Triangulating Autohomosexuality

Autogynephilia in natal males and autoandrophilia in natal females can be thought of as a notion of “autoheterosexuality”; a sexual interest in being the opposite sex. They are hypothesized to be variations on heterosexuality, in some sense applied to the self. (Or at least, autogynephilia is; autoandrophilia is a bit weirder.)

It seems like in theory some symmetric notion of autohomosexuality should exist. Not necessarily be common, mind you, but if it’s possible for gynephilic men and androphilic women to invert their sexuality in some sense, then the same should be possible for androphilic men and gynephilic women. That autogynephilia doesn’t go away with MtF transition is further evidence for this hypothesis.

Anecdotally, I’ve heard some cases of this too. I know a trans man who post-transition has sexual interests that are unambiguously like those of AGP cis men, and he says that pre-transition, he had similar interests, e.g. fantasizing about being the women he found attractive.

One characteristic of autoheterosexuality is that it often gets reified into arousal to the idea of being the opposite sex. It’s unclear to me whether autohomosexuals would reify their sexuality the same way, thinking of it as arousal to the idea of being their current sex, because this might merely be a side-effect of not being the desired sex. It’s also not fully clear to me what autohomosexuals would be into. For instance, it’s often proposed that they might masturbate to themselves in a mirror, but to me it also seems plausible that they might instead more be into taking on the appearance of other people of their sex that they find attractive (at least, I’ve had more success with that reification in some initial surveys on the topic).

Based on what we know or suspect about autogynephilia, I would propose the following characteristics for determining the validity of a putative autohomosexuality measure:

  • Since autoheterosexuality is more common in heterosexuals than homosexuals (at least in natal males… autoandrophilia is weird), we should expect autohomosexuality to be associated with typical homosexuality (though it would not necessarily be common among gay people).
  • If autogynephilia and autoandrophilia are hypothesized to be the same sorts of variance applied to different underlying orientations, then we should expect them to be strongly correlated in bisexual women. This expectation might also be applied to bisexual men, but on the other hand, it is often not clear how bisexual the bisexual men actually are, so it shouldn’t necessarily be expected there.
  • Similarly, autogynephilia is hypothesized to “run in families”; if this is the case, then we might expect the bisexual sisters of AGP men to also be AGP.
  • Autohomosexuality should have some elements in common with autoheterosexuality that makes it recognizably similar; but I wouldn’t require exact identity, and it might look like a bit of a stretch when it has actually been fully discovered. However, at the very least I would expect autohomosexuals and autoheterosexuals to “recognize” their own sexuality in each other when having them described, at least moderately better than chance.
  • There’s no clear reason why we would expect autohomosexuality to be more common than autoheterosexuality, so we probably shouldn’t expect this.
  • If the measure works for trans women too (and some AGP autohomosexuality measures might not), we should expect trans women to score highly. Similarly, if it applies to cis men, we should expect it to be in agreement with other autogynephilia measures.

One thing that can be noted is that these validation rules are based on the assumption that autogynephilia is a variation of attraction to women. Some instead argue that it is a form of feminized sexuality. I find this argument somewhat questionable, as I would think that it is androphilia that is feminized sexuality. I’ve seen some quite conflicting anecdotes and arguments that focus on the form female sexuality takes, and I don’t think there’s any clear conclusion here, so it’s difficult to just rely on experience. Instead, one way I would test this would be by looking at whether they are correlated with other forms of feminized behavior. IME this holds for attraction to men but not for autogynephilia, so therefore I wouldn’t think of autogynephilia as being a feminized sexuality.

Using these guidelines, we can try to evaluate a number of proposed autohomosexuality measures. They generally focus on autogynephilia in women, rather than autohomosexuality in general, as very few people seem to pay attention to autoheterosexuality.

1. Moser’s approach

In his paper Autogynephilia in women, Charles Moser took the Cross-Gender Fetishism scale and translated it to better apply to cis women. He then did a survey where he tested how many women answered it affirmatively.

Very little additional data was collected for Moser’s scale. A lot of his sample was heterosexual, and he got a significant amount of affirmative answers, so this could be interpreted as evidence that his scale doesn’t correlate with sexual orientation, or perhaps even correlates with heterosexual orientation. However, it’s not very clear, and more research would be needed to say for sure. Similarly, the other validity checks are also impossible to evaluate currently. As a result, Moser’s scale can only really be interpreted as a suggestion, rather than a validated approach.

Perhaps the most notable thing that is missing is a comparison to trans women’s responses. The first of his items might not generalize well to trans women, but the rest should probably work relatively OK.

2. Lawrence’s approach

Anne Lawrence criticized Moser’s approach for not using scales that reified the “attraction to being a woman” enough, and instead suggested a different set of items that strongly reify this concept.

Nobody has collected data with Lawrence’s items, but I once collected data with a similar approach. It did not pass many of the validity checks that I was able to run.


Picture: comparison of the results from trans women and cis women on a scale much like Lawrence’s.

In particular, it did not correlate by sexual orientation; it and AAP were uncorrelated, perhaps with a negative trend, in bisexual cis women; and trans women had the same distribution as cis women. The main validity check that it passed was that in men, it matched the results from another autogynephilia measure.

I also tested a similar measure in men. Here, it seemed to partially pass the sexual orientation criterion. However, among bisexual men, it was negatively correlated with autogynephilia, and for some reason autoandrophilia was more common than autogynephilia regardless of sexual orientation. In women, this autoandrophilia also matched the more-standard ones I use well. On the other hand, trans men didn’t score super high, so it’s hard to say what to make of that. (It may be that autoandrophilia is not a primary cause of gender dysphoria on reddit, but instead that something else, e.g. masculinity, is. If that is the case, we should be able to identify this by finding that autoandrophilia in trans men is negatively correlated with this “something else”.)


Picture: different degrees of autogynephilia and autoandrophilia in different groups of men.

3. Veale’s approach

In her Master’s thesis, Jaimie Veale changed Blanchard’s Core Autogynephilia Scale to be more relevant for cis women by asking whether they had ever been sexually aroused by imagining having “more attractive” physically female features. Her thesis is much more extensive than Moser’s paper, so this time we can evaluate some new things.

On page 66, she has a correlation table which found that her measure of autogynephilia was associated with attraction to men rather than with attraction to women. In general the associations were weak, and so it’s hard to say anything for sure, but it makes me question the validity of her scale.

A lot of the other validity checks were not examined, and so I don’t know whether they held, but that’s not surprising considering they’re relatively obscure. It might be useful to research this in future studies, though.

Veale found that trans women scored slightly but not much higher than cis women on her scale.

4. The Self-Attraction approach

Some people feel that attraction to oneself would be the way autohomosexuality works. On the one hand, I can sorta understand that, and I could totally see the counterfactual female!me be attracted to herself. In fact, there’s quite a few anecdotes of things that seem like autogynephilia and include a heavy element of self-attraction. But I’m not sure this is how it would work, and part of it is the evidence I got when I tried to test it.


Pictured: the average self-reported degree of self-attraction in my Survey on Sexuality, Masculinity and Femininity.

While there does seem to be an effect, where queer people report greater levels of self-attraction than straight people, the effect size is modest. As such, it’s not a very convincing case of passing the sexual orientation test.

I do not have the data to evaluate this approach on many things other than the sexual orientation test yet.

In another survey I tried a variant, asking about arousal by own body and sexual experiences (such as masturbation sessions) focused on admiring one’s own body. This yielded some more-promising results:


Pictured: results from Survey on Personal Sexual Arousal.

This suggests that perhaps the most-effective way to ask about this would be the third approach, asking about whether people have sexual experiences where they focus on their own body.

This general approach strikes me as similar to autosexuality, so perhaps this is something that needs to be researched.

5. The Mimicry-A*P approach

When talking with a trans man I know who is AGP, he suggested focusing on fantasizing about being other women, rather than on sexual interest in oneself. Before he transitioned, he had found it arousing to imagine having a body like the women he was attracted to.

This leads to the concept I call “mimicry-autohomosexuality”. Here, I ask something like “Picture a handsome man/beautiful woman. How arousing would you find it to imagine being her?”. This approach has seemed to pass quite a few tests.


Pictured: mimicry-A*P results from my Survey on Gender, Sexuality and Other things.

The expected ordering with sexual orientation is there; queer > straight. We also see trans women score higher than other groups, though this doesn’t apply to trans men for some reason. Among trans men, there was no statistically significant correlation between masculinity and autoandrophilia (r~-0.266, p~0.1). Among bi women, there was a strong correlation between mimicry-AGP and standard ways of asking about AAP (r~0.32; p~0.0003), and also between mimicry-AGP and mimicry-AAP (r~0.45). The mirrored correlations didn’t exist in bi men (r~-0.1 and r~0.23).

Mimicry-autoheterosexuality had adequate agreement with my standard way of measuring autoheterosexuality in men (r~0.6) and in women (r~0.63). However, I got much higher rates of affirmative answers for mimicry-autoheterosexuality in men (66% vs 45%) and slightly higher in women (54% vs 43%).

The range on this measure seems limited; for instance, trans women seem to tend to hit the ceiling, and so their degree of mimicry-AGP is likely underestimated:


Pictured: response distribution for various groups.

One potential issue with mimicry-A*P is that people seem systematically more likely to endorse the variant that matches their gender than the variant that doesn’t. For instance, gay men were 1.4x more likely to endorse mimicry-AAP than straight men were to endorse mimicry-AGP, and straight men were 1.4x more likely to endorse mimicry-AAP than gay men were to endorse mimicry-AGP.


Pictured: amount of mimicry-A*P endorsed by various groups in my fourth porn survey.

There may also be other potential flaws; e.g. in women, I found mimicry-AGP to be correlated with narcissism (r~0.24, p~5E-4), even though I found no such connection in men. Mimicry-AGP also seemed correlated with femininity in women (r~0.13, p~0.03), despite no such connection in men or trans women. My conclusion from this is that most likely, mimicry-AGP picks up on additional things beyond just autohomosexuality.


Perhaps it might be relevant to create an overview of the different approaches:

Criterion 1 2 3 4 5
Queer higher than het ? No No Yes Yes
Correlated with AAP in bi women ? No ? ? Yes
Runs in same families as autohet ? ? ? ? ?
Surface-level similarity to autohet Yes Yes Yes Yes Yes
Same prevalence as autohet ? No No Maybe Kinda
High scores for trans women Likely Kinda Yes N/A Yes
Concordance with other autohet measures N/A N/A N/A N/A Kinda

Overall, I believe that the most promising approach is mimicry-AGP, but there may be value in considering other approaches and expanding the scales. In particular, it seems that it would be very valuable to discover that multiple different approaches agree with each other, as that could be useful for studying this in a more stable manner.