# Epigenetic Homosexuality Model Objection Defeater ## Intro In 2012 three evolutionary biologists (Rice, Friberg, and Gavrilets) published a model with a memorable idea: homosexuality might be caused by "epi-marks," chemical tags that ride on top of DNA and are usually wiped clean between generations. On their picture, an unborn child is tuned to be more or less sensitive to the hormone testosterone, a strong tag escapes the usual erasure, passes from a parent to an opposite-sex child, and nudges just the brain systems that set who a person is attracted to. The result, they suggest, is same-sex attraction. The paper is clever, and it gets cited in arguments about whether people are "born gay." So it is worth being clear about what it does and does not show. It is a *hypothesis*, a proposed mechanism, and by the authors' own admission it comes with no direct evidence. They never point to an actual epi-mark, an actual gene it controls, an actual case of parent-to-child transmission, or an actual brain circuit it changes. The model shows that the idea is *conceivable*. It does not show that it is *true*. The reply below does two things. First, it walks through why the model, as impressive as it is, cannot bear the weight people put on it: it strings together a long line of separately-possible steps and treats the finished chain as if it were established, its confident reading of human epigenetics runs ahead of the evidence, one of its headline predictions was later contradicted, and its equations fit numbers to the theory rather than testing the theory. Second, it notes a twist worth knowing. The paper actually argues *against* a simple "gay gene," so it cannot be waved as proof of "born this way"; and even if every step were confirmed tomorrow, a biological cause still would not tell you whether acting on a desire is right. That last point is the one that never depends on the science at all. A note on tone before the substance: what follows tests a *paper*, not a person. Every human being bears the image of God and is loved by him. The target here is an overreaching scientific claim, not anyone's dignity. ## In full Defeater for the claim: *"Rice, Friberg, and Gavrilets (2012) have shown that homosexuality is produced by inherited epigenetic marks that canalize fetal sensitivity to androgens. So same-sex orientation is a hardwired, heritable biological condition, settled by the developmental biology, and the traditional Christian ethic is simply at war with the science."* The model, stated fairly, is this. Male (XY) and female (XX) embryos differ in how strongly their tissues respond to prenatal androgens. The authors propose that this difference is stabilized ("canalized") by sex-specific epigenetic marks: marks that raise androgen sensitivity in XY development and lower it in XX development. These marks are normally erased and re-made each generation. Occasionally, they argue, a strong mark escapes erasure and is transmitted to an opposite-sex child, where, if it acts on the neural systems governing partner preference but spares the genitals and gender identity, it yields same-sex attraction. A mathematical model then shows that mutations producing such "sexually antagonistic" marks could spread under some parameter values, because the canalization benefits the sex that makes the mark even at a cost to the occasional opposite-sex descendant. The critique is **seven-pronged**, and it grants the model's ingenuity throughout. (1) **No direct evidence.** The paper identifies no specific inherited mark, no gene it regulates, no observed transmission event, no altered fetal brain circuit, and no demonstrated causal link to adult orientation; the authors themselves concede they cannot supply definitive evidence, so the model is at most *conceivable*, not shown to occur. (2) **Mechanistic gap accumulation.** The mechanism requires roughly eight separate "might" steps to all hold at once (a mark exists, it is epigenetic, it is occasionally inherited, it survives two rounds of reprogramming, it acts tissue-specifically, it targets partner-preference neurons, it spares genitals and gender identity, and it overrides the child's own newly-made marks); each step may be individually possible, but the probability of the whole conjunction cannot be read off the plausibility of its parts. (3) **Overstated human transgenerational epigenetics.** Mammalian germ cells and early embryos undergo extensive epigenetic reprogramming; genuine germline transmission in humans has to be distinguished from continued environmental exposure, maternal physiology, DNA variants that themselves influence methylation, and direct exposure of a fetus's own germ cells, and mouse findings establish possibility, not the proposed human orientation mechanism or its rate. (4) **A central prediction failed.** The paper predicted that large genetic studies would find negligible association; the Ganna et al. 2019 GWAS (about 477,000 people) instead found five loci and a broadly polygenic contribution, so DNA variation does contribute detectably, which undercuts the paper's sharper suggestion that familial clustering is carried mainly by epi-marks rather than DNA. (5) **Under-determined inferences.** Low identical-twin concordance does not uniquely point to epigenetics ("nonshared environment" is a statistical residual, not a measured epi-mark), the required tissue-specificity is asserted rather than demonstrated (it functions as a rescue device), and the model collapses a heterogeneous phenotype (attraction, arousal, fantasy, identity, behavior, exclusivity versus bisexuality, and male versus female patterns) into one androgen-sensitivity axis. (6) **Question-begging terminology.** Describing same-sex orientation as a "gonad-trait discordance" and a "feminized male preference" builds the conclusion into the vocabulary, presupposing that opposite-sex attraction is the one biologically concordant endpoint and that same-sex attraction is a partial reversal of sexual differentiation, which is the very thesis to be proven. (7) **Parameter-fitting, not evidence.** The mathematics shows only that *if* such marks had the stipulated properties they could spread; the headline "about 50% non-erasure" figure is back-derived from chosen prevalence and twin-concordance values *on the assumption the model is already correct*, which is fitting parameters inside the hypothesis, not independent confirmation of it. The **logical diagnosis** ties these together. The paper is an *inference to the best explanation* (an abduction): it proposes a mechanism to explain familial clustering plus incomplete twin concordance. That is a legitimate move, but an abduction earns confidence only when it predicts novel observations, outperforms rival explanations, survives attempts to falsify it, and receives direct mechanistic confirmation. This model has met none of those standards, and inferring "therefore inherited epi-marks *cause* homosexuality" from "epi-marks *could* explain the pattern" is a textbook case of affirming the consequent, since many different mechanisms produce the same broad pattern. The **double-edged-sword observation** is the payload for debate. This paper is cited in *both* directions, and neither citation is earned. It cannot be used as "born this way" proof, because it is unconfirmed *and* because its authors expressly argue against a simple genetic cause. It also cannot be used, in the other direction, to reduce orientation to a mere developmental accident, because it is equally unconfirmed. And underneath all of it sits the point that never needs the biology at all: even a fully proven cause, genetic or epigenetic or developmental, would not by itself establish any moral conclusion, because "this is how a desire arises" never entails "acting on it is right" ([Is-Ought Fallacy](/codex/is-ought-fallacy/), [Moral Argument](/codex/moral-arguments/)). The science, however it resolves, leaves the moral question exactly where it was. ## Cheatsheet **The 30-second reply:** > That 2012 paper is a model, not a finding, and the authors say so themselves: no actual epi-mark, no gene, no transmission event, no brain circuit, no measured causal link. It works by chaining eight separate "maybes" into one story and treating the finished chain as proven, but a string of maybes multiplies down to something very unlikely. It leans hard on human epigenetic inheritance that is still mostly demonstrated in mice, not in human sexual orientation. Its big prediction, that genetics would turn up almost nothing, was contradicted in 2019 when the largest gene study found a real polygenic signal. And its math just fits numbers to the theory rather than testing it. So it is a testable idea, which is to its credit, but it is not evidence that anyone is "born gay." Notice too: the paper argues *against* a simple gay gene, so it cannot be your proof of "born this way" either. And even if it were all true tomorrow, a biological cause still would not tell you whether acting on the desire is right. That question never turned on the science. **The 5 fast facts:** 1. **The authors concede there is no direct evidence.** No specific inherited mark, no regulated gene, no observed parent-to-child transmission, no altered fetal brain circuit, no demonstrated causal link to adult orientation. The paper shows the mechanism is conceivable, not that it occurs. 2. **It is a chain of about eight "might" steps.** A mark exists, is epigenetic, is inherited, survives two rounds of reprogramming, acts tissue-specifically, hits partner-preference neurons, spares genitals and gender identity, and overrides the child's own marks. Each step being possible does not make the whole chain probable. 3. **Human transgenerational epigenetic inheritance is overstated.** Germ cells and embryos reprogram extensively; solid cases are mostly in mice, and must be told apart from environmental exposure, maternal effects, and DNA variants that drive methylation. Extrapolating mouse stress studies to human sexual orientation is a very large leap. 4. **A headline prediction failed.** The paper predicted genetics would explain almost nothing. Ganna et al. 2019 (about 477,000 people) found five loci and a broadly polygenic contribution, so DNA does contribute at the population level, contradicting the paper's sharper claim. 5. **The math fits the hypothesis, it does not confirm it.** The equations show marks *could* spread *if* they had the assumed properties. The "50% non-erasure" figure is derived from prevalence and twin numbers by assuming the model is already right. That is parameter-fitting, not evidence. **The 3 strongest counter-moves:** - *"Name one actual epi-mark, gene, or transmission event from the paper."* There is none. Force the admission that the model is conceivable, not demonstrated. This is the whole game. - *"Multiply the maybes."* Walk the eight-step chain out loud and ask for the probability of all of it holding at once. Plausible-per-step does not survive conjunction. - *"The paper argues against a gay gene, so how is it your 'born this way' proof?"* Name the double edge. Then add: even confirmed, a cause does not make an act right. **Concessions to make freely (each collects a larger one):** - Yes, the model is genuinely ingenious and, to its real credit, explicitly falsifiable. Concede it warmly. In exchange the objector must concede that a falsifiable model which has *not yet passed its tests* is a hypothesis, not a result, and cannot be cited as settled science. - Yes, epigenetics really does regulate sexual differentiation, and identical twins really can differ in gene expression. Concede it. In exchange the objector must concede that "epigenetics is real" does not establish *this particular* inherited-mark-causes-orientation pathway, which is the contested claim. - Yes, there is almost certainly *some* biological contribution to orientation, prenatal and developmental. Concede it plainly. In exchange the objector must concede it is a contribution to a complex, heterogeneous trait, not a single decoded mechanism, and certainly not one this paper has demonstrated. - Yes, orientation is generally not a simple choice. Concede it. In exchange the objector must concede that "unchosen" is a separate question from "caused by inherited epi-marks," and a further separate question from "therefore morally permissible." **What NOT to defend:** - Do not argue that epigenetics is fake or that orientation is purely chosen; the science is against you and the case does not need it. Win on what the *paper* has and has not shown. - Do not claim the paper is worthless; it is a serious, testable model. Overstating the critique concedes the high ground. The precise claim is that it is unconfirmed, not that it is stupid. - Do not rest the moral conclusion on the biology turning out one way; the is-ought point holds whichever way the science lands. - Do not treat "it's not genetic" as a win for the traditional ethic; the ethic does not depend on orientation being chosen, and the paper does not deliver "chosen" anyway. **The closing line:** > *"It is a smart, testable idea, and I respect that they made it falsifiable. But a model that names no actual mark, no gene, no transmission, and no circuit, and whose one sharp prediction was later contradicted, is a hypothesis waiting on evidence, not evidence. It can't prove 'born this way,' the authors themselves argue against a gay gene. And if it were confirmed tomorrow, we'd still be exactly where we started on the only question that was ever in dispute: not where the desire comes from, but whether acting on it is right. Biology has never answered that question, about any desire, for anyone."* ## Argument structure | | Premise | Notes | |---|---|---| | **P1** | **The paper supplies no direct evidence for its proposed marks; it establishes conceivability, not occurrence.** It identifies no specific inherited epi-mark, no gene the mark regulates, no observed parent-to-child transmission event, no altered fetal brain circuit, and no demonstrated causal link from any such alteration to adult sexual orientation. The authors concede they cannot provide definitive evidence for a strong epigenetic basis. A model with every mechanistic slot still empty shows at most that the mechanism is *possible*, which is a claim about imagination, not about the world. | Evidential floor | | **P2** | **The mechanism is a conjunction of about eight separately-plausible steps, and joint probability cannot be read off the parts.** The story needs all of the following to hold together: a canalizing mark exists; it is epigenetic; it is occasionally inherited; it survives two rounds of germline and embryonic reprogramming; it acts in a tissue-specific way; it targets the neural systems for partner preference; it spares genital development and gender identity; and it overpowers the child's own freshly-made sex-concordant marks. Grant each step a generous individual probability and the product is still small. Treating the finished chain as supported because each link is imaginable is *mechanistic gap accumulation*: plausibility per step is silently swapped for probability of the whole. | Conjunction fallacy / gap accumulation | | **P3** | **The model overstates human transgenerational epigenetic inheritance.** It treats multi-generational transmission of relevant human marks as well established and occurring at "nontrivial" rates, but mammalian germ cells and early embryos reprogram extensively, and genuine germline transmission must be distinguished from continued environmental exposure, maternal physiology and uterine effects, inherited DNA variants that drive methylation, and direct exposure of a developing fetus's own germ cells. The best-demonstrated cases are largely in mice; extrapolating from rodent stress studies to human sexual orientation, and asserting a "nontrivial" transmission rate, runs well past the evidence. | Empirical overreach | | **P4** | **A central genetic prediction of the paper was later contradicted.** The model predicted that large-scale studies would fail to find genetic markers for most homosexuality and that any hits would explain almost nothing. The Ganna et al. 2019 GWAS (about 477,000 participants) identified five loci associated with same-sex sexual behavior and found a broadly polygenic contribution; individual variants had tiny effects and behavior was not predictable from the genome, but DNA variation nonetheless contributes detectably at the population level. That does not eliminate an epigenetic role, but it undercuts the paper's sharper suggestion that familial clustering should be explained mainly by inherited epi-marks rather than DNA polymorphisms. (Qualifier to state fairly: the GWAS measured same-sex *behavior*, not a single unified measure of enduring orientation, so it is not a complete test of every version of the model.) | Failed prediction | | **P5** | **The supporting inferences are under-determined; each is consistent with the model but does not select it.** (a) Low identical-twin concordance does not uniquely indicate epigenetics; it is equally consistent with polygenic influence at incomplete penetrance, prenatal environmental and placental differences, stochastic neural development, and measurement noise, and "nonshared environment" in a twin model is a *residual* category, not a measured cause. (b) The required tissue-specificity (a mark that hits partner-preference neurons while sparing genitals and gender identity) is asserted, not demonstrated, and functions as a rescue device: no evidence is given that such marks exist in the relevant human fetal neurons, survive into them, regulate preference, and leave neighbors untouched. (c) The model collapses a heterogeneous phenotype (attraction, arousal, fantasy, identity, behavior, exclusivity versus bisexuality, and likely-distinct male and female developmental architectures) into one androgen-sensitivity axis. An explanation compatible with the data but not favored over its rivals has not been confirmed by that data. | Under-determination | | **P6** | **The framing terminology quietly builds in the conclusion.** Calling same-sex orientation a "gonad-trait discordance," and speaking of a "feminized male preference" or a "masculinized female preference," presupposes that opposite-sex attraction is the single biologically concordant developmental endpoint and that same-sex attraction is a partial reversal of sexual differentiation. That is a model of the phenomenon, not a neutral description of it, and it is precisely the thesis the paper is meant to establish. When the vocabulary already encodes the conclusion, agreement with the vocabulary is not evidence for the conclusion. | Question-begging | | **P7** | **The mathematics fits parameters to the hypothesis; it does not test it.** The formal model shows only that *if* sexually antagonistic epi-marks have the stipulated transmission probabilities, fitness benefits, penetrance, tissue-specificity, and independence, then mutations producing them can spread under some parameter values. That is a conditional about a stipulated object, not evidence that the object exists. The headline "about 50% non-erasure" rate is especially telling: it is derived from selected prevalence and twin-concordance figures *under the assumption that the model is already correct*, which is parameter-fitting within the hypothesis. Different prevalence estimates, concordance estimates, or definitions of homosexuality would yield different numbers. Consistency-with is not confirmation-of. | Parameter-fitting, not evidence | | **Surprise** | **The paper's own logic, and its own thesis, block both ways it is cited.** Logically, the model is an inference to the best explanation, and a strong abduction must predict novel observations, beat its rivals, survive falsification attempts, and win direct mechanistic confirmation; this one has done none of these, so inferring "epi-marks cause homosexuality" from "epi-marks could explain the pattern" is affirming the consequent. Rhetorically, the paper is deployed in opposite directions and earns neither: it cannot serve as "born this way" proof, because it is unconfirmed *and* its authors argue against a simple genetic cause; and it cannot serve to reduce orientation to a bare developmental accident, because it is equally unconfirmed. The honest reading discharges both citations at once. | Double-edged-sword / affirming-the-consequent | | **C** | Establishing "homosexuality is caused by inherited epigenetic marks" from this paper requires (a) mistaking a mechanism with every slot empty for a demonstrated cause; (b) reading the plausibility of eight separate steps as the probability of their conjunction; (c) treating still-contested human transgenerational epigenetics as settled; (d) holding a genetic prediction the largest later study contradicted; (e) selecting one under-determined explanation over equally-compatible rivals; (f) accepting terminology that presupposes the conclusion; and (g) mistaking parameter-fitting for evidence. Each step fails. **The model is conceivable, not confirmed; a testable hypothesis, not a result.** It cannot ground "born this way," it cannot reduce orientation to an accident, and, decisively, a biological cause of *any* kind would still leave untouched the only question in dispute, whether acting on the desire is right, which no fact about a desire's origin has ever settled. | | ## Master objections to the whole argument **MO1: "You're just a science denier moving the goalposts. Epigenetics is real, transgenerational effects are documented, and you're dismissing peer-reviewed biology because you don't like the conclusion."** - The critique concedes that epigenetics is real, that it regulates sexual differentiation, and that transgenerational effects are documented in some systems. None of that is in dispute. The dispute is narrow and specific: whether *this paper* has demonstrated that *inherited epi-marks cause human homosexuality*. It has not, and the reason is not ideology but the paper's own contents, it names no mark, gene, transmission event, or circuit, and its authors concede the lack of definitive evidence. Distinguishing "epigenetics exists" from "this exact epigenetic pathway to orientation is established" is not goalpost-moving; it is the ordinary difference between a field and a specific unverified claim within it. Refusing that distinction is what would let any speculative mechanism ride on the credibility of the whole discipline. **MO2: "Requiring direct observation of the mark is an unfair standard. Plenty of good science proposes mechanisms before they're directly observed. This is how hypotheses work."** - Exactly right, and the critique agrees: proposing a mechanism ahead of direct observation is legitimate and often fruitful. The point is not that the paper is illegitimate *as a hypothesis*; it is that a hypothesis in this state is not yet *evidence for its conclusion* and must not be cited as though it were. A proposed-but-unobserved mechanism earns belief in proportion as it makes novel predictions that come true and outperforms its rivals. This one made a sharp genetic prediction that was later contradicted (P4), and its supporting inferences are equally compatible with several rival mechanisms it does not exclude (P5). So the objection actually states the standard correctly and then exempts this paper from it. Hold the paper to the very standard the objection names, and it comes up short, not because unobserved mechanisms are forbidden, but because this one has not yet done what would license belief. **MO3: "Fine, it's not proven. But it's still the best available explanation for familiality plus incomplete twin concordance. Best-explanation reasoning is valid, so you should provisionally accept it."** - Inference to the best explanation is valid, and provisional acceptance of a genuinely best explanation is reasonable. Two problems block the move here. First, "best" is comparative, and the paper has not shown it beats the field of rivals; polygenic influence at incomplete penetrance, prenatal and placental environmental differences, stochastic neural development, and combinations of these account for the same familiality-plus-discordance pattern without any inherited-mark machinery (P5). An explanation that has not been shown to outperform its competitors is not established as the best; it is one candidate among several. Second, even a provisional best explanation of a *pattern* is not a demonstrated *cause*, and the leap from "could best explain the correlation" to "does cause the trait" is affirming the consequent. Provisional acceptance, if warranted at all, would license "worth investigating," not "settled biology, cite in a culture-war argument." **MO4: "This is all a smokescreen. You're attacking the biology so you can keep condemning gay people. If there's any biological cause at all, your ethic collapses."** - The moral conclusion does not depend on the biology, in either direction, which is exactly why the critique keeps them separate. Suppose the model were fully confirmed tomorrow: a biological cause of a desire still would not entail that acting on the desire is permissible, any more than a heritable component of alcoholism makes drunkenness good. That is the is-ought gap ([Is-Ought Fallacy](/codex/is-ought-fallacy/)), and it is handled at length on the companion page ([Born This Way Objection Defeater](/codex/born-this-way-objection-defeater/)). So the ethic does not "collapse" if a biological cause is found, and it does not depend on orientation being chosen. Note also the double bind for the objection: the paper it leans on argues *against* a simple genetic cause, so it cannot simultaneously be the objector's proof of biological hardwiring and a problem for the traditional view. The critique attacks an overreaching scientific claim; the moral question is argued on its own ground, and it was never going to be won or lost in a genetics journal. **MO5: "The GWAS you cite actually supports a biological basis, five loci is not nothing. You're using it against Rice-Friberg-Gavrilets while it cuts your way too."** - The GWAS is cited for a precise and limited purpose, and its limits are stated plainly. It is used only to show that the paper's *sharp prediction* (that genetics would explain essentially nothing and that any hits would be negligible) did not hold: five loci and a broadly polygenic signal did emerge (P4). It is not cited to show that homosexuality is "genetic" in any strong sense, and it does not show that: the same study found the effects tiny, behavior unpredictable from the genome, and it measured same-sex *behavior* rather than a unified measure of orientation. So the GWAS simultaneously (a) contradicts this paper's specific "genetics will find nothing" prediction and (b) refutes any strong "gay gene" reading, which is why the companion page uses it against the strong-genetic-determinism claim as well. Both uses are consistent: the honest reading of the GWAS is a detectable-but-small, non-deterministic polygenic contribution, which is bad news for *both* "it's negligible epi-marks not DNA" and "it's a gay gene." **MO6: "Even granting the critique, isn't this just Christians grasping for any way to say being gay is a choice? The pastoral effect is cruel regardless of the logic."** - The critique does not claim, anywhere, that orientation is a choice; it explicitly concedes that orientation is generally not simply chosen, and it declines the "it's not genetic, so it's chosen" inference as a further non-sequitur (see What NOT to defend). The aim is narrower and, properly used, kinder: to stop a *hypothesis* from being wielded as a settled scientific verdict in an argument about persons' lives, in either direction. The historic Christian position holds three things at once, that every person bears God's image and is owed dignity and compassion, that the moral standard on sexual conduct is universal rather than a singling-out, and that love does not consist in affirming every desire. Testing an overreaching claim is compatible with all three, and it should be done with the tenderness owed to people made in God's image. A true account of what the science does and does not show is not cruelty; misrepresenting a hypothesis as proof, whichever side does it, is what misleads people. ## Premise 2, Mechanistic gap accumulation (the decisive methodological move) ### Affirmative case 1. **The conjunction, laid out.** The proposed causal path is not one claim but a chain, and it fails unless every link holds *together*: (i) a canalizing epigenetic mark on androgen-signaling exists; (ii) the relevant sensitivity difference is epigenetic rather than genetic or hormonal; (iii) the mark is sometimes inherited across generations; (iv) it survives both germline reprogramming and early-embryo reprogramming; (v) it acts tissue-specifically rather than body-wide; (vi) it acts specifically on the neural systems that set partner preference; (vii) it spares genital development and gender identity; and (viii) it overrides the child's own newly-made, sex-concordant marks. 2. **Why the chain is the problem.** Probability of a conjunction is the product of the (conditional) probabilities of its parts. Even if a debater generously grants each step a 60 to 70% chance, eight such steps multiply to roughly a one-in-forty to one-in-sixty outcome for the whole path, and several of these steps (surviving two reprogramming rounds; hitting *only* preference neurons) plausibly deserve far lower individual numbers. The rhetorical error the paper invites is to feel the plausibility of each step in turn and never multiply. 3. **This is a general failure mode.** *Mechanistic gap accumulation* is the pattern of treating a series of "this could happen" links as though their combination were already supported. It appears across speculative biology and is a close cousin of the just-so story: internally coherent, individually plausible at each node, and unconstrained by any measurement that would fix the joint probability. Naming it converts a vague unease ("this feels like a stretch") into a precise objection ("show me the conjunction, not the links"). 4. **What would repair it.** The remedy is not rhetoric but data at the nodes: identify the mark (i), show it is epigenetic (ii), catch a transmission (iii-iv), and demonstrate the tissue- and circuit-specificity (v-vii) in relevant human tissue. Each measured node collapses a "might" to a known value and shortens the chain of speculation. Until then, the honest description is a long conditional, not a mechanism shown to run. ### Anticipated objections 1. *"Biological pathways are always multi-step. By your logic no developmental mechanism could ever be proposed, since they all chain many events."* 2. *"The steps aren't independent, so you can't just multiply. Canalization makes several of them go together."* ### Rebuttals 1. The objection conflates *proposing* a multi-step pathway with *having demonstrated* one. Established developmental pathways are multi-step too, but their steps are individually *measured*, which is exactly what pins the joint account down; the complaint here is not that the pathway has many steps but that many of its steps are *unmeasured assumptions*. A multi-step mechanism with every step observed is science; a multi-step mechanism with the key steps stipulated is a hypothesis. The number of steps is not the issue; the number of *unverified* steps is. 2. Granting non-independence actually helps the critique as much as it hurts. If canalization makes several steps covary, that is itself a substantive empirical claim requiring its own evidence (that the marks bundle in the specified way), so it adds an assumption rather than removing one. And the least-correlated, most-demanding steps, surviving two rounds of reprogramming, and acting on preference neurons while sparing genitals and gender identity, are precisely the ones the model most needs and least supports. Even a charitable correlation among the easy steps leaves the hard, load-bearing steps unmet and unmeasured. ## Premise 7, The mathematics fits the hypothesis, it does not confirm it ### Affirmative case 1. **What the model proves and what it does not.** The formal result is a conditional: *given* marks with the assumed transmission probability, canalization benefit, penetrance, tissue-specificity, and independence, selection can favor and spread the mutation producing them within certain parameter ranges. That establishes internal consistency and evolutionary feasibility. It does not establish that any such mark exists, which is a separate, empirical question the equations cannot answer. 2. **The tell-tale back-derived parameter.** The paper's memorable "about 50% non-erasure" figure is not measured; it is inferred from chosen prevalence and twin-concordance values under the assumption that the model is the correct account of those values. That is circular for the purpose of confirmation: the data are used to *tune* the model, then the tuned model is presented as explaining the data. Tuning is legitimate for exploring a hypothesis; it is not evidence for the hypothesis, because a false model can usually be tuned to fit the same summary statistics. 3. **Sensitivity to definition.** The fitted numbers ride on contested inputs, the prevalence of homosexuality, the concordance rate, and, upstream of both, the *definition* of the phenotype (exclusive versus bisexual, behavior versus attraction versus identity, male versus female). Change the definition or the prevalence estimate and the fitted "non-erasure" rate moves. A quantity that floats with the analyst's definitional choices is a fitted parameter, not a discovered constant of nature. 4. **The standard a confirming model would meet.** A model earns evidential weight when it makes a *novel, risky* prediction, distinct from the data used to build it, that then comes true, ideally a measurement at one of the empty mechanistic nodes (an identified mark, an observed transmission, a mapped circuit). The paper's one sharp, out-of-sample prediction, that genetics would find essentially nothing, went the other way (P4). So on the very yardstick that would have converted the math into evidence, the result to date is negative. ### Anticipated objections 1. *"Fitting a model to data and checking consistency is normal, legitimate science. You're describing ordinary model-building as if it were a fallacy."* 2. *"Feasibility modeling matters: showing the mechanism is evolutionarily *possible* answers the old objection that homosexuality can't persist under selection. That's a real contribution."* ### Rebuttals 1. Model-fitting is indeed normal and legitimate, *as a stage*. The fallacy is not fitting; it is presenting the fitted model as *confirmed* by the same data it was fitted to. Consistency with the training data is the minimum a serious hypothesis must clear, not evidence that distinguishes it from rivals fitted to the same data. The critique targets the citation of a fitted, untested model as though it were an established cause, not the ordinary practice of building and tuning models. 2. Granted, and this is the paper's real and fair contribution: it dissolves the "how could it persist under selection?" puzzle by exhibiting a mechanism under which persistence is possible. But "evolutionarily possible" is a modal claim, and it answers a *modal* objection; it does not upgrade to "actual." Showing that a trait *could* persist by this route leaves fully open whether it *does* persist by this route rather than by any of the other routes that are also evolutionarily possible. Feasibility earns the hypothesis a seat at the table; it does not seat it at the head. ## Christian satisfaction, why the reply is coherent and unthreatened The critique needs no anti-science posture and takes on no burden the biology could later overturn: - **On the science**, it concedes everything actually established, that epigenetics regulates development, that some transgenerational effects are real, that orientation is generally unchosen and probably has biological contributors, and disputes only the specific, unverified claim that this paper demonstrates an inherited-mark cause of homosexuality. If the empty nodes are one day filled and the mechanism is confirmed, nothing in the Christian position moves, because the position never rested on orientation being non-biological. - **On the logic**, the load-bearing move (a cause of a desire does not entail the rightness of acting on it) is biology-independent, so the argument cannot be flanked by a future result. The debate over sexual ethics is relocated to where it actually lives, moral grounding and design ([Moral Argument](/codex/moral-arguments/), [Worldviews](/codex/worldviews/)), and is carried on the companion pages. - **On the person**, the whole treatment distinguishes an overreaching *claim* from the *people* the claim is about. Every human being is an image-bearer of infinite worth ([Imago Dei](/codex/genesis-1-27/)), the moral standard on sexual conduct is universal rather than a singling-out, and Scripture already grants that desire is often deep and unchosen ([Psalm 51:5](/codex/psalms-51-5/)). Correcting a misused hypothesis is compatible with full compassion, and is owed as much to honesty as to charity. The residue to hold with care is pastoral, not logical. Real people carry real burdens here, and the point of getting the science right is not to score against them but to keep an unfinished hypothesis from being wielded, by anyone, as if it had already spoken. ## Live-cite kit **Scientific and scholarly (for credibility):** - **The model under critique:** William R. Rice, Urban Friberg, and Sergey Gavrilets, "Homosexuality as a Consequence of Epigenetically Canalized Sexual Development," *Quarterly Review of Biology* 87 (2012). The authors' own concession that they cannot supply definitive evidence is the single most useful citation in the exchange. - **The contradicting genetics:** Andrea Ganna et al., "Large-scale GWAS reveals insights into the genetic architecture of same-sex sexual behavior," *Science* 365 (2019), about 477,000 participants, five loci, broadly polygenic, behavior not predictable from the genome. Contradicts the paper's "genetics will find nothing" prediction *and* any strong "gay gene." - **Germline reprogramming:** the standard developmental-biology point that mammalian germ cells and early embryos undergo two extensive waves of epigenetic reprogramming, which is what makes stable transgenerational transmission the exception requiring proof, not the default. - **Confound structure:** the standard caution that apparent transgenerational epigenetic effects must be separated from continued environmental exposure, maternal and uterine physiology, DNA sequence variants that themselves drive methylation, and direct exposure of a fetus's own germ cells. - **Neural distribution:** the observation, consistent with later structural-imaging work on orientation, that sexual attraction is not localized to a single "orientation center" (and certainly not to the frontal cortex alone) but involves distributed reward, salience, emotional, endocrine, and social-cognition systems, so a single tissue-specific mark sparing all neighbors is a strong and unmet requirement. **Logic and philosophy of science (for the structure):** - **Affirming the consequent:** "epi-marks *could* explain the pattern" does not yield "epi-marks *cause* the trait," because many mechanisms produce the same pattern. - **Standards for a strong abduction:** novel prediction, superiority over rivals, survival of falsification attempts, and direct mechanistic confirmation, none yet met. - **Conjunction:** the probability of a multi-step mechanism is the product of its steps, so plausible-per-step does not entail probable-overall. **Aphorisms (for the close):** - *"A model with every slot still empty shows that a thing is imaginable, not that it happened."* - *"Multiply the maybes. Eight of them do not add up to a fact."* - *"Even a proven cause of a desire has never told anyone whether acting on it is right."*
## Common questions this page answers **Q: Does the Rice, Friberg, and Gavrilets 2012 paper prove that homosexuality is biological?** No. It proposes a model, it does not demonstrate one. The paper suggests that inherited "epi-marks" (epigenetic tags that adjust how strongly a fetus responds to androgens) could occasionally pass to an opposite-sex child and produce same-sex attraction. But it identifies no actual mark, no gene the mark controls, no observed case of parent-to-child transmission, and no altered brain circuit, and the authors themselves concede they cannot provide definitive evidence. So it shows the mechanism is conceivable, not that it occurs. It is a testable hypothesis, which is to its credit, but a hypothesis is not a finding. **Q: What is wrong with the epigenetic "epi-marks" theory of homosexuality?** Several things, none of which require denying that epigenetics is real. The proposed mechanism is a chain of about eight separate steps that must all hold at once (a mark exists, is epigenetic, is inherited, survives two rounds of reprogramming, acts tissue-specifically, targets only partner-preference neurons, spares genitals and gender identity, and overrides the child's own marks), and the probability of the whole chain cannot be inferred from the plausibility of each link. It also overstates how well human transgenerational epigenetic inheritance is established (most solid cases are in mice), its supporting twin-study and tissue-specificity reasoning is equally consistent with rival explanations, and its mathematics fits parameters to the theory rather than testing it. **Q: Didn't a big genetic study support the idea that people are born gay?** The largest study to date, Ganna et al. 2019 (about 477,000 people), found five gene locations statistically associated with same-sex behavior and a broadly polygenic pattern, but the effects were tiny and it concluded you cannot predict a person's sexual behavior from their genome. So it points to a small, non-deterministic biological contribution, not a "gay gene." It is relevant to the 2012 epigenetic paper because that paper predicted genetics would find essentially nothing; the 2019 result contradicted that specific prediction while also refuting any strong genetic-determinism reading. The honest summary is a detectable but small and non-deterministic contribution. **Q: Can this paper be used to argue that people are "born this way"?** No, and this is the twist worth knowing. The paper actually argues *against* a simple genetic cause of homosexuality, so it cannot be cited as proof of "born this way" in the genetic sense. And because its own mechanism is unconfirmed, it cannot be cited as proof of biological hardwiring in the epigenetic sense either. It is deployed in both directions in popular debate, and it earns neither, because in both cases it is an unverified model being treated as a settled result. **Q: If homosexuality did turn out to have a biological cause, would that make it morally acceptable?** That is a separate question that the biology cannot answer. "This desire has a biological cause" is a claim about where a desire comes from; "acting on it is right" is a moral claim, and you cannot derive the second from the first. This is the is-ought gap. Many traits have biological or heritable components (alcoholism is roughly 50% heritable) without the associated behavior being thereby good. So whichever way the science lands, on genes, epi-marks, prenatal hormones, or a mix, it leaves the moral question exactly where it was. That question is handled on the companion page, [Born This Way Objection Defeater](/codex/born-this-way-objection-defeater/).
## See also - [Born This Way Objection Defeater](/codex/born-this-way-objection-defeater/), the companion defeater on the broader "born this way / it's genetic / animals do it" argument, where the is-ought core is developed - [Homosexuality](/codex/homosexuality/), the anchor hub, orientation/desire/act distinction and the Christian position - [Biblical Sexual Ethics Objection Defeater](/codex/biblical-sexual-ethics-objection-defeater/), the companion defeater on the biblical-text objections - [Is-Ought Fallacy](/codex/is-ought-fallacy/), why a demonstrated biological cause still would not settle the moral question - [Moral Argument](/codex/moral-arguments/), the grounding problem the moral question actually turns on - [Methodological Naturalism Critique](/codex/methodological-naturalism-critique/), the wider pattern of treating an unconfirmed naturalistic mechanism as a settled result - [Is Being Gay Biological](/codex/is-being-gay-biological/), the source page on the biological-causation question - [Worldviews](/codex/worldviews/), the level the sexual-ethics debate belongs at once the biology is set aside