorders, intelligence, stress, institutions, racism, identity, poverty, social roles, or mental health: complex concepts that bring together different mechanisms and are useful and necessary for fruitful research programs. We can debate these notions, but I believe that efforts to discard them are often driven more by ideology than by sound methodological reasoning. The relevant question is whether the notion allows us to study and explain phenomena that would otherwise remain scattered and difficult to analyze or account for. And in the case of “gender,” I believe it fulfills that function more effectively than eliminativists are willing to acknowledge.

You say that my model brings together biological, statistical, sociological, and psychological elements without becoming a single construct. But my model is not the mere addition of those elements—as your formula “sex + trait typicality + group response + individual response” suggests—but rather the characterization of a pattern of interactions among them. It concerns the interaction between individual and group responses in shaping how typical and atypical traits are negotiated within each sex population, and this phenomenon has enough relevance, interest, and distinctive interaction patterns to warrant both a name and its own field of study—just as other psychosocial phenomena that meet similar criteria do. The relevant point is whether those interactions produce a stable pattern that is useful for explanation and prediction, and that does occur at a level comparable to concepts such as intelligence or depressive disorders. If your criterion is that patterns of interaction never deserve a name, I don’t see how that criterion could be defended, nor do I see it applied consistently in science.

You say that my model does not constitute a natural kind because it is heterogeneous. But if we apply that standard, we should also have to discard many other psychological and social notions that are useful and necessary for current, fruitful research programs. That criterion would be methodologically harmful for the social, cognitive, and behavioral sciences, and that is a strong reason to reject it. Concepts that group patterns of interaction allow us to identify stable regularities and accumulate knowledge. So here, your objection depends on an assumption that I don’t share.

You say that my definition of “gender” is a verbal wrapper with no explanatory unity, adding nothing beyond what could be studied separately. But this does not address my central point: when the components are analyzed in isolation, the study of their interactions is lost. And there is solid evidence of identifiable interaction patterns with specific causes and specific consequences. If you accept that these interactions should be analyzed, then the disagreement becomes terminological—whether we call them “gender” or something else. I have already said that I am willing to adopt alternative terminology, provided it is rigorous and offers advantages. In that sense, the notion is not an empty wrapper but a tool for studying a set of interaction patterns that explain important phenomena—just as we do with concepts like depressive disorders, intelligence, stress, institutions, identity, poverty, social roles, or mental health.

James
I still haven’t been convinced that there is justification for the word ‘gender’ as a concept with any definitional utility. It offers far more confusion and invitation for abuse of language than it does utility.

Biological, statistical, sociological, and psychological phenomena do not magically form a unified construct called ‘gender’ simply because they interact. I maintain your model is still a list of interdependent variables, not a natural kind. Because “gender” is unnecessary, conceptually redundant, and adds ambiguity rather than clarity, it has become a repository for imaginary constructs, false additional attributions - and therefore, even if you could come up with a satisfactory justification for its inclusion (which to me you haven’t) - it still would do much more harm than it would do good.

Note: Conversation ended at this point