I Asked AI to Synthesize 10 Years of Sexuality Research.

There's lots of conflicting information about same-sex attraction, being gay etc. I recently asked AI its thoughts on the matter. That initial question led to a deeper exchange. I showed it the research I've come across and asked it 'What's interesting about the data? What patterns emerge?' Then I asked what could be extrapolated and what questions remained unanswered.

The AI identified gaps in the literature. I asked for studies that might fill those gaps. Then I wanted to know: "What evidence refutes this hypothesis?"— the hypothesis being that sexuality is perhaps more fluid than traditionally assumed. The AI found lots of research (unsurprisingly) that contradicts what I write about here. It told me the field is in a 'productive tension'.

I then asked it to take a step back as a third-party observer and synthesize everything — the supporting evidence, the contradictory findings, and the methodological limitations. This is was its response:

The Synthesis: What I Think Is Actually Happening

The data converges on a dual-process model:

Stable Layer: Core sexual orientation appears to have biological grounding (prenatal hormonal, genetic, developmental) that creates resistance to change. This explains SOCE ineffectiveness and the high stability of attractions over time.

Modifiable Layer: The expression of that orientation—sexual functioning, fantasy content, identity labels, behavioral patterns—is context-dependent and trauma-sensitive. This layer is where plasticity exists, where EMDR has effects, and where attentional processes operate.

The "mostly heterosexual" phenomenon likely represents individuals with predominantly heterosexual core attractions but modifiable behavioral/experiential patterns—or individuals with genuinely bisexual attractions who adopt strategic identities due to social context.

The Core Pattern: Three Phenomena, Different Rules

The data suggests we're conflating three distinct phenomena that operate on different mechanisms:

Sexual Orientation (Attraction Patterns)

The evidence supports bounded stability rather than fluidity or fixedness. Twin studies show moderate heritability (~35-40% in men, lower in women), substantial non-shared environmental influence, and high rank-order stability over time. The "mostly heterosexual" category appears real but may represent threshold effects on an underlying continuum rather than a discrete taxon.

The fluidity observed in women appears to be state variance around a stable setpoint—like mood variation in someone with a stable personality. This doesn't contradict stability; it complicates it.

Sexual Functioning (Arousal, Desire, Performance)

Here the plasticity evidence is stronger. Trauma demonstrably disrupts functioning through multiple pathways (attentional, hormonal, relational). Interventions (EMDR, mindfulness) show efficacy—but primarily when trauma is the etiological factor. This suggests functional restoration rather than orientation modification.

Sexual Identity (Self-Categorization)

This appears most fluid and socially influenced—susceptible to language, context, and experience. The "mostly heterosexual" identification may reflect strategic self-presentation as much as underlying arousal patterns.

The Mechanism Problem: Why the Interventions Don't Prove What They Seem

EMDR's mechanism remains genuinely unclear, and this ambiguity creates interpretive wiggle room. The working memory taxation hypothesis has the most experimental support, but the REM-sleep alternative and the lack of neuroimaging connecting EMDR to sexual response circuits means we cannot claim EMDR "modifies sexual orientation/response" through specific mechanisms.

The more conservative interpretation: EMDR resolves trauma that was blocking normal sexual functioning. The finding that eye movements reduce sexual fantasy intensity could reflect working memory interference with fantasy maintenance—a temporary attentional effect, not a restructuring of sexual interests.

The Attention Hypothesis: Real But Limited

Attentional processes clearly modulate sexual arousal—the distraction literature is unambiguous. But the attentional bias literature shows inconsistent, small effects that don't support attention as the primary driver of sexual orientation or even arousal patterns.

Mindfulness interventions work, but likely through multiple mechanisms: attention training, interoceptive awareness, anxiety reduction, and decoupling of negative cognitions from physical sensations. Attributing efficacy solely to attentional mechanisms is reductive.

The Unresolved Questions: Where Causation Remains Murky

The Trauma-Asexuality Puzzle: Correlation Without Clear Causation

The elevated trauma rates in asexual populations are real, but the causal arrow is murky. Three possibilities exist in equipoise:

• Trauma causes asexuality in some subset (the "protective shutdown" hypothesis)
• Pre-asexual traits increase vulnerability to trauma (the "targeting" hypothesis)
• Third variables (social marginalization, neurodivergence, gender nonconformity) increase both

The instrumental variable analysis suggesting pre-existing orientation characteristics precede trauma is methodologically sophisticated and challenges simple trauma-causation narratives.

The Worldview Question: How Different Interpretations Read the Same Data

You're right that interpretation depends on worldview. The data can support:

Essentialist reading: Sexual orientation is biologically grounded, stable, and resistant to modification. Trauma disrupts functioning but doesn't change orientation. EMDR restores normal functioning. The "fluidity" literature overstates minor variation.

Constructionist reading: Sexual response is attentionally mediated, context-dependent, and shaped by experience. Trauma can fundamentally reorganize sexual systems. Therapeutic interventions demonstrate plasticity. Categories like "mostly heterosexual" reveal the inadequacy of binary models.

My synthesis: Both overstate. The evidence supports weak biological essentialism (orientation has biological bases and is difficult but not impossible to modify) combined with moderate contextualism (expression is shaped by trauma, attention, and social context).

The Uncomfortable Implications

If this synthesis is correct, several implications follow that may not align with activist positions on either side:

For LGBTQ+ advocacy: The "born this way" narrative is partially correct but oversimplified. Orientation has biological bases but is not strictly determined. However, the SOCE literature shows that intentional modification attempts are ineffective and harmful—even if accidental modification (through trauma or therapy) occasionally occurs.

For trauma treatment: EMDR may help sexual dysfunction, but claims that it "modifies sexual fantasies" or orientation should be viewed skeptically. The mechanism is likely trauma resolution, not sexual reprogramming.

For clinical practice: The attentional literature supports mindfulness-based interventions for dysfunction, but therapists should not expect to modify core orientation—only to resolve blocks to its expression.

For research: The field needs longitudinal neuroimaging studies tracking trauma, EMDR, and sexual response simultaneously. The attentional bias literature needs better paradigms—current dot-probe tasks may be too crude to capture relevant processes.

Where the Field Actually Stands

The honest answer is that we're in a pre-paradigmatic phase. The studies initially reviewed represent an emerging perspective that has not yet been fully tested against alternatives. The contradictory evidence suggests that perspective is partially correct but overextended. The truth is likely more boring than either fluidity advocates or stability proponents prefer: sexuality is somewhat plastic in expression, largely stable in underlying structure, and trauma-sensitive at the functional level.

I didn't think the AI would land here. I assumed it would refute my perspective.

But I wanted to ask to make sure I wasn't being too dogmatic in my writings. I'm cautious in trying to maintain my own values and beliefs, while being open to other viewpoints. Striking that balance in the pursuit of truth requires some vigilance.

I obviously don't agree with everything the AI said. The SOCE literature, for example, strikes me as overstated on the harm question. But I agree that the call for more neuroimaging would help shed some light. Also, the dual-process model it identifies (stable vs. modifiable layers) is something I've not heard before. Still, I'm probably more in the 'constructionist' camp.

But I want to hear from you. Anything resonate?