Nathan Cofnas is a philosopher of biology whose serious academic work is broader than the controversy that made him publicly known. His peer-reviewed research concerns natural selection, innateness, evolutionary mismatch, moral psychology, cultural evolution, scientific norms, nutrition, intelligence, and Kevin MacDonald’s theory of Judaism. His best-supported controversial contribution is his sustained critique of MacDonald’s antisemitic group-strategy theory. His weakest and most consequential move is different: in popular essays after 2023, he treats a substantially genetic explanation of racial IQ gaps as already established and builds a political program on it. The measured gaps, individual-level heritability of intelligence, and genetic population structure are real; they do not by themselves establish the genetic cause or size of a between-group gap. Current molecular-genetic methods do not settle that causal question. Some critics caricature or misquote him, but the core scientific criticism—that his public certainty outruns the direct evidence—is substantive, not merely ideological hostility.
Evidence and attribution boundary
This note distinguishes four things that are often collapsed in coverage of Cofnas:
peer-reviewed philosophical argument;
peer-reviewed empirical or review work;
popular political argument and prediction;
claims made about him by journalists, institutions, and opponents.
“Published in a peer-reviewed journal” does not mean “empirically demonstrated,” especially when the article is conceptual, ethical, or a defense of inquiry. Conversely, finding a critic’s language polemical does not refute the critic’s methodological point.
One-sentence orientation
Cofnas combines analytic philosophy of biology with an increasingly explicit hereditarian, anti-DEI political project, while also opposing Kevin MacDonald’s antisemitic evolutionary narrative and criticizing conspiratorial, populist, and anti-intellectual tendencies on the right.
Why he is difficult to classify
Cofnas does not fit cleanly into a conventional left–right or “far right” box.
He calls himself a race realist and argues that average racial differences in intelligence are substantially genetic.
He wants elites to reject racial equality of innate psychological distributions and describes this as a “hereditarian revolution.”
He has proposed preserving racial distinctions through social barriers and anticipates more racially homogeneous communities, although he explicitly rejects white nationalism, racial hatred, and legal apartheid.
He is a market liberal who attacks protectionism, MAGA communism, right-wing conspiracy culture, and the anti-intellectual “podcast bro” ecosystem.
He argues that many self-described race realists are low-quality thinkers attracted by antisemitism and grievance.
He has written the most sustained scholarly critique of Kevin MacDonald’s claim that Judaism is a consciously or functionally coordinated evolutionary group strategy.
He defends institutions, expertise, and at times paternalistic information control more strongly than many populists on the right.
The most accurate short classification is therefore: a secular, elitist, classical-liberal hereditarian with an anti-populist streak, not a straightforward white nationalist or MacDonald-style ethnonationalist.
Biography and current institutional status
BA, Columbia University.
Graduate study in the history and philosophy of science at Cambridge, followed by a DPhil at Oxford.
Leverhulme Early Career Fellow at the University of Cambridge, 2022–2025.
Former research associate of Emmanuel College; the college ended the affiliation during the controversy over his 2024 essay.
His Cambridge university employment reached its ordinary term; reporting that simply says “Cambridge fired him” collapses the university post and the separate college affiliation.
Cofnas’s website lists him as a postdoctoral researcher at Ghent University. As of 21 August 2026, Ghent had placed him on precautionary suspension while a preliminary disciplinary inquiry considered complaints and “concerning signals.” Suspension is not a completed finding: UPI report on Ghent’s announcement.
flowchart TD
A["Philosophy continuous with science"] --> B["Concepts in evolutionary biology"]
A --> C["Evolutionary accounts of morality"]
A --> D["Science, activism, and free inquiry"]
D --> E["Race and intelligence controversy"]
D --> F["Academic freedom and deplatforming"]
C --> G["Power and cultural evolution"]
E --> H["Hereditarian political program"]
G --> H
D --> I["Critique of Kevin MacDonald"]
I --> J["Antisemitic conspiracy claims rejected"]
Core research programs
1. Philosophy of evolutionary biology
Cofnas’s least political work is often his most conventionally philosophical. He treats biological concepts as explanatory tools that should be revised when standard definitions fail to distinguish different causal processes.
Natural selection without teleology
In “Natural Selection Requires No Teleology in Addition to Heritable Variation in Fitness” (2024), he argues that the familiar conditions—variation, differential fitness, and heritability—are not sufficient to identify natural selection. Artificial selection, intelligent design, forward-looking orthogenesis, and processes in which variation is produced because of its future usefulness can satisfy versions of those conditions. His proposed additional condition is negative: natural selection lacks a future-directed or goal-guided source of adaptive variation.
Assessment: This is a conceptual proposal, not an empirical discovery. Its value lies in sharpening how different evolutionary explanations are classified.
Innateness
In “Innateness as Genetic Adaptation: Lorenz Redivivus (and Revised)” (2017), Cofnas identifies innateness with being a genetic adaptation, while allowing that development still depends on environments and that organisms can inherit non-genetic, phylogenetic information. The proposal tries to preserve a scientifically useful nature–nurture distinction without treating innate traits as developmentally fixed.
Assessment: The account is clear and naturalistic, but it is one contested philosophical definition among several; it should not be imported into political debates as if biology had legislated a single ordinary-language meaning of “innate.”
Evolutionary mismatch
In “A Teleofunctional Account of Evolutionary Mismatch” (2016), he distinguishes forms of mismatch by reference to what a trait was selected to do and the environment in which that function now fails. The paper aims to prevent “mismatch” from becoming a loose label for any modern discomfort or pathology.
2. Morality, culture, and power
Moral progress can be debunked without disproving morality
In “A Debunking Explanation for Moral Progress” (2020), Cofnas argues that convergence toward liberal moral norms may be explained by non-truth-tracking causes: prosperity, declining violence, broadened empathy, and changes in group self-interest. This weakens the inference from apparent moral progress to moral realism. It does not, by itself, prove that liberal norms are false or that objective morality is impossible.
Gene–culture coevolution and moral truth
In “How Gene–Culture Coevolution Can—but Probably Did Not—Track Mind-Independent Moral Truth” (2023), he considers whether socially selected norms could converge on moral truth. He emphasizes that selection for coordination, stability, power, or group advantage is not automatically selection for truth. The paper is best read as a constraint on evolutionary vindications of morality, not a decisive refutation of moral realism.
The Golden Rule
In “The Golden Rule: A Naturalistic Perspective” (2022), Cofnas argues that Golden-Rule-like norms arise from ordinary social cognition and reciprocal reasoning. He also contends that the rule is a distinctive cornerstone of Christianity and does not, in its ordinary forms, entail impartial treatment of everyone.
Assessment: The cross-cultural origin claim is plausible as a naturalistic model; the comparative claim about Christianity depends on how broadly one identifies analogues in Jewish, Confucian, Buddhist, Hindu, Islamic, and classical sources.
Powerful coalitions shape morality
In “Power in Cultural Evolution and the Spread of Prosocial Norms” (2018), he argues that many cultural-evolution models begin from a false picture of prosocial norms as emerging mainly through blind group competition. Powerful majority coalitions can deliberately create and enforce norms, sometimes for a perceived common good and sometimes for their own interests.
This concern with who has power to define norms later becomes central to his political writing on universities, DEI, and hereditarianism.
Incest taboo and moral instincts
In “Are Moral Norms Rooted in Instincts? The Sibling Incest Taboo as a Case Study” (2020), Cofnas challenges a strong Westermarck-effect account. Aversion to childhood co-residents does not straightforwardly supply the represented kin relation or the moralized, general prohibition found in cultural rules. He uses the case to argue that moral norms are less tightly constrained by evolved instincts than some evolutionary psychologists assume.
3. Science, politics, and institutional trust
Cofnas’s recurring sociology-of-science thesis is that scientific institutions can be distorted by moral and political commitments, especially when a field’s conclusions bear on identity and policy.
“Science Is Not Always ‘Self-Correcting’” (2016) uses controversies in intelligence research to argue that reputational and institutional mechanisms can preserve errors as well as correct them.
“Does Activism in Social Science Explain Conservatives’ Distrust of Scientists?” (2018) connects declining conservative trust to the visible liberal activism of social scientists.
The related rejoinder acknowledges that the authors did not directly identify a complete causal chain. The inference is plausible, but the observational trend does not isolate activism from partisan sorting, media ecology, education, religion, or elite cues.
“Scientific Literacy and Attitudes toward Science in the Political Spectrum” (2016) finds a useful asymmetry: self-identified or social conservatives scored lower on scientific literacy and optimism than liberals/progressives, while economic conservatives performed as well as or better than the economic left.
Important tension: Cofnas objects to ideological gatekeeping in mainstream institutions, yet his later popular work endorses a counter-elite project that would use appointments and political power to remove “woke” decision-makers. His principle is not institutional neutrality in every context; it is closer to open inquiry followed by elite contestation for institutional control.
4. Applied ethics and health controversies
Coercive paternalism and intelligence
In “Coercive Paternalism and the Intelligence Continuum” (2020), Cofnas argues that lower general intelligence can make certain reasoning errors and self-destructive choices more likely, potentially justifying targeted paternalistic intervention.
Assessment: The argument raises genuine questions about competence and autonomy but risks moving too quickly from group-level correlations to coercion. Any practical policy would need decision-specific capacity standards, evidence of benefit, due process, the least restrictive alternative, and safeguards against class or disability discrimination.
Red meat guidelines
In the coauthored “Should Dietary Guidelines Recommend Low Red Meat Intake?” (2020), he and Frédéric Leroy argue that the evidence base did not justify categorical population-wide advice to minimize unprocessed red meat.
Vegetarian and vegan diets for children
In “Is Vegetarianism Healthy for Children?” (2019), he argues that available evidence did not justify confident assurances that vegetarian—and especially vegan—diets were safe without careful supplementation and monitoring.
Assessment of the nutrition papers: These are evidence reviews and critiques of guideline certainty, not demonstrations that meat is harmless or that well-planned vegetarian diets are necessarily unhealthy. Guideline evidence and pediatric nutrition literature continue to change, so these papers should not be treated as current clinical advice.
Full study analyses from the “Analyze Moral Progress” conversation
Provenance and evidence boundary
The five collapsed callouts below reproduce the complete assistant responses from the referenced ChatGPT conversation Analyze Moral Progress (6a90f219-7150-83e9-88f2-d9cc5759ead1). They are preserved as prior conversation-derived analysis, including their original qualifications, examples, tables, links, and source lists. Their inclusion records the analysis; it does not convert every statement or citation into a newly verified finding of this note.
Full response — A Debunking Explanation for Moral Progress
The paper is Nathan Cofnas, “A Debunking Explanation for Moral Progress,” published in Philosophical Studies in 2020. It is primarily a response to Michael Huemer’s argument that humanity’s long-term convergence toward liberal values is evidence for objective moral truth. (Springer)
The basic philosophical problem
The debate concerns moral realism.
Moral realism says something like:
Slavery is wrong, not merely because we dislike it or because our culture condemns it, but because there is an objective fact that slavery is wrong.
An antirealist might instead say our moral judgments ultimately arise from human psychology, evolution, culture, interests, emotions, social arrangements, and similar factors. There need not be a mind-independent moral fact “out there” corresponding to them.
The problem for antirealism is that human morality seems to have moved in a surprisingly consistent direction.
Over the last several centuries, societies have increasingly condemned:
slavery, torture, arbitrary killing, racial discrimination, denial of women’s rights, political repression, aggressive warfare, cruel punishment, and so forth.
At the same time, ideas like individual dignity, equality before the law, personal liberty, democracy, and concern for suffering have spread.
Huemer asks:
Why would many different societies independently move in roughly the same moral direction if there were no moral truth for them to discover?
His answer is that this looks somewhat like convergence in science. Scientists from different cultures converge on the germ theory of disease because germs really exist.
Perhaps societies converge on “slavery is wrong” for the same reason: because slavery really is objectively wrong. (Springer)
Cofnas’s paper tries to show that Huemer has moved too quickly.
Cofnas’s central argument in plain English
Cofnas says:
We can explain why humanity became more liberal without assuming that people were discovering objective moral truths.
He proposes a naturalistic causal story.
Very roughly:
self-interest → resistance to domination → liberal institutions → greater peace → more empathy and sensitivity to suffering → broader liberal morality
If that mechanism naturally pushes societies toward liberalism, then convergence does not necessarily mean discovery of moral truth. (Springer)
That is the entire paper in miniature.
A simple analogy
Imagine that isolated civilizations all eventually invent shoes.
You might say:
“Interesting. They all converged on shoes. There must be some objective truth that shoes are the correct form of footwear.”
But you do not need that explanation.
Humans have vulnerable feet. Rocks hurt. Shoes protect feet. Similar human needs interacting with similar environments produce similar solutions.
Cofnas thinks something analogous can happen morally.
Humans generally:
do not like being dominated,
do not want to be murdered,
do not want their property taken,
dislike living in fear,
sympathize with suffering people,
cooperate when cooperation benefits them.
Those common psychological characteristics can push very different societies toward similar institutions.
No perception of an independent moral realm is required.
Where the historical story comes in
Cofnas tells a fairly ambitious story about human political evolution.
Stage 1: Humans became unusually egalitarian
Our primate ancestors probably lived in dominance hierarchies. Cofnas follows Christopher Boehm in arguing that hunter-gatherer humans developed what is sometimes called a reverse dominance hierarchy.
Instead of one alpha dominating everyone else, coalitions prevented would-be alphas from dominating the group.
If someone became too aggressive or controlling, other members could ridicule, ostracize, punish, or sometimes kill him.
So hunter-gatherers frequently had strong norms of:
equality among adult males, food sharing, autonomy, and opposition to domination.
Cofnas therefore says something counterintuitive: modern liberalism is partly a return to an ancient human pattern, rather than something completely new. (Springer)
There is real anthropological support for the broad claim that humans spent much of their history in relatively egalitarian small-scale societies. More recent work still supports that general picture, although it emphasizes considerably more variation than Cofnas’s simplified story suggests. A 2024 review, for example, finds that enduring inequality became much more prominent relatively late in the Neolithic, rather than appearing automatically with agriculture. (PubMed)
Stage 2: Agriculture creates hierarchy
Agriculture allowed:
larger populations, stored wealth, armies, bureaucracies, specialization, rulers, and states.
Large agricultural societies defeated or absorbed smaller societies.
Consequently, societies became much more hierarchical even though the average person might not actually have preferred hierarchy.
The point is important:
institutions can spread because they are militarily successful rather than because individuals think they are morally good. (Springer)
Current research makes this story more complicated. Agriculture itself is neither strictly necessary nor sufficient for hierarchy. Some hunter-gatherers developed substantial inequality, while some farming societies remained relatively egalitarian. Control of concentrated resources, property, institutions, and coercive power appears especially important. (PubMed Central (PMC))
So Cofnas’s broad historical mechanism remains plausible, but his presentation is somewhat cleaner and more linear than contemporary anthropology warrants.
Stage 3: The oppressed push back
This is probably the most intuitive part of his argument.
Suppose 5 percent of society has enormous privileges while 95 percent is exploited.
The 95 percent does not need a philosophical insight into objective moral reality to think:
“I would prefer not to be exploited.”
If ordinary people gain political power, they have self-interested reasons to support:
equality before the law, restrictions on rulers, property protections, political participation, restrictions on violence, and individual rights.
So democratization can generate liberalization because liberal rules protect ordinary people from powerful people. (Springer)
That gives Cofnas a natural explanation for why liberalism repeatedly emerges.
Stage 4: Rights become contagious
This is one of the more interesting parts of the paper.
Imagine group A demands:
“We deserve equality.”
They win.
Group B then says:
“Why doesn’t that principle apply to us?”
Cofnas uses the American and French Revolutions.
American colonists proclaimed equality while maintaining slavery. But slaves and abolitionists could take the language of equality and turn it against slavery.
French revolutionary men proclaimed liberty and equality while excluding women. Women could then ask why the principle did not apply to them.
So liberal principles become politically generalizable.
Cofnas calls this the “contagiousness of revolutions.” (Springer)
This produces something that looks like moral reasoning:
Men deserve political equality.↓Why not women?Whites deserve liberty.↓Why not Blacks?Christians deserve religious freedom.↓Why not Jews?Citizens deserve protection from cruelty.↓Why not foreigners?Humans shouldn't suffer unnecessarily.↓Why not animals?
Huemer might interpret this as people gradually noticing the logical implications of true moral principles.
Cofnas says you can tell much of the same story through groups pursuing their interests once they obtain enough power to do so.
Stage 5: Peace changes our psychology
Cofnas then makes another interesting claim.
It is tempting to say:
Medieval people tolerated brutal executions because their morality was worse. Modern people reject them because we discovered that cruelty is wrong.
Cofnas suggests the causation may sometimes run partly in the opposite direction:
society becomes safer first, and then people’s moral sensitivities change.
Effective states reduce:
murder, raiding, feuds, private revenge, dueling, and interpersonal violence.
When people grow up surrounded by less violence, they become less desensitized to violence.
Something that barely shocked a medieval audience may horrify a modern audience because modern people encounter much less severe violence in ordinary life.
So:
better institutions↓less violence↓less exposure to violence↓greater sensitivity to suffering↓stronger anti-violence norms
Cofnas therefore argues that changing morality may sometimes be the effect of pacification rather than its primary cause. (Springer)
That is a useful challenge to simple “humanity became morally wiser” histories.
His specific claim that the post-1990 American crime decline was mainly due to policing and social controls is too confident. Criminologists have debated the decline extensively, with policing, incarceration, the waning crack epidemic, demographic and economic changes, and other factors receiving varying weights. The broader point, however, survives: violent behavior can fall dramatically without requiring an equally dramatic change in moral beliefs. (AEA Publications)
Stage 6: Empathy expands the moral circle
Human beings evolved empathy.
We particularly empathize with:
family, friends, neighbors, familiar people, and people whose suffering we directly observe.
As groups interact more, outsiders stop looking like abstract enemies and begin looking like recognizable individuals.
That can make suffering psychologically harder to ignore.
Cofnas therefore thinks things like abolitionism can partly spread through emotional mechanisms.
Graphic descriptions of slavery, novels such as Uncle Tom’s Cabin, images of abused slaves, personal relationships, and public testimony make suffering psychologically salient.
Once empathy becomes attached to a norm such as:
“Don’t inflict unnecessary suffering”
that norm can become culturally durable. (Springer)
This helps explain why the moral circle expands.
But here we reach the paper’s biggest philosophical weakness
Cofnas says empathy evolved because it helped our ancestors survive and reproduce.
Therefore, he argues, empathy does not automatically track moral truth.
But the realist has a powerful response:
Why couldn’t evolution give us emotional capacities that allow us to recognize morally relevant facts?
Consider vision.
Evolution gave us eyesight because seeing predators, food, cliffs, and other people increased reproductive success.
But that does not mean:
“Because eyesight evolved for survival, vision doesn’t tell us anything true about reality.”
Vision evolved because accurately detecting reality was often useful.
A moral realist could say something similar about empathy:
Empathy evolved for biological reasons, but it nevertheless helps us recognize a morally relevant fact: other conscious beings suffer.
Cofnas acknowledges exactly this objection and says the realist therefore carries the burden of explaining why empathy should count as truth-tracking. (Springer)
That moves the debate forward, but it doesn’t settle it.
The deepest philosophical issue
This is the distinction I would focus on when evaluating the paper:
Explaining why someone believes something is not the same as showing the belief is false or unjustified.
Suppose I say:
“You believe smoking causes cancer because your biology gave you a desire not to die.”
That explains why you’re interested in the question.
But it does not undermine the evidence that smoking causes cancer.
Likewise:
“You oppose slavery partly because you empathize with slaves.”
doesn’t show:
“Slavery isn’t objectively wrong.”
It might simply explain why you became motivated to notice its wrongness.
This is the potential danger of a genetic fallacy, dismissing a belief merely because of its causal origin.
Sophisticated debunking arguments avoid that fallacy by making the stronger claim that the causal process generating the belief is insensitive to whether the belief is true.
And that is precisely where the real fight lies.
Huemer vs. Cofnas
Question
Huemer
Cofnas
Why did slavery become condemned?
Humanity increasingly recognized its objective wrongness.
Oppressed people resisted, political power shifted, and empathy made abuse harder to tolerate.
Why did equality expand?
People recognized increasingly general moral truths.
Groups demanded for themselves rights already claimed by others.
Why has violence declined?
Moral attitudes toward violence improved.
States, prosperity and institutions reduced violence, which subsequently changed attitudes.
Why has concern expanded?
Reason helps us recognize equal moral worth.
Empathy and changing social circumstances expand concern.
Why do cultures converge?
They are discovering the same moral reality.
Similar humans facing similar incentives produce similar moral systems.
That explanation contains no reference to electrons.
But it obviously doesn’t follow that electrons aren’t real.
Likewise, explaining the social mechanisms that allowed abolitionism to spread doesn’t tell us whether:
Slavery is objectively wrong.
The causal and metaphysical questions are separate.
The most important insight to take from the paper
I would condense Cofnas’s contribution to this:
Moral progress does not necessarily mean moral discovery.
What appears retrospectively as:
ignorance↓reason↓recognition of moral truth↓liberalism
might actually involve:
changing technology↓changing institutions↓changing distributions of power↓greater security and prosperity↓less violence↓greater empathy↓changing moral intuitions↓liberalism
And, importantly, both processes could be operating simultaneously.
That last possibility is where I think the debate ultimately lands. Cofnas gives good reasons to reject a simplistic inference from convergence to moral realism. But he does considerably less to show that reason, argument, consistency, and genuine moral insight played no role in the convergence.
So my assessment would be:
Cofnas successfully weakens Huemer’s empirical argument for moral realism, but he does not successfully debunk moral realism itself. His sociological account is compatible with both:
liberal values spread because of self-interest, institutions, and empathy even though there are no objective moral truths, or
those same social changes removed barriers that had prevented people from recognizing moral truths that were there all along.
The empirical history alone probably cannot decide between those two metaethical interpretations. That is also very close to Cofnas’s more modest official conclusion: convergence by itself does not favor realism or antirealism. (Springer)
Paper: Nathan Cofnas, “A Debunking Explanation for Moral Progress,” Philosophical Studies 177 (2020): 3171–3191. DOI: 10.1007/s11098-019-01365-2. (Springer)
Full response — The Golden Rule: A Naturalistic Perspective
Nathan Cofnas’s “The Golden Rule: A Naturalistic Perspective” (Utilitas, 2022) is closely connected to the previous Cofnas paper. Again, the target is not morality itself. The target is a particular argument for non-natural moral realism.
The paper asks:
Why does something resembling the Golden Rule appear in many different civilizations?
One possible answer is:
Because independent thinkers were rationally discovering the same objective moral truth.
Cofnas argues that we have a simpler naturalistic explanation:
The Golden Rule is an effective way of getting humans, who already possess evolved capacities for empathy and cooperation, to imagine themselves in another person’s position and behave more prosocially.
His larger conclusion is therefore quite modest: cross-cultural convergence on Golden-Rule-like sayings does not provide strong evidence that humans independently perceived a non-natural moral truth. (Cambridge University Press)
The argument Cofnas is attacking
The most important figures here are Katarzyna de Lazari-Radek and Peter Singer, drawing on Henry Sidgwick.
They defend something called the Principle of Universal Benevolence, or PUB.
Very roughly:
From an impartial moral standpoint, everyone’s good matters equally.
Combined with a utilitarian account of “the good,” this becomes something close to:
We ought to impartially maximize the well-being of all sentient beings.
De Lazari-Radek and Singer argue that certain moral intuitions have especially high credibility when three conditions are met:
Careful reflection makes the principle seem self-evident.
Other careful thinkers independently converge on it.
We cannot explain the belief as resulting from an evolutionary or otherwise non-truth-tracking process. (Cambridge University Press)
The Golden Rule appears useful for conditions 2 and 3.
Christianity:
Treat others as you would want to be treated.
Judaism:
Don’t do to others what is hateful to you.
Confucianism:
Don’t impose on others what you wouldn’t want imposed on yourself.
Buddhism, Hinduism, and other traditions contain similar formulations.
So the realist argument starts looking like:
Independent civilizations ↓similar moral principle ↓not obviously explained by natural selection ↓perhaps reason independently discovered it ↓evidence for objective moral truth
Singer and de Lazari-Radek compare this to mathematics. Humans did not evolve specifically to perform advanced mathematics, yet reasoning enables us to discover mathematical truths. Perhaps moral reasoning similarly enables us to discover impartial moral truths. (Cambridge University Press)
Cofnas thinks this argument has two major problems.
Problem 1: The Golden Rule is not actually the same as impartial morality
This is probably the paper’s strongest point.
The Golden Rule sounds like universal impartiality, but Cofnas argues that the two principles are fundamentally different.
Golden Rule
Treat another person as you would want to be treated.
This involves:
perspective-taking
reciprocity
empathy
interpersonal conduct
Principle of Universal Benevolence
Everyone’s good counts equally, and we should impartially promote the good.
This involves:
impartiality
aggregation
everyone’s interests
no privileged moral standpoint
potentially all sentient beings
Those are not equivalent.
For example, imagine you have $100.
The Golden Rule might tell you:
If you were hungry, you’d want someone to help you, so help the hungry person.
Utilitarian impartiality asks something much more demanding:
Of every possible use of this $100 anywhere in the world, which produces the most good?
Nothing in:
“Treat others as you’d want to be treated”
logically entails:
“Maximize total well-being impartially.”
That gap matters enormously.
Cofnas therefore thinks Singer and de Lazari-Radek are reading a modern philosophical conception of impartial morality back into ancient religious sayings. (Cambridge University Press)
Cofnas then goes religion by religion
This is the historical core of the paper.
1. Christianity
Here Cofnas actually concedes a lot.
Jesus explicitly presents the Golden Rule as summarizing religious law:
“Whatever you want men to do to you, do also to them.”
And Christian tradition frequently treats love of one’s neighbor as central.
So Cofnas accepts that Christianity genuinely gives the Golden Rule an unusually central position. (Cambridge University Press)
But he says it still isn’t utilitarian impartiality.
Jesus is not saying:
Take a neutral “view from nowhere” and maximize everyone’s welfare.
Jesus’s reasoning operates within a religious worldview concerning:
God
divine commands
salvation
sin
judgment
love of neighbors and enemies
And it concerns humans, not sentient creatures generally.
This especially matters for Singer because Singer’s moral impartiality includes animals.
So:
Christian Golden Rule≠Singer/Sidgwick impartial utilitarianism
Even if they overlap morally.
Cofnas’s point is not that Christianity is selfish. Obviously Jesus’s commands can be extraordinarily demanding and universalistic. His point is that Christian universal love and utilitarian impartial maximization have different philosophical foundations and contents. (Cambridge University Press)
2. Judaism
A famous Jewish version comes from Hillel:
“What you hate, do not do to others.”
Cofnas argues that this is often presented by modern writers as though Hillel were announcing:
“Here is the fundamental principle underlying all morality.”
But he says that interpretation exaggerates its role in Judaism.
Jewish ethical and religious tradition contains many foundational principles concerning:
Torah
covenant
commandments
Sabbath
justice
charity
holiness
religious duties
Moreover, Cofnas notes that Leviticus’s “love your neighbor as yourself” was historically interpreted by many rabbinic authorities as referring especially to fellow Jews or observant Jews, rather than expressing universal impartial concern for every human being. (Cambridge University Press)
Even if you interpret Hillel universally, Cofnas says:
“Don’t do something to someone that you would hate”
still does not mean:
“Everyone’s welfare counts equally and should be maximized.”
So once again:
reciprocity is not the same thing as utilitarian impartiality.
3. Hinduism
De Lazari-Radek and Singer point to a passage in the Mahabharata that looks remarkably Golden-Rule-like.
Cofnas accepts that such passages exist.
But he argues that extracting them from their religious context creates a misleading impression.
Traditional Hindu ethics revolves around concepts such as:
dharma
ahimsa, non-harm
self-control
caste and social duties
asceticism
karma
afterlife
Cofnas cites scholarship suggesting that Indian commentators historically did not isolate the Golden Rule and present it as the foundational insight from which the rest of morality follows. (Cambridge University Press)
So the problem is not:
“Hinduism doesn’t have anything resembling the Golden Rule.”
It does.
The problem is the inference:
“Therefore Hindu philosophers independently discovered the same foundational principle of impartial morality that Sidgwick discovered.”
Cofnas thinks that substantially overstates the similarity.
4. Confucianism
Confucius gives perhaps the clearest non-Western example:
“Do not impose on others what you yourself do not desire.”
Again, this looks strikingly similar to the Golden Rule.
But Confucian morality is famously role-based and relational.
Your obligations depend partly on whether someone is:
your father
your child
your ruler
your friend
your neighbor
a stranger
Traditional Confucianism therefore contains substantial moral partiality.
You can legitimately owe more to your parents than to strangers.
That is almost the opposite of strict Sidgwickian impartiality.
Cofnas points to the famous Confucian opposition to Mozi, who advocated “inclusive care.” Mencius attacked Mohism precisely because it seemed to eliminate the special concern people should have for parents. (Cambridge University Press)
So:
Confucian reciprocity:Respect others' perspectives.Sidgwickian impartiality:No person's good has intrinsically greater weight.
Very different propositions.
5. Buddhism
Buddhist texts also contain statements resembling:
Since something is unpleasant to me, I should realize that it is unpleasant to another and refrain from inflicting it.
Again, Cofnas doesn’t deny the similarity.
But he argues there is little reason to interpret this as saying:
“The foundational principle of morality is impartial maximization of everyone’s welfare.”
It functions naturally as an injunction against harming others.
Imagine completely isolated parents in five civilizations independently discover:
“If I want my child to understand another person’s suffering, I should ask her to imagine being that person.”
Civilization A says:
“Imagine yourself in his place.”
Civilization B says:
“Would you want someone doing that to you?”
Civilization C says:
“His pain is like your pain.”
Civilization D says:
“Treat your neighbor as yourself.”
You don’t need to hypothesize that all four cultures have apprehended a mysterious non-natural property.
They’ve independently discovered an effective psychological teaching device.
Why does it work?
Because humans possess empathy.
So Cofnas’s model is:
natural selection↓capacity for empathy and perspective-taking↓humans form cooperative societies↓moral teachers want more prosocial behavior↓they discover perspective-taking works↓"How would you like it?"↓Golden-Rule-like teachings emerge independently
Cofnas therefore wants debunking arguments to encompass sociological and cultural causes, not merely biological evolution. (Cambridge University Press)
The fascinating exception: Mozi
Cofnas knows there is an obvious problem.
Some ancient thinkers really did advocate something much closer to impartial morality.
The most important example is Mozi, fourth century BCE China.
Mozi promoted jian ai, usually translated as something like inclusive care or impartial concern.
His basic position was much closer to:
Care about everyone, not merely your own family or state.
And he evaluated institutions partly according to whether they produced general benefit.
That sounds remarkably utilitarian.
Cofnas acknowledges this.
But here’s the interesting part:
Mozi’s argument was religious.
Mozi believed Heaven was a quasi-personal divine authority that cared impartially about all humans.
From Heaven’s perspective:
my kingdom=your kingdomrich=poornoble=commonmy family=your family
Therefore humans should adopt something resembling Heaven’s impartial standpoint. (Cambridge University Press)
Cofnas argues:
That isn’t independently discovering “the point of view of the universe.” It is reasoning from theological premises concerning what Heaven wants.
So you can potentially explain:
religious belief↓belief in impartial Heaven↓impartial ethics
without assuming perception of a non-natural moral truth.
Surprisingly, early Western utilitarianism has a similar story
This is one of the paper’s most interesting historical sections.
Before Bentham’s famous secular utilitarianism, eighteenth-century Anglican theologians such as George Berkeley and John Gay developed theories resembling utilitarianism.
Their reasoning went something like:
God is perfectly good.
God created everyone.
God doesn’t arbitrarily prefer Anthony over John.
Therefore God desires the happiness of mankind generally.
Morality consists in acting according to God’s will.
Therefore we should promote general human happiness.
So:
God loves humanity impartially↓God wants universal happiness↓we should promote universal happiness↓proto-utilitarianism
Cofnas notes that this really is much closer to impartial morality than the Golden Rule.
But again the impartiality is derived from theology.
Why did cultures develop beliefs in morally concerned gods in the first place?
Cofnas appeals partly to work associated with Ara Norenzayan’s “Big Gods” hypothesis.
The basic idea is:
Small societies can police cooperation through:
reputation
kinship
face-to-face observation
retaliation
But large anonymous societies create a problem:
How do you stop strangers cheating one another?
One cultural solution is belief in:
an all-seeing supernatural agent who punishes wrongdoing.
Even when nobody sees you steal:
God sees you.
Groups with such beliefs may cooperate more effectively, allowing their institutions and religions to spread.
Cofnas therefore proposes something like:
need for large-scale cooperation↓culturally successful belief in moralizing gods↓increasingly universal conception of God↓God cares about everyone↓theologians reason from God's perspective↓universalistic moral doctrine
That produces seemingly impartial morality through natural and cultural processes. (Cambridge University Press)
This is speculative in places, and Cofnas presents it as a plausible debunking route rather than a demonstrated genealogy for every case.
Now we reach the most important philosophical question
Just as with the previous Cofnas paper:
Does explaining the origin of a belief actually debunk the belief?
This is where I think Cofnas’s paper becomes much less decisive.
Evolution → empathy → moral reflection → Golden Rule
doesn’t automatically undermine the Golden Rule either.
Cofnas knows this. He explicitly says debunking arguments do not establish that the targeted moral belief is false. They attempt to undermine its justification by showing that its causal origin is insensitive to the relevant truth. (Cambridge University Press)
So everything turns on truth-tracking.
The realist’s strongest response
The moral realist can grant Cofnas almost everything and reply:
“Yes, empathy evolved. Yes, cultural teachers discovered ways of activating it. Yes, the Golden Rule has social functions. But why assume empathy isn’t helping us perceive morally relevant features of reality?”
Consider pain.
Evolution obviously made us sensitive to pain for survival reasons.
Yet:
someone screaming in agony really is suffering.
Empathy might draw our attention to a morally relevant fact:
Other people have experiences just as real as ours.
Perspective-taking then makes something cognitively obvious:
My suffering matters to me.Other people have the same capacity for suffering.There is nothing metaphysically special about "me."Therefore I need some justification for treating my suffering as intrinsically more important.
At that stage, perhaps reason really is doing something.
Cofnas has explained why we engage in the reasoning process.
He has not necessarily shown that the reasoning process fails to track moral truth.
That is exactly analogous to the problem we encountered in his previous paper.
An important distinction: psychological origin vs rational justification
Suppose I tell someone:
“Imagine if you were the slave.”
That might trigger empathy.
But once the person thinks about the situation, she might reason:
I hate being arbitrarily dominated.
The slave is relevantly similar to me.
Race isn’t a morally relevant difference.
Therefore it would be inconsistent to demand liberty for myself while denying it to him.
Cofnas’s naturalistic account explains the first very well.
It is much less obvious that it explains away the second.
And that is where moral realists can resist the debunking conclusion.
Another important issue: the Golden Rule isn’t actually sufficient as morality
Cofnas is right about this.
Taken literally:
“Do to others what you want done to yourself”
can produce absurd results.
Suppose:
I love extremely spicy food.
You hate it.
The literal Golden Rule says:
Give you extremely spicy food.
Obviously wrong.
Or imagine someone who enjoys pain.
Should he therefore hurt other people?
Obviously not.
That’s why philosophers introduce what is sometimes called the Platinum Rule:
Treat others as they would reasonably want to be treated.
Cofnas discusses this and treats both Golden and Platinum rules primarily as useful moral heuristics, rather than complete moral theories. (Cambridge University Press)
That is an important insight.
The Golden Rule functions better as:
“Get outside your own perspective.”
than:
“Here is an algorithm capable of solving every moral problem.”
What I think Cofnas establishes convincingly
I would give him three fairly substantial victories.
1. Golden Rule ≠ utilitarianism
This seems clearly correct.
The distance between:
“Treat others as you’d want to be treated”
and:
“Impartially maximize the aggregate welfare of all sentient beings”
is enormous.
Singer and de Lazari-Radek need additional philosophical premises to bridge that gap.
2. Cross-cultural similarity has probably been overstated
There really is an important methodological warning here.
Finding superficially similar quotations in:
Christianity
Judaism
Hinduism
Buddhism
Confucianism
doesn’t establish that the traditions understood them identically.
The surrounding ethical systems differ dramatically.
A Confucian principle embedded within filial hierarchy cannot automatically be equated with Singerian impartiality.
Cofnas’s contextualization makes the “all religions independently discovered utilitarian impartiality” claim look substantially weaker. (Cambridge University Press)
3. Debunking explanations don’t have to be genetic
This is probably the paper’s most important general point.
“Natural selection wouldn’t directly favor this belief”
does not imply:
“There is no naturalistic explanation for the belief.”
That is a genuine challenge for some evolutionary arguments for moral realism. (Cambridge University Press)
Where I think Cofnas goes too far
The sentence I’d scrutinize most closely is his claim that because empathy is a product of natural selection, there is no reason to think it is a moral-truth-tracking emotion. (Cambridge University Press)
That doesn’t straightforwardly follow.
Evolutionary origins can coexist with truth-tracking.
Evolution selected:
vision
memory
causal reasoning
object permanence
numerical approximation
precisely because organisms capable of representing reality reasonably accurately often survive better.
Of course evolution can also produce systematic biases.
So the proper conclusion is not:
evolved → non-truth-tracking.
It is:
evolved → truth-tracking status requires additional argument.
The realist therefore still has room to argue that empathy supplies morally relevant information while rational reflection corrects its biases.
For example, empathy is notoriously partial:
One suffering child in front of me can move me more than 10,000 anonymous people overseas.
Reason may tell me:
Their suffering isn’t less real simply because I can’t see them.
Ironically, that could support Singer’s larger picture:
evolution gives us limited empathy↓reason notices inconsistencies↓reason universalizes concern↓impartiality
Cofnas hasn’t eliminated that possibility.
Cofnas vs. Singer in one table
Question
Singer / de Lazari-Radek
Cofnas
Why does the Golden Rule appear widely?
Independent rational convergence
Similar humans discovering similar moral teaching techniques
Debunking needn’t involve direct genetic selection
Why did impartial ethics emerge?
Rational thinkers take the impartial standpoint
Sometimes theological reasoning produces impartiality
Does convergence support non-natural realism?
Yes, potentially
Much less than realists claim
How this paper connects to Cofnas’s previous paper
The two papers fit together extremely well.
The previous paper argued:
Moral progress toward liberalism can have a naturalistic explanation.
This paper argues:
Cross-cultural convergence on the Golden Rule can have a naturalistic explanation.
The structure is almost identical:
REALIST ARGUMENTIndependent convergence↓unlikely to be evolutionary↓best explained by rational discovery↓evidence of objective moral truth
Cofnas replies:
NATURALISTIC ALTERNATIVEshared human psychology+similar social problems+cultural evolution+empathy/cooperation+religious and educational institutions↓independent convergence
Therefore:
Convergence alone cannot establish moral realism.
That is the common thesis connecting both papers.
The deepest lesson from the paper
The question is not:
“Can morality be explained naturally?”
Almost certainly many aspects of morality can.
The harder question is:
Does the naturalistic explanation make moral truth causally irrelevant to why we hold the belief?
That is what a genuine debunking argument needs.
For example:
Strong debunking case
Suppose people believe:
“Friday the 13th is unlucky.”
And historians show exactly how the superstition spread through cultural transmission, confirmation bias, and selective memory.
None of those mechanisms cares whether Friday the 13th genuinely produces bad luck.
That’s strongly debunking.
Much harder moral case
Suppose people believe:
“Inflicting extreme suffering on someone for amusement is wrong.”
We discover that:
empathy evolved,
cooperation was adaptive,
societies punish sadists,
parents teach perspective-taking.
Does that show the belief has nothing to do with the fact that the victim is actually experiencing terrible suffering?
Not obviously.
The naturalist still has to show:
Our process would produce approximately the same belief regardless of whether the moral proposition were true.
Cofnas supplies a plausible causal story, but he doesn’t conclusively demonstrate that counterfactual.
My overall assessment
I would rate the paper’s arguments separately.
Very strong: The Golden Rule has been incorrectly equated with Sidgwickian or Singerian impartiality.
Strong: Quoting similar Golden-Rule passages across religions obscures major differences in their historical meaning and importance.
Strong: Moral debunking explanations can be cultural and sociological rather than directly evolutionary.
Plausible but not decisive: Empathy plus cultural evolution provides a naturalistic explanation for the recurrence of Golden-Rule-like teachings.
Much weaker: Therefore our belief in the Golden Rule is epistemically debunked as a possible non-natural moral truth.
That last inference still requires an argument showing that empathy, perspective-taking, consistency reasoning and moral reflection fail to track moral reasons.
Cofnas does not really settle that.
The paper in one paragraph
Cofnas argues that philosophers such as Singer, Parfit and de Lazari-Radek have made too much of the Golden Rule’s appearance across civilizations. The sayings found in Christianity, Judaism, Hinduism, Buddhism and Confucianism are not all formulations of the same impartial utilitarian principle, and outside Christianity the Golden Rule generally was not regarded as the foundation of morality. Its recurrence also has a straightforward naturalistic explanation: humans evolved empathy and cooperative tendencies, and moral teachers independently discovered that asking people to imagine themselves in someone else’s position is an effective way to encourage prosocial behavior. Even genuinely impartial systems such as Mohism and early Anglican proto-utilitarianism emerged through theological reasoning that may itself have naturalistic cultural explanations. Cofnas therefore substantially undermines the argument that cross-cultural agreement on the Golden Rule is evidence for non-natural moral realism, but he does not show that the Golden Rule is false, nor does he conclusively show that empathy and rational reflection cannot help us discover genuine moral truths. (Cambridge University Press)
Citation: Nathan Cofnas, “The Golden Rule: A Naturalistic Perspective,” Utilitas 34, no. 3 (2022): 262–274. DOI: 10.1017/S0953820822000073. (Cambridge University Press)
Full response — Power in Cultural Evolution and the Spread of Prosocial Norms
The paper is Nathan Cofnas, “Power in Cultural Evolution and the Spread of Prosocial Norms,” published in The Quarterly Review of Biology in 2018. It is less directly philosophical than the two Cofnas papers we discussed before. This one is primarily about how morality and social norms evolved, and it supplies part of the empirical machinery Cofnas later uses in his arguments about moral progress and moral realism. (RCNi Company Limited)
Its central claim is:
Human moral norms did not simply emerge because individuals independently copied successful behaviors. Humans often deliberately designed, imposed, and enforced moral rules through collective power.
That sounds straightforward, but it is a fairly serious challenge to one influential family of theories about cultural evolution.
The basic problem Cofnas is addressing
Cofnas is criticizing the cultural-evolutionary tradition associated especially with Robert Boyd, Peter Richerson, and Joseph Henrich.
Their basic approach treats culture somewhat analogously to biological evolution.
Instead of genes spreading, you have cultural variants spreading:
beliefs, customs, technologies, norms, institutions, religious practices, food habits, and so on.
People don’t copy randomly. They have learning biases.
For example:
Copy successful peopleCopy prestigious peopleCopy what most people are doingCopy practices that seem usefulCopy members of your own group
Eventually some cultural practices become common.
Then groups with particularly effective practices may outperform groups with ineffective practices.
So you can get:
individual imitation ↓different cultural practices ↓some practices stabilize ↓some groups become more successful ↓successful practices spread ↓cultural evolution
This is cultural selection.
Cofnas does not reject cultural evolution in general. His argument is narrower.
A standard cultural-learning explanation might ask:
Why did Religion A spread?
Perhaps:
A had more appealing doctrines↓people preferentially convertedorprestigious people practiced A↓others copied themorcommunities practicing A prospered↓other communities copied A
But now imagine the king says:
Convert to Religion A or leave the country.
Suddenly the psychological appeal of Religion A might not matter very much.
The ruler has changed people’s available options.
Cofnas uses the 1492 Alhambra Decree as an example. Ferdinand and Isabella ordered practicing Jews in Spain either to leave or convert. Whatever cultural-learning biases Jews possessed, coercive state power radically changed which cultural practices could persist publicly. (RCNi Company Limited)
So Cofnas wants us to distinguish:
"I adopted this because I found it attractive."
from:
"I adopted this because people with power made me."
Those are completely different mechanisms of cultural change.
What Cofnas means by “power”
He distinguishes several forms of influence.
Consider Michael Jordan advertising a shirt.
People might buy it because:
“Jordan is successful and prestigious, so I want what he wears.”
That fits perfectly well with prestige-biased cultural learning.
But compare that with a king saying:
“Wear this clothing or I will imprison you.”
Now you aren’t copying the king because he is prestigious.
You are complying because he controls consequences.
Cofnas thinks two kinds of power are particularly important:
Force: someone removes alternatives through coercion.
Authority: people recognize someone’s right to issue certain commands.
The Pope’s authority over committed Catholics is therefore different from a celebrity’s prestige. The Pope can potentially define acceptable doctrine precisely because followers recognize an institutional authority relationship. (RCNi Company Limited)
This becomes crucial when Cofnas turns to morality.
The really interesting part: hunter-gatherers
Someone could reply:
“Fine. Power matters in kingdoms and modern states. But our moral psychology evolved hundreds of thousands of years ago among small hunter-gatherer bands without kings, governments, armies, or police.”
If hunter-gatherers lacked significant power structures, perhaps the standard cultural-evolution model still works for the period in which human moral psychology evolved.
Cofnas’s answer is:
Hunter-gatherers did have power. It was simply organized differently.
And this is where Christopher Boehm’s theory of reverse dominance hierarchy becomes central. (RCNi Company Limited)
Chimp hierarchy vs. human reverse hierarchy
Cofnas starts with our primate relatives.
A chimpanzee troop normally has something resembling:
Alpha ↓Beta ↓Gamma ↓lower-ranking individuals
Dominant individuals use physical force and alliances to control resources and mating opportunities.
Cofnas, following Boehm, argues that human beings inherited conflicting tendencies:
desire to dominate+capacity to submit+strong dislike of being dominated
The crucial human innovation was the ability of many weaker individuals to coordinate against one powerful individual.
Language, weapons, coalition-building, gossip, and planning allowed several ordinary people to overwhelm a would-be alpha.
So instead of:
Alpha ↓everyone else
you get:
majority coalition ↓ ↓ ↓ ↓ ↓potential dominator
The hierarchy is effectively reversed.
The group dominates the individual who attempts to dominate the group. (RCNi Company Limited)
This is one of the most important ideas in the paper
Hunter-gatherer egalitarianism, on this account, wasn’t simply an absence of power.
It was actively maintained.
If someone tried to become too domineering, others might escalate through:
Cofnas therefore says egalitarianism itself can be understood as a political achievement maintained through collective power. (RCNi Company Limited)
This is philosophically interesting because egalitarianism can look like:
“Nobody has power.”
Cofnas/Boehm’s account says it may actually be:
“Everyone collectively exercises enough power to stop any individual from acquiring too much power.”
That is a very different conception of equality.
Morality enters the picture
Cofnas argues that this majority coalition did not merely suppress aspiring dictators.
It also enforced rules concerning:
sharing, violence, theft, deception, sexual behavior, cooperation, hunting, taboos, and other socially important conduct.
In Boehm’s ethnographic sample of 50 mobile foraging groups, lethal punishment was recorded in 24, although less extreme sanctions were far more common. Cofnas’s central point is not that hunter-gatherers were constantly executing one another, but that behind ordinary norms stood a credible possibility of collective sanction. (RCNi Company Limited)
This produces:
group agrees on acceptable behavior↓members monitor one another↓violations become public through gossip↓group sanctions violators↓individuals learn to follow norms↓norms become internalized
And eventually:
external enforcement↓selection favors people good at learning norms↓shame / guilt / reputation sensitivity↓internal conscience
This is Cofnas’s account of part of the evolution of norm psychology. (RCNi Company Limited)
What is “norm psychology”?
It means humans seem psychologically prepared to think:
“There is a correct way people are supposed to behave.”
Children don’t merely observe that adults behave a certain way.
They readily infer:
“That’s how you’re supposed to behave.”
And they sometimes police other children who break those rules.
So humans seem to possess psychological mechanisms for:
recognizing norms, internalizing norms, feeling shame when violating norms, caring about reputation, condemning violations, and punishing violators.
Both Cofnas and the theorists he criticizes accept something like this.
People interact↓some randomly/conformistically adopt cooperation↓some adopt punishment of cheaters↓cooperative groups perform better↓their norms spread↓genes favor brains adapted to learning those norms↓norm psychology evolves
Notice how bottom-up this is.
Nobody needs to understand the entire system.
It is somewhat like biological evolution.
Birds didn’t consciously design wings.
Likewise humans need not have consciously designed cooperative institutions.
Cultural variation plus cultural selection can produce functioning institutions without anyone understanding why they work. (RCNi Company Limited)
Cofnas’s alternative
Cofnas says:
That may describe some cultural evolution, but it badly mischaracterizes morality.
His proposed history looks more like:
humans develop language + coalition abilities↓majority overthrows alpha domination↓group deliberately establishes rules↓group collectively enforces rules↓individuals must conform↓selection favors ability to internalize norms↓norm psychology evolves
The key difference is intentional institutional design.
People weren’t necessarily blindly stumbling into cooperation.
They could reason:
“If everybody cheats during the hunt, we’re screwed.”
“If one guy takes all the meat, the rest of us suffer.”
“We need everyone to contribute.”
“Therefore we’ll punish anyone who cheats.”
Cofnas thinks humans possessed enough foresight to deliberately create these arrangements. (RCNi Company Limited)
The Cephu story makes Cofnas’s argument concrete
Cofnas uses Colin Turnbull’s famous ethnography of the Mbuti.
The Mbuti hunted by arranging nets cooperatively.
A man named Cephu secretly moved his net forward so that more animals would enter his net before reaching everyone else’s.
Essentially:
Everyone cooperates+Cephu manipulates the system=Cephu gets disproportionate benefit
The others discovered what he had done.
Information spread.
The group collectively condemned him.
Cephu tried to defend himself by claiming importance and status.
The others rejected that claim and threatened the social consequences of separation from the group.
Eventually he apologized and surrendered his day’s meat. Soon after, relations returned to normal. (RCNi Company Limited)
For Cofnas, this illustrates something that a simple “individual punishment” model misses.
Nobody independently thought:
“I personally feel like punishing Cephu.”
Instead:
violation↓discussion/gossip↓collective moral judgment↓group consensus↓coordinated sanction↓compliance
The group itself functions as a political actor.
That’s central to the paper.
Why Cofnas thinks laboratory experiments can mislead us
He discusses experiments in which people can choose between:
a system allowing punishment of free riders, and a system without punishment.
Over time, people often migrate toward the sanctioning system because cooperation works better there.
A cultural evolutionist might say:
See? Cooperative institutions can emerge through individuals choosing successful systems.
Cofnas objects that real hunter-gatherer life didn’t necessarily offer:
“Would you personally like to subscribe to the punishment system?”
If the group established meat-sharing rules, an individual couldn’t simply say:
“No thanks. I personally subscribe to a sanction-free institutional regime.”
They were inside a political community.
This is analogous to modern law.
You don’t individually choose whether the prohibition against theft applies to you.
The institution applies to everyone within its jurisdiction. (RCNi Company Limited)
This is why Cofnas thinks some experiments artificially “atomize” people.
They model:
individualindividualindividualindividual
while actual societies sometimes act as:
GROUP / | | \ individuals
Collective agency is really Cofnas’s deeper argument
The paper repeatedly attacks what Durham called “radical individualism.”
Suppose five people meet and decide:
“From now on everyone must contribute to group defense.”
Then they enforce the agreement.
Cofnas says it can be misleading to describe what happened as nothing more than:
Individual A copied something.
Individual B copied something.
Individual C copied something.
Something additional happened:
They deliberated and collectively created an institution.
That is much closer to political science, sociology, or institutional economics than simple cultural imitation.
Cofnas thinks cultural evolutionary theory needs to take this seriously. (RCNi Company Limited)
This is where Elinor Ostrom becomes relevant
Cofnas invokes experiments associated with Elinor Ostrom.
When people were allowed to communicate about common-pool resource problems, groups could sometimes deliberately establish sanctioning arrangements and achieve very high levels of cooperation.
This matters because it challenges the idea that effective institutions necessarily arise through blind cultural selection.
People can sometimes:
identify collective-action problem↓discuss possible solution↓anticipate free riding↓design sanction↓implement institution↓cooperate successfully
In other words:
Humans are not merely subjects of cultural evolution. They can partly become agents of cultural evolution.
That is probably the deepest insight of the paper.
Cofnas is NOT saying everything is consciously designed
This qualification is important.
He accepts that enormous amounts of cultural knowledge emerge without deliberate design.
Nobody individually invented:
languages, cooking traditions, thousands of technological tricks, or many religious customs.
This resembles Hayek’s idea of spontaneous order.
Useful systems can emerge without any single person’s understanding them.
Cofnas’s argument is instead:
Don’t turn spontaneous order into a universal explanation of human culture.
Humans sometimes do consciously design rules.
And morality, because it concerns behavior affecting everyone else, may be particularly likely to involve collective enforcement and institutional design. (RCNi Company Limited)
Why morality is different from spear-making
This distinction is extremely useful.
Imagine two hunters make spearheads differently.
One hunter prefers:
Technique A
Another prefers:
Technique B
The group may not care.
If Technique A works better, others may eventually copy it.
That’s standard cultural evolution.
But compare:
Hunter A shares meat.Hunter B refuses to share.
Now everyone has an interest.
The group might say:
“You don’t get to choose your preferred personal meat-sharing philosophy. These are our rules.”
So Cofnas distinguishes approximately:
technical culture→ often individual learning/imitationmoral/social culture→ often collective decision + enforcement
He therefore does not conclude that mathematical cultural-evolution models are worthless.
His claim is that their assumptions are especially problematic when explaining the historical evolution of morality and prosocial norms. (RCNi Company Limited)
An important qualification: “prosocial” doesn’t necessarily mean morally good
This is worth emphasizing.
A norm can promote cooperation within a group while being terrible toward outsiders.
Imagine:
Everyone contributes to the army.Nobody deserts.Everyone obeys commanders.Everyone sacrifices for the group.
Those are highly cooperative norms.
But the society might use that cooperation to invade its neighbors.
Likewise moral enforcement could support:
egalitarianism, food sharing, mutual defense,
but also:
religious taboos, oppressive gender norms, persecution, caste systems, or aggressive warfare.
Cofnas himself recognizes that power can sustain both beneficial and harmful cultural arrangements. (RCNi Company Limited)
So the paper is explaining how norms become powerful, not demonstrating that the resulting norms are ethically correct.
How hierarchy later changes the mechanism
There is an interesting historical implication.
Hunter-gatherer society:
majority coalition↓enforces rules against potential dominators↓egalitarian norms
Later sedentary agricultural society:
stored wealth+property+armies+political hierarchy↓elite gains coercive power↓elite can enforce rules favorable to itself
So the same basic technology of morality:
“People ought to behave this way”
can serve radically different power structures.
Cofnas puts the contrast starkly: hunter-gatherers can use morality to reinforce egalitarianism, while hierarchical societies can use morality to reinforce hierarchy. (RCNi Company Limited)
This anticipates his later paper on moral progress.
There is also an alternative evolutionary theory Cofnas takes seriously
Cofnas discusses Nicolas Baumard and colleagues’ mutualistic theory of morality.
This theory doesn’t focus primarily on collective punishment.
Instead imagine early humans choosing cooperation partners.
If you’re:
reliable, fair, generous, competent,
other people want to cooperate with you.
If you’re:
selfish, lazy, dishonest, exploitative,
people avoid you.
So natural selection favors a psychology concerned with fairness and reputation.
The basic mechanism is:
I need cooperation partners↓people prefer fair partners↓I benefit from being perceived as fair↓selection favors fairness psychology
Cofnas thinks this probably explains part of morality.
He does not insist that collective punishment explains everything.
He suggests the two mechanisms may be complementary:
partner choice+reciprocity+collective norm enforcement=human moral psychology
"This norm spread because institutions enforced it."
is real and extremely important.
Even within contemporary cultural-evolution research, later work has increasingly modeled intentional institutional design alongside selective imitation. A 2021 paper by Sergey Gavrilets and Mahendra Duwal Shrestha explicitly compares selective imitation with self-interested institutional design, including people’s ability to anticipate future benefits from punishment institutions. That is very close to the kind of mechanism Cofnas said needed greater attention. (ScienceDirect)
So on that broad methodological point, Cofnas looks prescient rather than misguided.
Where I think Cofnas overstates the case
There are several places where I would be more cautious.
1. His hunter-gatherer picture is probably too uniform
The paper says very strongly that nomadic hunter-gatherers universally maintained extreme male egalitarianism and that powerful majority coalitions deliberately enforced prosocial norms. (RCNi Company Limited)
The broad picture of strong egalitarian and anti-dominance norms remains influential. A 2022 review of human dominance, for example, describes community sanctions against aggression, intimidation, and domineering behavior among contemporary hunter-gatherers. (PubMed Central (PMC))
But modern research emphasizes substantially more variation in leadership and enforcement.
A 2025 Nature Communications analysis of 23 egalitarian hunter-gatherer societies found that leaders play an important role in teaching cultural norms, which actually supports Cofnas’s larger argument that norm transmission isn’t merely anonymous imitation. But the authors also emphasize that leaders in these societies generally lack coercive authority, relying heavily on prestige, persuasion, teaching, and social norms. (Nature)
Cofnas’s mechanism probably belongs in the explanation, but it needn’t monopolize it.
2. The emphasis on execution may be too strong
Cofnas repeatedly points to lethal sanctions as evidence that hunter-gatherer norms rested on real coercive power. (RCNi Company Limited)
But a large 2023 cross-cultural study of 131 nonindustrial societies found that the form of punishment varies systematically with social organization. Reputational punishment was associated with egalitarianism and absence of food storage, while execution was moderately associated with greater social stratification. (PubMed)
That doesn’t refute Cofnas. His argument only requires meaningful collective enforcement, and gossip, ridicule, ostracism, and reputation can accomplish that.
But it suggests that the strongest version:
“egalitarian forager morality was backed by serious physical coercion”
probably needs softening.
A more defensible formulation would be:
Egalitarian norms were actively socially enforced, with sanctions ranging from reputation and ridicule to occasional severe coercion.
3. “Deliberate design” and “cultural evolution” aren’t really opposites
This is, I think, the paper’s biggest conceptual weakness.
Suppose people consciously design a new law.
That’s still a cultural variant.
Different groups can then:
copy it, modify it, compete with it, preserve it, or abandon it.
So you can have:
deliberate design↓cultural variation↓selection among designs↓further deliberate modification↓more cultural selection
The fact that institutions are intentionally designed doesn’t mean they aren’t evolving culturally.
Modern cultural-evolution research increasingly treats these processes as complementary rather than mutually exclusive. The 2021 Gavrilets and Shrestha model is a good example: foresight and institutional design can interact with selective imitation, and the two mechanisms can reinforce one another. (ScienceDirect)
So I think Cofnas is stronger when he says:
“existing models omit an important causal variable”
than when he suggests:
“the phenomenon can’t legitimately be modeled using cultural evolutionary theory.”
The former is convincing.
The latter is too strong.
4. “Power makes learning biases irrelevant” is also too categorical
Imagine the Pope issues a moral teaching.
People might comply partly because:
authority+prestige+religious belief+conformity+fear of social exclusion
Similarly, a government law can change behavior, but people may subsequently internalize the behavior through conformity or moral education.
How this connects to the two previous Cofnas papers
This is where the three papers form a coherent project.
This 2018 paper establishes:
morality is shaped bypower + institutions + coalition interests + social enforcement
Then “A Debunking Explanation for Moral Progress” argues:
historical changes in power + institutions↓liberalization↓apparent moral progresstherefore:moral convergence need not imply discovery of objective moral truth
And “The Golden Rule: A Naturalistic Perspective” adds:
evolved empathy + social cooperation + cultural teaching↓Golden-Rule-like doctrinestherefore:cross-cultural convergence need not imply apprehension of non-natural moral truth
So the larger Cofnas project looks approximately like this:
evolved psychology+coalition politics+power+institutions+cultural transmission+historical circumstances↓moral norms and moral change
Then Cofnas asks:
If these mechanisms can explain why our moral beliefs arose, how much evidential weight should we place on moral convergence as evidence for objective moral truth?
This 2018 paper therefore provides a significant part of the causal background for the later debunking papers.
But notice what this paper does NOT establish
This is the same distinction I emphasized with the previous papers.
Suppose Cofnas is completely correct that:
group power↓meat-sharing rule↓norm internalization↓fairness intuition
That explains:
Why humans believe certain forms of sharing are morally required.
The historical genealogy and the metaethical truth question remain distinct.
To get a genuine debunking argument, Cofnas later needs an additional premise:
These causal mechanisms would generate our moral beliefs regardless of whether those beliefs corresponded to objective moral truths.
This paper itself largely stays on the empirical side of that issue.
The deepest insight I would take from the paper
I would summarize it as:
Morality is not merely something individuals believe. Morality is also a technology for coordinating and governing groups.
That changes how you think about moral history.
Instead of imagining:
individual gets moral idea↓others find idea persuasive↓idea spreads
Cofnas wants us to recognize:
people face collective problem↓group discusses desired behavior↓rule is established↓social power enforces rule↓people grow up internalizing rule↓rule begins to feel morally obvious
That is a powerful point.
It helps explain why moral intuitions can feel deeply personal while being profoundly shaped by institutions.
But it does mean that philosophers should be careful about treating the apparent obviousness or universality of a moral intuition as straightforward evidence of independent moral perception.
My overall assessment
I think this is actually one of the more interesting Cofnas papers for understanding his larger project.
His strongest contribution is identifying a genuine blind spot in highly individualistic models of cultural evolution: human beings deliberately create institutions, collectively enforce norms, exercise authority, and use coercion. Moral rules are particularly susceptible to these processes because they concern conduct that affects other group members.
His Boehm-inspired account of reverse dominance is plausible and remains influential. There is good evidence that egalitarian foragers actively suppress domineering behavior through collective social sanctions. (PubMed Central (PMC))
His weakest claim is that this seriously undermines cultural-evolutionary modeling itself. Later researchers have demonstrated that institutional design, leadership, foresight, punishment, selective imitation, and cultural selection can all be incorporated into broader evolutionary models. (ScienceDirect)
And his description of hunter-gatherer societies is too uniform and coercion-heavy. Recent comparative evidence suggests that teaching, prestige, consensus, reputation, leadership, partner choice, and noncoercive influence also matter considerably. (Nature)
Nathan Cofnas argues that influential models of cultural evolution underestimate power and collective agency. Human moral norms did not necessarily emerge because isolated individuals happened to imitate cooperative behaviors that were subsequently favored by cultural group selection. Following Christopher Boehm, Cofnas argues that hunter-gatherer humans formed egalitarian “reverse dominance” coalitions that deliberately suppressed would-be dominators and collectively created and enforced rules concerning sharing, violence, cheating, and cooperation. Living under these enforced norms may then have created biological selection pressures favoring conscience, shame, reputation sensitivity, and a general psychology for internalizing social rules. The paper’s broad insight, that moral evolution involves institutions, power, intentional rule design, and collective enforcement rather than merely imitation, is strong and has been reinforced by subsequent work on institutional design and leadership. Its stronger claims, especially that traditional cultural-evolutionary models are fundamentally incapable of handling these processes and that hunter-gatherer moral regulation was uniformly coercive, are much less secure. (RCNi Company Limited)
Citation: Nathan Cofnas, “Power in Cultural Evolution and the Spread of Prosocial Norms,” The Quarterly Review of Biology 93, no. 4 (2018): 297–318. DOI: 10.1086/700767. University of Chicago Press article
Full response — Are Moral Norms Rooted in Instincts? The Sibling Incest Taboo as a Case Study
The paper is Nathan Cofnas, “Are Moral Norms Rooted in Instincts? The Sibling Incest Taboo as a Case Study,” published in Biology & Philosophy in 2020.
It is particularly useful alongside the three Cofnas papers we just discussed because it makes an important qualification to evolutionary explanations of morality:
Even if evolution gives humans an instinct that produces behavior resembling a moral rule, that does not mean the instinct produced the moral rule.
Cofnas uses the incest taboo as his main case study, but his real target is much broader. He thinks philosophers and evolutionary psychologists too often move too quickly from:
“Evolution gave us psychological tendency X”
to:
“Therefore moral norm Y arose from X.”
He argues that there can be a surprisingly large conceptual gap between the two. (Springer)
The basic idea in plain English
Consider two propositions:
Personal instinct: “I don’t want to have sex with the person I grew up with.”
and:
Moral norm: “Brothers and sisters ought not have sex with each other.”
They sound related.
But Cofnas says they are actually different in two major ways.
EVOLVED INSTINCTI avoid sexwith people I grew up with. ↓ ? ↓MORAL RULEEveryone should avoid sexwith their biological siblings.
The question mark is the entire paper.
Cofnas argues that standard evolutionary explanations often simply assume that this transition occurs.
He wants an explanation of how it occurs. (Springer)
First, what is the Westermarck effect?
Edward Westermarck proposed in the nineteenth century that humans evolved a mechanism for avoiding inbreeding.
Evolution faces a problem.
Having children with a close genetic relative increases the probability that harmful recessive mutations will be expressed in offspring.
But your brain does not come equipped with a DNA-testing machine.
So evolution needs some reasonably reliable cue of relatedness.
One such cue is:
Did I grow up in intimate contact with this person?
Historically, someone you lived with from very early childhood was likely to be your sibling.
So natural selection could produce:
early childhood coresidence↓sexual disinterest / aversion↓less sibling mating↓less inbreeding
This is the Westermarck effect.
There is evidence broadly consistent with some kind of early coresidence effect. Researchers have studied cases such as Taiwanese “minor marriages,” in which unrelated future spouses were raised together from childhood, as well as communal child-rearing in Israeli kibbutzim. Some studies have found associations between early coresidence and reduced attraction. (Springer)
Cofnas does not need to deny that some such mechanism exists.
His argument is:
Even if the Westermarck effect is completely real, it does not automatically explain the incest taboo.
That distinction is crucial.
Problem 1: The representation problem
This idea comes originally from Bernard Williams.
Ask what the Westermarck mechanism actually detects.
It cannot detect:
“biological sibling.”
Instead, approximately:
“person with whom I experienced intimate childhood coresidence.”
Those usually overlap.
But not always.
Imagine:
Case A
You are adopted as an infant and raised alongside another unrelated adopted child.
The Westermarck mechanism might say:
sexually unattractive.
But the incest norm says:
not siblings, therefore no incest.
Case B
You have a biological sibling separated from you at birth.
You meet at age 25.
The Westermarck mechanism might not generate aversion because you never coresided.
But the incest norm still says:
siblings, therefore prohibited.
So:
Westermarck mechanism tracks:CHILDHOOD CORESIDENCEIncest taboo tracks:BIOLOGICAL / SOCIAL KINSHIP
“Our moral prohibition on eating poisonous berries evolved from our bitter-taste aversion.”
You would expect the instinct and norm to approximately target the same thing.
But imagine the instinct actually targeted:
“anything colored blue”
while the cultural prohibition targeted:
“berries containing toxin X.”
Even if blue berries often contained toxin X historically, you’d need an additional explanation for how:
avoid blue things
became:
everyone should avoid toxin X
That’s essentially what Cofnas thinks has been glossed over in the incest literature.
A proposed solution: maybe humans reinterpret the instinct
Neven Sesardić offers an interesting response.
Perhaps humans don’t consciously experience:
“I am sexually repelled by people with whom I spent ages zero through six.”
Instead, we notice:
“I’m not sexually attracted to my sister.”
And our brain interprets the cause as:
She’s my sister.
So:
actual mechanism:coresidence → aversionsubjective interpretation:siblinghood → aversion
Then perhaps the incest taboo naturally becomes framed around siblings.
This is a clever solution to the representation problem. (Springer)
Cofnas’s reply: this may be circular
Cofnas asks:
Why is “sibling” such an obvious and socially significant category in the first place?
Today it obviously is.
We have entire systems of rules about:
parents
children
siblings
cousins
inheritance
marriage
family obligations
But some of the reason those categories are so socially salient is precisely because culture already regulates them.
So Sesardić may inadvertently be saying:
Why did sibling norms arise?Because "sibling" was alreadyan important social category.Why was sibling important?Partly because society hadnorms about siblings.
Cofnas doesn’t claim this completely disproves the proposed solution, but he thinks it means more historical explanation is required.
Problem 2: The moralization problem
This is, in my view, the more important point.
Suppose we completely solve the representation problem.
Imagine your biological instinct literally says:
“I do not want to have sex with my sibling.”
We still haven’t obtained:
“Nobody should have sex with their sibling.”
Look at the change:
PERSONAL PREFERENCEI don't want X.↓???↓MORAL NORMNobody ought to do X.
Why should my preference determine what you are allowed to do?
This is Arthur Wolf’s moralization problem. (Springer)
A simple nonsexual example makes the point obvious
Suppose evolution gives me an intense dislike of broccoli.
It obviously doesn’t follow:
“Nobody should eat broccoli.”
Or suppose humans naturally prefer:
helping their own children.
That does not automatically generate:
“Every person has a moral duty to preferentially help their own children.”
Indeed, Cofnas points out an interesting complication.
An evolved parental instinct says approximately:
I want MY child to receive resources.
But the moral norm says:
Everyone should preferentially help THEIR OWN children.
Those can conflict.
If I were purely following my evolved self-interest, I might prefer:
I help my children.+You also help my children.
rather than:
I help mine.You help yours.
So even here, the moral principle doesn’t simply reproduce the content of the instinct. (Springer)
This is one of the paper’s strongest examples.
This directly challenges Sharon Street
This is philosophically important.
Sharon Street’s famous Darwinian Dilemma argues that evolution substantially shaped our basic evaluative tendencies.
Examples might include intuitions like:
Protect your children.
Help your relatives.
Avoid serious bodily harm.
Punish cheaters.
Street’s broader point is that natural selection strongly influenced what humans came to regard as reasons or values.
Cofnas says there is a missing step.
Evolution might produce:
I love my child.
But our moral system contains:
Parents have special dutiestoward their children.
Those aren’t identical.
Likewise:
I retaliate against someonewho harms me.
isn’t equivalent to:
People who violate others'rights ought to be punished.
So Cofnas thinks even some sophisticated evolutionary accounts may overestimate how directly evolved instincts determine the content of morality. (Springer)
That has interesting consequences for evolutionary debunking arguments, which I’ll return to below.
The obvious response: maybe disgust gets moralized
Westermarck himself proposed something resembling this.
Then when I imagine you committing incest, I experience something like the same disgust.
So:
I imagine sibling sex↓sexual disgust↓I imagine other people doing it↓vicarious disgust↓"That's morally wrong."
Cofnas calls this the vicarious disgust hypothesis. (Springer)
That would potentially solve the moralization problem.
My personal aversion becomes generalized to third parties.
But disgust does not automatically become morality
Cofnas points out an obvious problem.
Humans find many things disgusting without considering them immoral.
For example:
bodily fluids
rotten food
picking one’s nose privately
certain unusual foods
You can think:
“That’s revolting.”
without thinking:
“That person has committed a moral wrong.”
So:
DISGUST≠ automaticallyMORAL CONDEMNATION
There must be some additional mechanism explaining why particular disgust reactions become moralized. (Springer)
Research since Cofnas’s paper broadly supports a nuanced version of this point. A 2023 meta-analysis covering 512 effects and more than 72,000 participants found a real association between trait disgust sensitivity and moral condemnation, about r = .23, but the relationship varied strongly by domain and was especially pronounced for sexual disgust and purity/sanctity judgments. So disgust clearly matters, but it is not a general-purpose mechanism automatically transforming anything disgusting into moral wrongdoing. (Sage Journals)
Experimental attempts to make people more morally condemnatory simply by inducing incidental disgust have generally produced much weaker effects, with some meta-analytic work suggesting effects become very small or disappear after correcting for publication bias. (Sage Journals)
So Cofnas’s basic caution here has aged reasonably well.
The kibbutz evidence
This is probably the most empirically detailed part of the paper.
Israeli kibbutzim historically provide something resembling a natural experiment.
Children who were not biologically related sometimes grew up extremely closely together in communal peer groups.
If Westermarck is right:
early coresidence↓sexual aversion
And if vicarious disgust produces morality:
sexual aversion toward peers↓condemnation of peer-group membershaving sex with each other
A 2012 study by Lieberman and Lobel reported evidence that longer coresidence predicted sexual aversion and that personal aversion was associated with attitudes concerning relationships between fellow peer-group members. (Springer)
At first glance, this looks like exactly what Westermarck predicts.
Cofnas argues that it isn’t.
Cofnas reexamines the data
There were two studies.
Study 1
Greater coresidence did not significantly predict moral condemnation of third-party sexual relationships between kibbutz peers.
That’s already a problem.
Study 2
There was an association between coresidence, sexual aversion, and how respondents ranked peer-group sex.
But Cofnas emphasizes that the effect was quite small.
More importantly:
respondents treated sibling sex as dramatically more morally problematic than sex between unrelated childhood peers.
That is strange for the strong Westermarck account.
These participants had spent more intimate childhood time with their peer group than with their biological siblings.
If coresidence-generated aversion were doing the main causal work, you’d expect those judgments to line up much more closely. (Springer)
Cofnas therefore concludes that the study does not establish:
personal Westermarck aversion↓general moral incest taboo
particularly because the cultural category sibling remains so much more morally important than childhood coresident.
An even more interesting complication
Other work on the kibbutzim complicates the classic Westermarck interpretation further.
Shor and Simchai conducted detailed interviews with former kibbutz members and found that many respondents recalled sexual attraction, rather than strong sexual disgust, toward at least some peers.
They also found evidence that peer groups themselves sometimes developed social norms discouraging intragroup romantic relationships.
Respondents described concerns such as:
embarrassment
group cohesion
awkwardness after breakups
disruption of close relationships
So perhaps:
not:coresidence → automatic disgust → no relationshipsbut partly:cohesive social group↓romantic relationships create risks↓group discourages them↓attraction is suppressed or not acted upon
Cofnas regards this as evidence that culture may sometimes produce or reinforce aversion, rather than aversion simply producing culture. (Springer)
That causal reversal is significant.
Alternative explanation 1: preserve family cohesion
Perhaps incest rules developed because sexual relationships inside tightly connected families create social problems.
Imagine siblings begin a sexual relationship and then have a bitter breakup.
They cannot simply stop seeing each other.
Parents, siblings and relatives become involved.
Or sexual rivalry within a family could destabilize relationships.
So societies might reason:
family is a permanent cooperative unit+sexual relationships create rivalry/conflict↓keep sex outside family
This is the social cohesion hypothesis.
Cofnas thinks this probably identifies something real. (Springer)
But he says it suffers from another representation problem.
If the real principle is:
Don’t have sex inside highly cohesive nonvoluntary groups,
why does morality specifically say:
Don’t have sex with biological relatives?
There are plenty of cohesive groups of unrelated people.
And plenty of biological relatives don’t live together.
So again:
underlying mechanism≠actual norm
Cofnas therefore thinks social cohesion may reinforce incest taboos without completely explaining their origin. (Springer)
Alternative explanation 2: people noticed inbreeding was dangerous
Cofnas finds another hypothesis more promising.
Perhaps humans simply discovered:
Close relatives who reproduce together have offspring with more problems.
Then they created rules against those matings.
Call this the recognition hypothesis.
The logic is:
observation:close-relative reproductionsometimes produces unhealthy offspring↓causal inference:close inbreeding is dangerous↓collective interest:avoid costly unhealthy offspring↓social rule:close relatives should not reproduce
This avoids Cofnas’s two central objections.
No representation problem
The observed danger tracks:
biological relatedness
which is what the rule tracks.
No moralization problem
There is a clear reason to care about third-party behavior.
In small communities, severely impaired offspring could impose costs not merely on the parents but on the larger group.
Cofnas thinks there is some historical evidence for this
He discusses William Durham’s cross-cultural work suggesting that many societies explicitly connected incest with bad outcomes for offspring.
Durham also reported a relationship between the breadth of incest prohibitions and the presumed genetic costs of inbreeding in different population structures.
Cofnas interprets this as at least suggestive evidence that humans may have deliberately tailored some incest rules to perceived consequences. (Springer)
He does not claim this definitively explains every incest taboo.
In fact, he explicitly says different societies may have reached similar rules through different historical pathways. (Springer)
That’s an important qualification.
But there is a puzzle here
Today many people’s immediate reaction to sibling incest isn’t:
“I’m concerned about recessive alleles.”
It is:
“That’s disgusting and wrong.”
Sometimes people maintain that judgment even when a hypothetical case eliminates pregnancy.
So if the rule originated because of genetic risks, why does it now feel like an intrinsic moral prohibition?
Cofnas’s answer is cultural persistence.
A rule might begin:
Don't do X because X causes Y.
Generations later:
Don't do X.
Eventually:
X itself is morally disgusting.
The original justification can disappear while the norm remains culturally entrenched. (Springer)
This is another important idea in the paper.
A very simple modern analogy
Imagine a community establishes:
“Never eat fish from Lake A.”
Originally:
Lake A was contaminated↓fish caused illness↓don't eat Lake A fish
After 200 years:
nobody remembers the contamination,
the lake is clean,
children are taught that eating those fish is shameful,
people feel disgust watching someone eat one.
Now the cultural norm has become psychologically self-sustaining.
Its current emotional basis need not reveal its historical causal origin.
Cofnas thinks moral psychologists frequently overlook this possibility.
This produces a very different picture of morality
A simplistic evolutionary psychology story looks like:
But Cofnas emphasizes that actual incest regulations vary substantially between cultures and historical periods.
He cites societies that tolerated or even encouraged relationships that other societies classified as incestuous, including examples from Roman Egypt and historical Zoroastrian communities, as well as cultures permitting exceptions for ruling families. (Springer)
The important point isn’t that incest norms are completely arbitrary.
They aren’t.
It’s that:
Shared biology produces recurring themes without necessarily producing identical moral rules.
That is much subtler.
A useful three-level distinction
I think this is the easiest way to understand Cofnas.
Level 1: Biological disposition
I feel protective towardpeople I grew up with.
Level 2: Social problem
Families need stable systemsof cooperation and reproduction.
Level 3: Moral institution
Siblings have special duties.Parents must care for children.Sibling sex is prohibited.
A lot of evolutionary psychology implicitly jumps:
Level 1 → Level 3
Cofnas says:
Level 2 and cultural history matter enormously.
That is the core paper.
Where I think Cofnas is strongest
1. The moralization problem is genuinely powerful
This is probably his best argument.
Evolutionary psychologists sometimes explain:
why I want X
and then treat that as an explanation of:
why we believe everyone ought to X.
Those are simply different questions.
The transition from:
desire
to:
social prescription
needs explanation.
This applies far beyond incest.
For example:
Retaliation
Evolution:
I want revenge against someone who harmed me.
Moral law:
Impartial institutions should punish wrongdoers according to general rules.
Not the same thing.
Kin altruism
Evolution:
I favor my relatives.
Morality:
Everyone has special obligations toward their relatives.
Again, not the same.
Status
Evolution:
I want higher status.
Morality might say either:
respect hierarchy
or:
suppress status domination.
The same instinct could contribute to very different normative systems.
Cofnas is right to insist that the middle causal steps cannot simply be assumed. (Springer)
2. The representation problem is also important
Natural selection usually cannot directly detect abstract categories such as:
Getting from one to the other requires cognitive and cultural work.
Again:
evolved cue detector≠explicit moral concept
Cofnas is right about that.
3. His treatment of disgust is largely compatible with subsequent evidence
Cofnas argues against:
disgust → automatic moralization.
Later meta-analytic work supports a more qualified relationship.
Disgust sensitivity does correlate with moral judgment, especially around sexuality and purity, but it is neither sufficient nor universally moralizing. (Sage Journals)
So I would modify rather than reject Cofnas:
Not:disgust has nothing to do with moralityand not:disgust automatically produces moralityBetter:disgust is one input into moral judgment,especially in sexual/purity domains,whose effect depends on cultural interpretationand other cognitive processes.
That fits the current evidence better.
Where I think Cofnas is weaker
1. He may demand too close a match between instinct and norm
This is probably the strongest objection to the paper.
Suppose evolution gives us a crude mechanism:
avoid childhood coresidents sexually
Humans possess sophisticated cognition.
We eventually discover:
childhood coresidenceusually tracks kinship+close kin mating causes problems+"sibling" is a useful social category
The resulting norm becomes:
Don’t have sex with siblings.
Does the fact that the norm has more precise content really show that the instinct played no important causal role?
Not necessarily.
The instinct could provide the initial emotional bias while culture refines and generalizes it.
For example:
rough biological disposition↓cognitive interpretation↓social discussion↓conceptual categorization↓formal moral norm
Cofnas is correct that those intermediary stages need explanation.
But:
“The instinct does not directly entail the norm”
is weaker than:
“The instinct did not substantially cause the norm.”
The latter is much harder to establish.
An analogy
Human language capacity is biological.
But evolution didn’t encode:
English grammar.
Our biology instead supplies capacities that cultural systems elaborate.
The mismatch between:
innate linguistic capacities
and:
English rule:"add -ed for many past-tense verbs"
doesn’t mean biology had nothing important to do with English.
Likewise:
biological kin-avoidance mechanisms
could strongly shape:
cultural incest rules
even when they don’t contain identical representations.
So Cofnas’s argument is strongest against simple one-step evolutionary stories, not necessarily against every gene-culture coevolutionary explanation.
2. Moralization doesn’t need to be logically entailed by the instinct
Cofnas repeatedly emphasizes:
There is no logical connection between my aversion and a universal rule.
That’s true.
But causal explanations don’t require logical entailment.
Different cultures may assign different weights to each mechanism.
That seems more plausible than searching for one universal origin story.
Cofnas actually moves toward this pluralistic view himself when discussing alternative theories. (Springer)
The biggest philosophical implication: Sharon Street’s debunking argument gets harder
This is particularly interesting given the papers you’ve been having me analyze.
Street’s evolutionary debunking argument against moral realism relies heavily on the thought that evolution has significantly influenced our evaluative judgments.
This creates an interesting tension in Cofnas’s larger project
The previous Cofnas papers we discussed offer naturalistic debunking explanations for things like:
liberal moral progress
the Golden Rule
moral convergence
But this paper warns against overly simple evolutionary debunking explanations.
There is no contradiction.
His position across the papers is roughly:
BAD NATURALISTIC EXPLANATION:Evolution gave us instinct X.Therefore moral belief Y came from X.
versus:
BETTER NATURALISTIC EXPLANATION:evolved psychology+historical institutions+culture+power+interests+social learning+reasoning↓moral belief Y
That actually unifies the papers very well.
He isn’t saying:
Morality isn’t naturally explicable.
He’s saying:
The natural explanation is much more culturally mediated than simplistic evolutionary psychology assumes.
That is a significant distinction.
How this connects all four Cofnas papers
You can now see a fairly coherent intellectual architecture.
2018: Power in Cultural Evolution
Moral norms aren't merelyindividual instincts or copied behaviors.Power, coalitions and institutionsactively construct and enforce them.
2020: Are Moral Norms Rooted in Instincts?
Don't assume an evolved instinctdirectly produces a moral norm.The content may be completely different.
2020: A Debunking Explanation for Moral Progress
Liberal moral convergence can arisethrough institutions, power shifts,security and empathy.Therefore convergence alonedoesn't prove moral realism.
2022: The Golden Rule
Cross-cultural moral similaritycan emerge through shared psychologyplus cultural mechanisms.Therefore convergence alonedoesn't prove non-natural moral truth.
Put together:
EVOLUTIONprovides psychological raw materials ↓CULTUREorganizes them ↓POWER AND INSTITUTIONSenforce particular norms ↓HISTORICAL SELECTIONpreserves some norms ↓SOCIALIZATIONmakes norms intuitive ↓MORAL PSYCHOLOGYcan then feel innate or self-evident
I think that is the best way to understand Cofnas’s broader project.
The deepest insight from this paper
I’d reduce it to this distinction:
Having an instinct is not the same as endorsing a norm.
Or more formally:
"I want X"≠"X is morally good""I dislike X"≠"X is morally wrong""I avoid X"≠"Everyone ought to avoid X""I favor my relatives"≠"Everyone has special dutiesto their relatives"
The transition from the left side to the right side is one of the central problems in explaining the evolution of morality.
Cofnas argues that evolutionary psychology too often assumes that transition instead of explaining it.
That criticism is powerful.
Where I land overall
I would rate the paper’s major claims this way:
Claim
Assessment
The Westermarck effect and incest taboo have different contents
Strong
Personal aversion doesn’t automatically explain universal moral condemnation
The kibbutz evidence establishes Westermarckian moralization
Cofnas gives good reasons for skepticism
Westermarckian biology plays no important causal role in incest norms
Too strong
Incest taboos may result partly from recognition of inbreeding risks
Plausible but historically underdetermined
Specific moral norms are less biologically predetermined than some evolutionary accounts suggest
Strong
The distinction I would make is:
Cofnas does a better job undermining the simple Westermarck story than proving that evolved incest-avoidance psychology was historically unimportant.
A plausible synthesis remains:
evolution provideskin-recognition and incest-avoidance tendencies↓culture interprets and categorizes them↓observation and social interestssupply additional reasons↓institutions formulate explicit kinship rules↓socialization and disgust reinforce them↓incest becomes a moral taboo
That preserves Cofnas’s strongest insight while allowing biological predispositions to have substantial causal influence.
The paper in one paragraph
Nathan Cofnas argues that the incest taboo, often treated as a textbook example of an evolved instinct becoming a moral norm, actually exposes a major problem for evolutionary explanations of morality. The proposed Westermarck mechanism makes individuals sexually avoid childhood coresidents, while the incest taboo prohibits everyone from having sex with siblings. This creates two gaps: the representation problem, because the instinct and norm concern different categories, and the moralization problem, because a personal aversion does not explain a universal “ought.” Cofnas argues that neither reinterpretation of the aversion as kin-directed nor a proposed mechanism turning sexual disgust into condemnation of third parties adequately closes those gaps, and he argues that kibbutz evidence has been overstated. He considers cultural explanations, especially recognition of the harms of inbreeding, more promising. The larger lesson is that evolution may supply the psychological ingredients of morality without directly determining the content of moral norms. Culture, reasoning, institutions, history and social interests can transform those ingredients into rules whose content differs significantly from the instincts on which they partly depend. (Springer)
Citation: Nathan Cofnas, “Are Moral Norms Rooted in Instincts? The Sibling Incest Taboo as a Case Study,” Biology & Philosophy 35, article 47 (2020). DOI: 10.1007/s10539-020-09764-0. Springer article
Full response — Coercive Paternalism and the Intelligence Continuum
The paper is Nathan Cofnas, “Coercive Paternalism and the Intelligence Continuum,” published online in 2018 and appearing in Behavioural Public Policy 4(1) in 2020, pp. 88–107. DOI: 10.1017/bpp.2018.4. (Cambridge University Press)
This paper is quite different from the previous Cofnas papers. It is primarily political philosophy and applied ethics, not moral psychology.
Its central argument is provocative but fairly precise:
If we sometimes legitimately override people’s choices because predictable cognitive biases make them act against their own interests, then in principle we should also sometimes override choices when serious mistakes result from insufficient cognitive ability.
In other words, Cofnas argues that advocates of paternalism have focused too much on biases that everyone shares and not enough on differences in people’s ability to understand complicated decisions. (Cambridge University Press)
The basic problem
Start with John Stuart Mill.
Mill’s famous Harm Principle says, roughly:
Competent adults should be free to make their own choices, including bad ones, unless they harm other people.
So:
I smoke cigarettes. ↓I hurt myself. ↓Normally my business.I drive drunk. ↓I endanger others. ↓Government can intervene.
One reason Mill gives for respecting autonomy is epistemic:
You generally know your own interests better than other people do.
I know:
what I enjoy
what risks I’m willing to take
what goals matter to me
how I value different outcomes
Therefore the government should be extremely hesitant to say:
“We know what is good for you better than you do.”
Cofnas begins from this traditional liberal position. (ResearchGate)
Then behavioral economics creates a problem for Mill
Daniel Kahneman, Amos Tversky, Richard Thaler, Cass Sunstein and others showed that humans are not perfectly rational decision-makers.
We systematically suffer from things like:
present bias
status quo bias
framing effects
loss aversion
overconfidence
procrastination
anchoring
For example:
Long-term preference:"I want to save for retirement."Immediate behavior:"I'll start next month."Next month:"I'll start next month."
The person’s own long-term goals and actual behavior conflict.
That gives paternalists an opening.
Thaler and Sunstein: libertarian paternalism
Thaler and Sunstein propose libertarian paternalism.
Instead of banning bad choices, government or institutions can nudge people.
For example:
Retirement planOld system:You must opt in.↓Many procrastinate.↓Low savings.New system:You are automatically enrolled,but can opt out.↓More people save.
Nobody is forced to participate.
The choice architecture is changed so that ordinary cognitive biases are more likely to produce an outcome people themselves regard as desirable.
Hence:
Paternalistic because behavior is intentionally influenced.
but:
Libertarian because people retain the right to opt out.
Psychologists have repeatedly found a general factor of cognitive ability, conventionally called g.
Someone who performs relatively well on one cognitively demanding task tends, statistically, to perform relatively well on many others.
So:
verbal comprehension ↕quantitative reasoning ↕working through complicated instructions ↕abstract reasoning ↕problem solving → partly reflect g
Cofnas argues that this matters outside school.
Cognitive ability predicts performance in many real-life domains, including health management and other complex tasks. The broader association between intelligence and health outcomes remains well established in later research; a major 2021 review describes substantial prospective associations between cognitive ability, disease and mortality. (Nature)
So the claim:
“IQ merely measures ability to take IQ tests”
does not fit the evidence.
But that still doesn’t establish Cofnas’s paternalistic conclusion. We need several more steps.
Cofnas uses adult literacy to make the problem concrete
One of the strongest parts of the paper uses the old National Adult Literacy Survey.
It tested adults on realistic tasks involving:
prose
documents
quantitative reasoning
Rather than asking:
“Complete the number sequence.”
it might ask someone to:
interpret a schedule, understand a bill, calculate a cost, or extract information from a document.
The environment becomes increasingly cognitively demanding.
So differences in reasoning ability that might matter relatively little in a simpler environment can have much larger consequences in a bureaucratic and technological society.
That is an important and, at least in broad outline, plausible insight.
The medical example is Cofnas’s strongest case
Suppose a doctor tells a diabetic patient:
If you experience symptoms X, Y and Z, you need to take action A.
That requires the patient to:
understand instructions↓remember them↓recognize symptoms↓connect symptoms to instructions↓weigh consequences↓take appropriate action
Cofnas cites research showing that some patients with low functional health literacy have serious difficulty doing this.
That can have consequences far beyond choosing the wrong bus.
Mismanagement of chronic disease can lead to:
hospitalization
disability
blindness
amputation
death
Later systematic reviews strongly confirm the broad relationship Cofnas relies on: lower health literacy is associated with difficulty interpreting medical information, poorer medication use, greater hospitalization and emergency-care utilization, and worse outcomes. (PubMed)
So there is a genuine empirical problem here.
Cofnas then asks an uncomfortable question
Imagine a patient repeatedly misunderstands a complicated medication regimen.
You explain it.
You simplify it.
You provide written instructions.
You remind them.
They still cannot reliably comply.
And the foreseeable consequence is:
serious permanent disability or death.
At some point, Cofnas asks:
Why is preserving complete autonomy automatically more important than saving the person’s life?
His proposed answer is:
Sometimes it isn’t.
That’s where coercive paternalism enters. (PhilArchive)
His strongest example
For a patient unable to manage a dangerous chronic condition, Cofnas imagines escalating intervention:
education↓reminders↓monitoring↓supervised treatment↓coercive supervision if necessary
In extreme cases, he suggests someone might even be required to live in a supervised environment if the alternative is catastrophic self-harm that they are unable to understand or prevent.
He suggests something similar could conceivably apply to cases of severe obesity when a person cannot manage a medically dangerous regimen and faces imminent serious harm. (PhilArchive)
This is where the paper becomes much more controversial.
Why ordinary information campaigns may not be enough
Suppose a public-health campaign says:
“Smoking greatly increases your risk of lung cancer.”
Two very different problems could prevent behavioral change.
Problem A: cognitive bias
You understand the evidence perfectly well but think:
“I’ll quit next year.”
or:
“It probably won’t happen to me.”
That’s overconfidence or present bias.
Problem B: comprehension
You don’t actually understand:
probabilities
long-term risk
competing risks
medical evidence
Then simply adding more warnings may accomplish very little.
Cofnas says these require different interventions.
BIAS:understands evidence+misweights it↓nudge may helpLOW COMPREHENSION:cannot adequately process evidence↓more information may not help
That distinction is genuinely important. (PhilPapers)
High intelligence doesn’t mean perfect rationality
Cofnas explicitly rejects:
“Smart people don’t make stupid decisions.”
They obviously do.
Highly intelligent people:
procrastinate
become addicted
fail to save
make emotional decisions
succumb to framing effects
take irrational risks
So his model isn’t:
high IQ = rationallow IQ = irrational
It is:
everyone↓susceptible to biases+lower cognitive ability↓additional difficulty withcertain complex reasoning tasks
Therefore different individuals might require different kinds of assistance. (ResearchGate)
Cofnas’s argument in its strongest formal form
I would reconstruct it like this:
People sometimes seriously harm themselves because of reasoning errors.
Some such errors arise from widespread cognitive biases.
If Conly is right, sufficiently serious bias-induced errors can sometimes justify coercive paternalism.
But reasoning errors also arise from differences in cognitive ability.
General cognitive ability varies continuously rather than simply dividing adults into “competent” and “incompetent.”
Lower cognitive ability can make certain complex health, financial and practical decisions substantially harder.
There is no morally relevant reason why an error should justify intervention only when its psychological cause is a universal bias rather than limited reasoning capacity.
Therefore, if coercive paternalism is justified at all, differences in cognitive ability can sometimes be relevant to when it is justified.
That is actually a much narrower conclusion than:
“Smart people should control less intelligent people.”
And distinguishing those two claims is essential to reading the paper fairly. (PhilPapers)
The paper is conditional on accepting Conly
This is probably the single most important thing to understand.
Cofnas does not really prove coercive paternalism from scratch.
His argument is closer to:
IF Conly is right that reasoning failures sometimes justify coercion, THEN there is no obvious reason to limit that principle to errors caused by universal cognitive biases.
For example, a 2023 meta-analysis found a large association between general intelligence and financial literacy, approximately r = .62 across more than 62,000 participants. (ScienceDirect)
Health research similarly finds robust relationships among cognitive ability, health literacy and health outcomes. (PubMed)
So Cofnas is on solid ground saying:
Differences in cognitive ability have practical consequences.
3. “Everyone is biased” is not necessary for paternalism
Suppose Conly’s principle is:
Override a decision when a preventable reasoning error will cause catastrophic self-harm.
Then it seems strange to add:
“…but only if I personally share the same tendency to make that error.”
Why should that matter?
Cofnas is convincing here.
The relevant factors should seemingly be:
How serious is the mistake?How reliably can we identify it?Can less intrusive intervention fix it?What are the costs of coercion?
rather than:
“Does everyone make this particular kind of mistake?”
But this is where I think the paper becomes much weaker
The empirical argument gets Cofnas to:
Cognitive ability affects decision quality.
It does not easily get him to:
Therefore government should coercively restrict lower-ability adults.
There are several enormous steps between those propositions.
Problem 1: IQ is not the same thing as decision-making capacity
This is probably the paper’s most important weakness.
In medicine and law, capacity is generally assessed for a particular decision, not simply inferred from global intelligence.
The standard clinical framework looks at whether someone can:
Anthony’s particular financial choice is irrational.
This distinction is crucial:
STATISTICAL:lower cognitive ability→ greater probability of mistakes
doesn’t yield:
INDIVIDUAL:this person→ making a mistake right now
Using IQ itself as a trigger for coercion therefore produces false positives.
Some lower-IQ individuals will understand a decision perfectly well.
Some very high-IQ individuals will completely misunderstand it.
The higher the cost of mistakenly overriding a competent person’s autonomy, the more serious this problem becomes.
Problem 3: “Bad outcome” is not necessarily “bad decision”
This is perhaps the deepest philosophical objection.
Suppose I know smoking will shorten my life.
But I genuinely value:
20 years of smoking enjoyment>possible extra longevity
You might think that’s foolish.
But have I made a reasoning error?
Not necessarily.
I may simply have different preferences.
Cofnas actually acknowledges the possibility of what we might call the rational smoker: someone who understands the risks and nevertheless concludes that smoking fits his own goals. (PhilArchive)
That creates a huge problem for paternalism.
To justify coercion we need to distinguish:
"I don't understand the consequences."
from:
"I understand them and value things differently."
Otherwise paternalism quietly changes from:
helping people achieve their own interests
into:
forcing people to live according to our conception of their interests.
That is exactly the danger Mill worried about.
Problem 4: Intelligence does not tell us what someone’s interests are
Suppose a person values:
religious duty over longevity
adventure over safety
pleasure now over wealth later
family obligations over personal wealth
creative freedom over stable employment
A more intelligent observer may be better at calculating consequences.
But that observer isn’t automatically better positioned to decide:
which consequences ought to matter most to the person.
This distinction is fundamental:
MEANS:"How do I achieve X?" ↑intelligence may help considerablyENDS:"Should I value X?" ↑IQ does not answer this
Cofnas is strongest when discussing instrumental mistakes:
“I want to remain alive but literally cannot understand how to manage my insulin.”
He is much weaker when the paternalist must decide what the person’s ultimate interests ought to be.
Problem 5: Cofnas underestimates institutional error
This is a major omission.
Cofnas compares:
individual makes mistakevs.competent paternalist prevents mistake
But governments are made of fallible people.
The real comparison is:
ERROR Aindividual incorrectlychooses for himselfversusERROR Binstitution incorrectlyoverrides individual
And institutions suffer from:
bad information
bureaucratic incentives
ideological bias
political pressure
corruption
discrimination
regulatory capture
rigid rules
inability to understand individual circumstances
A paternalistic policy therefore needs to demonstrate not merely:
“Individuals make predictable errors.”
but:
“The institution is sufficiently better at identifying and correcting those errors that the gains exceed both coercion costs and institutional mistakes.”
That’s a much higher hurdle.
Problem 6: The “who watches the watchers?” problem
Cofnas’s proposal creates a potentially dangerous hierarchy:
Person A:judged less competentPerson B:judged more competent↓B receives authority over A
But then:
Who determines B’s competence?
And:
Who determines which preferences count as irrational?
Even if this produces benefits in carefully defined medical cases, extending it broadly into ordinary adult life creates significant liberal-democratic risks.
This is exactly why Conly herself worried that paternalism might seem to establish a superior class governing an inferior one. Cofnas acknowledges that concern but does not, in my view, adequately solve it. (ResearchGate)
The paper’s most troubling passage
Near the end, Cofnas suggests that people with very low IQ who face severe risks such as uncontrolled addiction or homelessness could, in extreme circumstances, be required to live under supervision.
He also acknowledges that because test-score distributions differ among demographic groups, such policies could fall disproportionately on some groups. His response is essentially that a genuinely paternalistic intervention is supposed to benefit its recipient rather than burden them. (PhilPapers)
I think this response is philosophically inadequate.
Saying:
“The coercion benefits you”
doesn’t make the coercion harmless.
People can lose:
liberty
privacy
dignity
opportunity
political equality
even when the state sincerely believes it is benefiting them.
Historically, paternalistic institutions have often described coercive interventions as being for the target’s own good.
So “beneficial intent” cannot itself settle whether there is an adverse effect.
Indeed, that’s precisely what needs to be demonstrated.
A much stronger version of Cofnas’s proposal
I think the paper becomes considerably more defensible if you replace:
Person faces serious decision↓Can person understand it?↓Can person appreciate consequences?↓Can person reason about alternatives?↓Can person communicate stable preferences?↓YES → respect autonomyNO↓Can education/support restore capacity?↓YES → provide supportNO↓Is catastrophic harm imminent?↓YES → consider least restrictive intervention
IQ might provide background evidence about which kinds of information need simplification or assistance.
But IQ would not determine whether someone’s liberty is taken away.
That approach is much closer to contemporary medical capacity practice. (PubMed Central (PMC))
This also changes what “paternalism” looks like
Instead of:
high intelligence↓controls↓low intelligence
we get:
specific decisional impairment+serious foreseeable harm+failed less-restrictive alternatives↓temporary, targeted intervention
That’s a much easier principle to defend.
And importantly:
It applies equally to a person with IQ 75 and a person with IQ 145.
The question becomes:
“Can you competently make this decision?”
not:
“How intelligent are you?”
Cofnas versus Mill
Question
Mill
Cofnas
Who normally knows someone’s interests best?
The individual
Often the individual, but not always
Competent adults
Strong presumption of sovereignty
Competence varies continuously
Bad choices
Adults have a right to make them
Some may result from inability to reason adequately
Government coercion for self-harm
Generally illegitimate
Sometimes legitimate
Cognitive differences
Limited political importance
Potentially relevant
Main danger
Tyranny and social conformity
Preventable self-destruction
Cofnas versus Thaler/Sunstein versus Conly
Thaler/Sunstein
Conly
Cofnas
Problem
Cognitive biases
Cognitive biases
Biases + differences in reasoning ability
Intervention
Nudges
Sometimes coercion
Sometimes targeted coercion
Choice preserved?
Usually
Not necessarily
Not necessarily
Are paternalists cognitively superior?
Not necessarily
Emphasizes shared fallibility
Sometimes, in relevant domains
Main justification
Help choices better reflect preferences
Prevent grave self-harm
Same, including ability-related errors
Where the evidence leaves us
I would divide Cofnas’s argument into three layers.
Layer 1: Psychological
Cognitive ability varies continuously.
Very strong.
Layer 2: Empirical
Those differences predict meaningful differences in the ability to handle complicated health, financial and practical information.
Strong. Later evidence on health and financial literacy reinforces this. (ScienceDirect)
Layer 3: Political
Therefore the government may coercively restrict the decisions of less cognitively able adults for their own benefit.
Much weaker.
The first two propositions don’t establish the third.
You still need a normative theory specifying:
how much autonomy is worth
what counts as catastrophic harm
how competence is determined
what evidence proves incapacity
whether assistance could replace coercion
how false positives are handled
who exercises paternalistic authority
how abuse is prevented
Cofnas largely inherits the crucial normative premise from Conly rather than independently establishing it.
The deepest philosophical issue
The paper forces a real tension between autonomy and welfare.
Imagine two cases.
Case 1
A competent adult says:
“I know this activity carries a 25% chance of killing me. I accept that risk.”
Most liberals say:
Let him decide.
Case 2
An adult says:
“There’s no risk.”
But he only says that because he literally cannot understand the information showing the 25% mortality risk.
Now autonomy becomes less clear.
What exactly are we respecting?
Case 1:informed preferenceCase 2:preference produced by inabilityto understand consequences
Cofnas is right that these are morally different cases.
And I think that’s the strongest insight of the paper.
Complete laissez-faire autonomy becomes harder to defend when someone’s “choice” is inseparable from an inability to comprehend what he is choosing.
But that insight does not imply that low intelligence generally licenses political control.
The legitimate target would have to be specific decisional incapacity, not cognitive rank as such.
My overall assessment
I would score the major claims roughly like this:
Claim
Assessment
Cognitive ability varies continuously
Very strong
These differences matter outside school
Very strong
Lower ability can make complex health/financial decisions harder
Strong
Traditional competent/incompetent models can conceal important variation
Strong
Universal cognitive bias isn’t the only possible basis for paternalism
Strong, conditional on accepting paternalism
IQ itself is an adequate measure of decisional competence
Weak
Lower cognitive ability generally warrants stronger state paternalism
Weak
Severe, demonstrable inability to understand a catastrophic decision may justify intervention
Much more plausible
Disproportionate coercion ceases to be problematic because it is intended to benefit recipients
Unconvincing
So I would modify Cofnas rather substantially.
His best argument is not:
“Less intelligent people should have less freedom.”
It is:
The liberal concept of competence should not pretend that every adult above a minimal disability threshold has identical capacities for every decision. Some adults genuinely cannot understand particular high-stakes choices, and in extreme cases this may weaken the normal presumption against paternalistic intervention.
That is a serious argument.
But the safest and philosophically stronger implementation is:
NOT:IQ↓degree of liberty
but:
specific decision+actual ability to understand+actual ability to appreciate+actual ability to reason+magnitude of harm↓strength of autonomy presumption
IQ may help predict where difficulties are more likely. It should not itself decide who gets autonomy.
The paper in one paragraph
Nathan Cofnas argues that paternalism theory has focused too narrowly on universal cognitive biases. Humans also differ substantially in general cognitive ability, and these differences affect how well people can understand complicated health, financial and practical information. If Sarah Conly is correct that severe self-destructive errors can sometimes justify coercive paternalism, Cofnas argues that it should not matter whether the error results from a bias everyone shares or from an individual’s limited ability to evaluate complex evidence. He uses adult literacy, health-management and related evidence to show that the traditional philosophical division between fully competent adults and a small incompetent minority understates real cognitive variation. The empirical part of the paper is fairly strong: cognitive ability is continuous and predicts many forms of real-world reasoning. The normative conclusion is much less secure. IQ is not equivalent to decision-making capacity, statistical differences cannot determine whether a particular individual understands a particular choice, and coercion creates risks involving institutional error, competing values, dignity and abuse. Cofnas therefore provides a strong challenge to simplistic accounts of equal cognitive competence, but a much weaker argument for intelligence-based restrictions on equal liberty. A more defensible conclusion is that paternalistic intervention may sometimes be warranted when a person demonstrably lacks the capacity to understand a specific catastrophic choice, after less restrictive forms of assistance have failed, regardless of that person’s overall IQ. (Cambridge University Press)
Citation: Nathan Cofnas, “Coercive Paternalism and the Intelligence Continuum,” Behavioural Public Policy 4, no. 1 (2020): 88–107. DOI: 10.1017/bpp.2018.4. Cambridge Core article
The Kevin MacDonald debate
What MacDonald claims
Kevin MacDonald’s “Judaism as a group evolutionary strategy” theory presents Jewish religious and social institutions as an adaptation that promotes in-group genetic and political interests. In its later political form, the theory alleges that disproportionately influential Jews advanced liberalism, multiculturalism, mass immigration, and anti-hereditarian social science partly to weaken non-Jewish—especially white—group cohesion.
Cofnas’s alternative: the default hypothesis
Cofnas argues that ordinary causes explain Jewish overrepresentation in intellectual and political movements:
historically high literacy and urban concentration;
barriers and opportunities that directed talent into particular occupations;
high average cognitive ability in some Jewish populations;
exclusion from overtly antisemitic right-wing movements, which mechanically skews visible participation toward the left;
individual ideology, professional networks, and ordinary political incentives rather than a coordinated ethnic strategy.
He calls this a default hypothesis because it begins with general mechanisms that would be invoked for any unusually urban, literate, cognitively selected, and historically excluded minority. A specifically Jewish evolutionary-strategy explanation needs incremental evidence beyond those mechanisms.
The exchange, in order
“Judaism as a Group Evolutionary Strategy: A Critical Analysis of Kevin MacDonald’s Theory” (Human Nature, 2018): argues that MacDonald misrepresents sources, selects confirming cases, and does not establish unusually high Jewish ethnocentrism or coordinated group strategy.
MacDonald replies that the default hypothesis fails and that Cofnas ignores asymmetric Jewish support for liberal immigration and multiculturalism. The reply was later retracted, which is an editorial event, not a logical refutation by itself.
“The Anti-Jewish Narrative” (Philosophia, 2021): Cofnas separates and tests three claims—unusual Jewish ethnocentrism, Jewish responsibility for liberal/blank-slate movements, and antisemitism as a reaction to Jewish wrongdoing—and argues that none is adequately supported.
“Still No Evidence for a Jewish Group Evolutionary Strategy” (Evolutionary Psychological Science, 2023): responds to later versions of the critique, including immigration, and maintains that the evidence does not show Jews acting as an unusually coordinated ethnic interest group.
The author’s debate page also collects the intervening essays, MacDonald’s replies, Cofnas’s rejoinders, a defense by Edward Dutton, slide decks, commentary about censorship, and a response to Jean-François Gariépy.
Full linked exchange inventory
The author-curated page places the materials in this order. Repeated appearances of the same peer-reviewed paper are consolidated here but the whole argumentative sequence is retained.
Sequence
Item
Function in the exchange
1
Cofnas, Judaism as a Group Evolutionary Strategy (Human Nature)
Original peer-reviewed critique
2
Cofnas and Jonathan Anomaly, Quillette article on MacDonald and the alt-right
Popular overview
3
MacDonald’s first reply
Challenges the default hypothesis
4
Cofnas’s comments on MacDonald’s response
Non-peer-reviewed clarification
5
MacDonald’s second reply
Continues dispute over evidence and examples
6
Cofnas, Genetic Literacy Project article on censorship and MacDonald’s appeal
Argues suppression makes bad theories more attractive
7
MacDonald’s GLP response
Restates group-strategy case
8
Cofnas’s GLP rejoinder
Popular counter-response
9
Edward Dutton’s defense of MacDonald in Evolutionary Psychological Science
Argues MacDonald’s theory remains plausible
10
Cofnas’s peer-reviewed response to Dutton
Tests “plausibility” against evidence and alternatives
11
MacDonald’s response to the Dutton paper
Further defense
12
Cofnas’s University of Buckingham presentation slides
Compressed presentation of the case
13
Cofnas’s response to Jean-François Gariépy
Addresses a prominent hereditarian defender/critic
14
Cofnas, The Anti-Jewish Narrative (Philosophia)
Adds immigration and a three-claim audit
15
MacDonald, The “Default Hypothesis” Fails to Explain Jewish Influence
Peer-reviewed reply, later retracted
16
Cofnas, Quillette protest against threatened retraction
Defends publication of MacDonald’s reply despite rejecting it
17
Cofnas, Still No Evidence for a Jewish Group Evolutionary Strategy
Final and most comprehensive peer-reviewed response
The secondary exchanges add examples and reveal each side’s burden-of-proof assumptions, but they do not materially overturn the four-paper core. Cofnas himself recommends the final 2023 paper if only one item is read.
What survives scrutiny
Strongest findings
MacDonald’s theory often treats Jewish participation in a movement as evidence of Jewish group interest without specifying the expected base rate.
It selectively counts left-liberal Jewish actors while discounting Jewish conservatives, libertarians, anti-communists, restrictionists, and defenders of free speech.
It moves between individual motive, aggregate overrepresentation, cultural effect, and evolved group strategy without showing the causal links.
MacDonald’s account is difficult to falsify: Jewish disagreement can be redescribed as strategic pluralism, while contrary cases are dismissed as insufficiently Jewish.
Historical Jewish overrepresentation does not by itself identify coordination, ethnocentric intent, or evolutionary adaptation.
Real criticisms of Cofnas’s reply
MacDonald-aligned critics raise some questions that are methodologically legitimate even when embedded in polemics:
Cofnas’s “default hypothesis” is more a family of possible ordinary explanations than a fully estimated quantitative model.
The level of Jewish overrepresentation should be compared across movements and periods, not illustrated with a handful of counterexamples.
Biography, assimilation, religiosity, and strength of Jewish identification matter when classifying a figure as evidence for or against group strategy.
High intelligence and urbanity do not automatically explain the direction of a political preference.
These limitations do not rescue MacDonald’s theory. They identify where Cofnas’s alternative needs better measurement. MacDonald’s theory still bears the larger burden because it posits a group-specific, evolved, interest-coordinated mechanism.
Verdict
Cofnas wins the central evidentiary dispute. His work does not prove that Jewish interests never influence Jewish organizations or individuals; no serious account should require that. It shows that MacDonald has not demonstrated an unusual, evolved, coordinated strategy that explains liberalism, immigration policy, or antisemitism. This is an important resource for Jewish Conspiracy Theories because it criticizes MacDonald from inside evolutionary psychology and hereditarian discourse rather than dismissing the thesis by label alone.
Race, intelligence, and free inquiry
What the 2020 peer-reviewed paper actually argues
“Research on Group Differences in Intelligence: A Defense of Free Inquiry” is primarily a methodological and ethical defense of research, not a report of a new dataset that discovers a genetic racial gap. It makes several distinct claims:
Natural selection and population history make perfectly identical distributions of all intelligence-related variants among geographically separated populations improbable.
Standard intelligence tests measure a consequential, partly heritable trait at the individual level.
Some behavioral-genetic evidence is consistent with genetic contributions to measured group differences.
Research on the question faces unusual reputational and institutional pressure.
Therefore, scholars should be free to test hereditarian and environmental hypotheses.
The paper explicitly recognizes two points that many summaries omit:
High within-group heritability does not logically imply that a between-group difference is genetic.
Research favoring both hereditarian and environmental positions has appeared in prominent journals; his claim is relative stigma, funding, and publication difficulty, not the literal nonexistence of dissenting work.
Where the evidence is solid
Standardized cognitive tests predict educational, occupational, and some health outcomes, although prediction is probabilistic and not destiny.
Individual differences in intelligence are substantially heritable within many studied populations.
Human genetic variation contains geography-correlated population structure; self-identified ancestry labels often correlate with such structure in particular national samples.
Average scores differ across socially defined racial and ethnic groups in many datasets.
Environmental interventions and social conditions can change cognitive performance, development, and measured gaps.
What is not established
The proportion of a contemporary Black–White, African–European, or other socially defined average test-score gap caused by inherited genetic differences.
A transportable polygenic score that can compare average intelligence-related genetic value across broad racial groups without major population stratification, ancestry portability, phenotype, environmental, and gene–environment problems.
That a folk racial label denotes a discrete natural kind with one uniform cognitive distribution.
That current representation in elite institutions is a direct measure of genetically determined merit.
That genetic causation, even if eventually established in part, would determine an individual’s rights, moral worth, or appropriate treatment.
Recent reviews report that intelligence polygenic scores predict a modest portion of within-sample individual variance, while emphasizing incomplete causal understanding and weak portability across ancestries. The international PRIMED consortium similarly warns that race, ethnicity, and genetic ancestry are not interchangeable and that cross-population score transfer remains methodologically difficult: genetics of intelligence review; principles for transferring polygenic scores. A 2024 PNAS critique argues that current genetics does not establish substantial racial genetic differences in intelligence and explains how population stratification, indirect genetic effects, and environmental amplification can mimic direct causal signal: “Genes, culture, and scientific racism”.
The leap in the popular essays
By 2024, Cofnas no longer presented hereditarianism merely as a permissible hypothesis. In “A Guide for the Hereditarian Revolution”, he calls “race realism” the “scientifically correct position,” assigns stylized mean IQs to continental racial populations, predicts that Black people would nearly disappear from high-profile positions under meritocracy, and makes elite acceptance of hereditarianism the foundation of an anti-DEI political program.
That conclusion is stronger than his 2020 paper’s evidentiary case. A defensible chain would need direct, replicable molecular evidence about the causal variants and their effects across ancestries and environments. Cofnas instead moves from:
individual heritability + average group gaps + ancestry structure + limits of environmental explanations
to:
a substantially genetic group-gap explanation is scientifically settled.
The first line does not logically entail the second. It may motivate research; it does not close the question.
What his Substack adds—and what it has not answered
Cofnas’s Substack is indispensable for understanding the strength, scope, and political function of his current position. It is not equivalent to his peer-reviewed work. The 2020 article defends the legitimacy of inquiry and assembles indirect considerations; the later essays treat hereditarianism as sufficiently secure to ground an institutional strategy. Reading the two genres together reveals a shift from “this hypothesis must be researchable” to “this conclusion must reorganize elite opinion.”
Race is ancestry in ordinary language. He argues that traditional American racial classifications track broad genetic clusters with high accuracy and that fuzzy boundaries no more invalidate race than dialect continua invalidate language. This supports the modest proposition that ancestry-correlated population structure is real. It does not show that every census label is a stable biological unit, that ancestry and race are interchangeable in every society, or that any particular psychological difference is genetic.
Culture is partly downstream of population traits. Cofnas rejects culture as an independent alternative to genes and proposes that genetically influenced average intelligence and personality help generate a population’s norms. This is a possible gene–culture feedback hypothesis, but the essay does not estimate its parameters or distinguish it empirically from social inheritance, institutional persistence, migration selection, or environmental amplification.
Selected migrants should regress toward an ancestral mean. His Nigerian-immigrant example predicts that children of positively selected migrants will score below their parents because some of the parents’ advantage is nonheritable. Regression after selection is real, but the numerical example assumes the ancestral mean and heritability it is invoked to illustrate. Without measured parental genotypes, direct and indirect genetic effects, assortative mating, and comparable environments, it is not an independent estimate of a racial genetic gap.
Elite conversion is the political lever. He proposes an information campaign to persuade intellectual elites, combined with institutional replacement and legal reform. Hereditarianism is therefore not incidental to his politics: it supplies the empirical premise for accepting large demographic disparities as compatible with meritocracy.
What this contributes
The essay supplies a much clearer account of Cofnas’s population concept, gene–culture model, migration argument, and political strategy than the peer-reviewed free-inquiry paper. It also makes the evidentiary vulnerability clearer. Ancestry structure, heritable individual differences, and regression after selection can all be real without identifying the genetic share of a socially defined group-mean difference.
empirical premise: races and sexes have the same innate distributions of socially relevant traits;
moral premise: people have an equal right to realize their potential;
political conclusion: persistent disparities create a moral emergency to find and correct environmental injustice.
He calls the empirical premise false and argues that anti-DEI victories will remain temporary until a critical mass of elites accepts hereditarianism. This sharpens his definition of wokeism and explains why he considers race-and-IQ research politically load-bearing.
The same essay also contains an important concession: heritability is relative to a population and environment, and high heritability does not mean that a trait is genetically determined or unresponsive to environmental change. That is methodologically sound. But Cofnas does not apply the concession symmetrically to his principal racial inference. A population-relative estimate of individual variance still does not calculate why two population means differ.
Do not attribute the comment section to Cofnas
Readers in the comments under “Beating Woke with Facts and Logic” discuss Sasha Gusev’s polygenic-score critique, Russell Warne’s “five converging lines of evidence,” missing heritability, and African genetic diversity. These are relevant objections and defenses, but they are predominantly reader comments, not Cofnas’s authored reply. His visible author responses in the thread do not provide a technical answer to the cross-ancestry, stratification, or indirect-effect problems. A comment appearing under his post—and even a comment marked as liked by him—should not be cited as his considered argument.
Has he answered the recent genetics papers?
Public-response audit — checked 27 August 2026
No dedicated Cofnas Substack post or publicly indexed Note was located that names and answers any of the following papers. Searches covered the paper titles, authors, DOI, and the terms polygenic-score portability, cross-ancestry, population stratification, and indirect genetic effects. Because Substack does not expose every deleted comment or ephemeral interaction equally well, the defensible conclusion is “no identifiable public paper-specific response located,” not “he has never mentioned them anywhere.”
Discusses intelligence heritability and polygenic evidence generally, but does not answer the review’s points about modest variance explained, incomplete causal understanding, and indirect effects. The review itself principally concerns individual differences and does not directly adjudicate the racial-gap question.
Acknowledges that cross-ancestry scores are less valid, then relies on a broader indirect case. PRIMED is not an intelligence-gap study; its relevance is that it explains why a score trained in one ancestry cannot straightforwardly rank genetic propensities in another.
Rejects similar claims that race is merely arbitrary and that culture is independent of genes, but does not answer their specific account of stratification, genetic nurture, cultural inheritance, and environmental amplification. Of the three, this paper most directly rejects a Cofnas-style racial-genetic inference.
The absence of a direct response matters because Cofnas’s general arguments do not automatically resolve the papers’ methodological objections. His strongest reconstructable reply would be that imperfect polygenic scores do not erase behavioral-genetic, adoption, migration, psychometric, and evolutionary evidence, and that several individually inconclusive lines can favor hereditarianism when combined. That is a fair reconstruction of his position, but it remains an inference from his broader writing, not a rebuttal he has published to these three papers.
The 2026 West Eurasian selection Note
In an August 2026 Substack Note, Cofnas points to the Reich-lab ancient-DNA paper as evidence of selection related to intelligence in West Eurasians and argues that acknowledging a relationship between race and intellect should not make a source disreputable. This shows how he currently answers critics in compressed form: human populations experienced different selection, cognitive traits are heritable and polygenic, and population cognitive differences are therefore scientifically plausible.
The cited study establishes neither the direction nor the magnitude of contemporary Black–White, African–European, or other racialized cognitive-test differences. Evidence that combinations of alleles associated with cognitive performance changed in ancient West Eurasia demonstrates that recent human selection occurred; it does not perform the modern cross-population causal comparison on which Cofnas’s political program depends.
Substack verdict
Cofnas’s Substack materially improves the reconstruction of what he believes and why he thinks it matters. It does not supply the decisive genomic evidence missing from the interviews and peer-reviewed paper. The strongest accurate description is: his later writing converts an unresolved but scientifically permissible hypothesis into a confident political premise, while answering the principal molecular-genetic objections indirectly or not at all.
The Harvard 0.7% claim
Cofnas says that a color-blind Harvard admitting “only by academic qualifications” would be 0.7% Black, then extrapolates that Black professors would approach zero and Black people would disappear from nearly all high-profile occupations outside sports and entertainment.
The first number was not invented. It comes from a 2013 Harvard Office of Institutional Research counterfactual, disclosed during the affirmative-action litigation and reported by Heather Mac Donald in The New Criterion. But the interpretation is much weaker than Cofnas suggests:
It is a model of an academic-index-only admissions rule, not a measurement of genetically determined merit.
Academic credentials reflect schooling, family resources, neighborhood, health, discrimination, culture, selection into the applicant pool, and individual choices as well as ability.
Harvard never uses only grades and test scores; athletes, legacies, extracurricular achievement, recommendations, personal ratings, geography, and institutional priorities also enter.
The model does not establish what employers, professions, or the whole economy would look like under a general meritocracy.
After the Supreme Court restricted race-conscious admissions, Harvard’s reported Black share fell from 18% to 14%, not to 0.7%, although the university cautioned that several cycles would be needed to understand the full effect: Harvard’s official Class of 2028 data.
Verdict: the 0.7% figure is traceable, but the genetic and society-wide conclusions drawn from it are not contained in the study.
Dedicated Comedy Cellar interview audit — 1 October 2026
Is “Race Realism” Simply Racism? Nathan Cofnas Under Fire fact-checks the complete 2h27m interview, corrects its opening account of Jason Arday’s resignation and death, traces the Harvard model and every substantive identifiable study, article, book, and media source mentioned, and separates four propositions too often collapsed together: ancestry structure, measured group averages, causal genetic attribution, and political or moral conclusions. It finds Dworman’s normative cross-examination substantially stronger than his scientific adjudication, while finding Cofnas’s move from individual heritability and population structure to a settled racial-genetic cause unsupported by the cited evidence.
Professor Dave/Hossenfelder video audit — 30 August 2026
Sabine Hossenfelder: More Exposed Than Ever! (Racism Edition) audits Professor Dave’s treatment of the Arday controversy, Ghent and Emmanuel actions, Hossenfelder’s defense of Cofnas, the race–IQ literature, and the video’s broader neuroscience, AI, and Peter Thiel claims. It confirms Cofnas’s explicit hereditarian program and Hossenfelder’s intervention while finding that the video repeatedly converts satire, preliminary allegations, association, and temporal sequence into quotation, guilt, coordination, and causation.
His theory of “wokism”
Focused 2026 interview synthesis
The focused interview synthesis below reconstructs Cofnas’s complete race–IQ–DEI argument from his August 2026 TRIGGERnometry, Gad Saad, and Coleman Hughes interviews. It separately audits IQ validity, individual heritability, the Minnesota adoption study, intervention fadeout, regression toward group means, cross-ancestry polygenic scores, the 2026 Reich-lab ancient-DNA paper, meritocracy, limited quotas, his institutional strategy, and his proposed boundary between elite knowledge and mass discussion. The central finding is that several premises are well supported, while the claimed majority-genetic share of broad racial IQ gaps remains an inference rather than a directly measured result.
Cofnas’s definition in his own words
Wokeism as the consequence of denying group differences
“My basic idea is that wokeism is what follows from the denial of group differences. If you believe that all groups of people innately have the same distribution of socially relevant traits—so a person born in the Congo is exactly the same as a person born in Japan, exactly the same as a person born in Germany, on average—and any differences in outcome among these societies therefore must be due to the environment, because you’re ruling out genes based on some moral or whatever principle.
“And what follows from that? Differences must be due to the environment. And the most obvious environmental explanation is white racism or the legacy of white racism, particularly when it comes to lower performance by groups that were historically persecuted—for example, Africans. If you are looking for an environmental explanation for why Black people or people of African descent, either in Africa or in Western countries—in most cases; there are some exceptions, for example, in the United Kingdom African immigrants do very well because of selective immigration—but in general that’s the trend, and the most obvious explanation is white racism.
“Wokeism is simply the project of recognizing a moral emergency to fight the racism and bring about conditions where all groups achieve at equal rates—to have equal outcomes—because then we will know that the environment has finally been equalized. Until we confront the false empirical claim that is the root of wokeism, then it will be impossible to get rid of wokeism. I think wokeism is a very destructive ideology, as we have seen. That’s why I think it’s important.”
— Nathan Cofnas, TRIGGERnometry interview, approximately 37:52–40:09; lightly edited for punctuation and transcript false starts
Why he thinks understanding IQ differences matters
Cofnas is not presenting the race-and-IQ question as an isolated scientific curiosity. In his political model, the causal interpretation of average cognitive and other socially relevant group differences is load-bearing:
If groups are assumed to have identical innate distributions, persistent differences in performance must be attributed entirely to environments.
For historically persecuted groups, racism or its legacy then becomes an especially salient environmental explanation.
Outcome parity becomes evidence that environments have finally been equalized, while continuing disparity becomes evidence of continuing injustice.
DEI, affirmative action, speech regulation, and other equalizing interventions appear morally necessary under that diagnosis.
Therefore, Cofnas believes anti-woke institutional reform cannot endure unless elites reject what he regards as the false empirical premise of identical innate group distributions.
This explains why he treats IQ research as politically important: he believes the explanation of cognitive-performance gaps determines whether demographic disparities are interpreted as evidence of discrimination and whether equal outcomes become a policy imperative. In his framework, establishing a partly genetic cause would weaken the inference from disparity to racism and, consequently, the moral rationale for many equalizing programs.
Attribution and evidentiary boundary
This is the internal logic of Cofnas’s argument, not an empirical conclusion adopted by this note. An environmental explanation is not automatically reducible to racism; racism can be real without explaining every disparity; and evidence for some genetic contribution would not establish its magnitude or disprove specific instances of discrimination. Conversely, uncertainty about genetic causation does not justify treating every disparity as proof of bias. The key analytical task is to distinguish observed gaps, their causal decomposition, and the independent moral premises used to design policy.
Cofnas defines wokism as the conjunction of:
an equality thesis—racial and sex groups have the same innate distributions of socially important psychological traits; and
a moral equality premise—unequal outcomes require correction or identification of injustice.
On his account, persistent disparities plus the equality thesis force elites to search for hidden discrimination. DEI, speech regulation, and bureaucratic equalization follow. He therefore thinks legal reforms associated with Richard Hanania or institutional counter-mobilization associated with Christopher Rufo cannot last unless elites first accept hereditarianism.
What this model explains
Why disagreement about causes of disparities has unusually high political stakes.
Why some institutions treat unexplained disparities as presumptive evidence of discrimination.
Why hereditarian claims threaten the moral narrative of some DEI programs.
Why legal reforms may not change the culture of institutions whose leaders independently endorse egalitarian outcomes.
What it compresses or misses
Progressive egalitarianism does not require literal equality of every innate distribution; many egalitarians ground equal rights in personhood, dignity, anti-caste rules, uncertainty, or institutional humility.
DEI has multiple causes: civil-rights law, litigation risk, organizational imitation, professional incentives, elite social networks, demographic change, moral ideas, partisan sorting, and reputational markets.
A disparity can justify investigation without proving discrimination.
Even genetically influenced traits can be responsive to environments and institutions.
Hereditarianism does not settle the normative design of admissions, employment, welfare, education, or rights.
This creates a productive disagreement with The Origins of Woke - Richard Hanania: Hanania gives law and institutional incentives causal priority; Cofnas gives elite beliefs about human equality causal priority. The mechanisms can reinforce each other, and neither source isolates its claimed primary cause with a decisive research design.
Focused interview synthesis: Race, IQ, Wokeness, and DEI
Merge provenance
This section preserves the complete substantive body of the former child note Nathan Cofnas - Race, IQ, Wokeness, and DEI (created and reviewed Thursday, August 27th 2026). That note pointed up to this profile and used the source-status label transcript-attributed-with-primary-science-audit. Its aliases, interview URLs, related links, tags, and categories are now merged into this note’s frontmatter.
Bottom line
Across these three interviews, Nathan Cofnas presents one unified scientific-political argument: average differences among ancestry-linked populations—including differences in measured cognitive ability—are partly genetic; suppressing or denying this possibility produces an “equality thesis”; persistent disparities then appear to prove hidden discrimination; and this perceived moral emergency produces wokeness and DEI. He therefore wants a critical mass of elites to accept hereditarianism, restore substantially more meritocratic selection, and contest control of universities and other institutions. The Coleman Hughes exchange adds a crucial qualification: Cofnas does not advocate indiscriminate mass promotion of race-and-IQ claims. He favors informed acceptance among decision-making elites alongside a “mild taboo” against rude or inflammatory public discussion.
The interviews contain several sound or defensible premises: cognitive tests measure a real and socially predictive construct; individual differences in intelligence are substantially heritable in many studied populations; human populations show ancestry-correlated genetic structure; environments affect cognitive development and test performance; and race categories are fuzzy, admixed, and internally heterogeneous. The interviews do not establish Cofnas’s central causal conclusion that inherited genetic differences explain most average racial IQ gaps. Cofnas concedes that current polygenic scores transfer poorly across ancestries, supplies no direct molecular decomposition of a Black–White gap, and identifies himself as a philosopher rather than a population geneticist. His strongest contribution here is an explicit, testable model of anti-DEI politics; his weakest move is treating indirect and ambiguous evidence as if it settled the genetic share of group differences.
Scope and attribution
This note deliberately excludes the Jason Arday allegations, media campaign, and questions of responsibility for Arday’s death. It extracts only Cofnas’s claims about intelligence, human population differences, wokeness, DEI, meritocracy, academic institutions, and the political right.
The transcripts are automated and contain spelling errors, duplicated material, and occasional speaker ambiguity. Timestamps are therefore locators, not independently authenticated quotations. “Cofnas argues” means the position expressed in these interviews or his linked writing, not a finding accepted by the note.
Executive synthesis
Cofnas’s position in ten propositions
General intelligence is real and consequential. IQ tests are imperfect and culturally conditioned, but they capture reasoning differences that predict education, occupation, health, accidents, and other outcomes.
Intelligence is substantially heritable within studied populations. Heritability varies by age, population, and environment; Cofnas cites this as background for his group-level hereditarianism.
Human populations are biologically structured but not discrete. Racial categories resemble languages or dialects: fuzzy at the borders, affected by classification interests, yet anchored in real patterns of ancestry and variation.
Average cognitive-test differences among socially racialized groups are observable. In the first interview he gives stylized means for African Americans, White Americans, East Asians, Hispanics, and Ashkenazi Jews.
He believes genetic differences explain a nontrivial—and probably majority—share of some group gaps. This is his definition of hereditarianism about group differences.
He regards standard environmental explanations as inadequate. He appeals to adoption studies, intervention “fadeout,” regression toward group means, repeated cross-national rankings, migration selection, and recent evidence of human genetic selection.
He denies that present genomics has directly solved the racial-IQ question. He accepts that European-derived polygenic scores lose predictive validity across ancestries, but thinks the total indirect case nevertheless supports hereditarianism.
He defines wokeness through the “equality thesis.” If groups have equal innate distributions but unequal outcomes, morally serious people will infer remediable environmental injustice and pursue equalization.
He treats DEI as the institutional expression of that inference. In his usage, DEI is not merely broad recruitment; it is the use of different standards and demographic balancing under a narrative of current or inherited injustice.
His political answer is an elite hereditarian realignment. He wants anti-woke scholars to remain inside major institutions, displace ideological gatekeepers, normalize discussion of group differences, and restore meritocracy—while allowing that limited representation-based preferences or quotas may sometimes be justified.
What is actually Cofnas’s “finding”?
Philosopher and synthesizer, not the author of a decisive genomic study
The interview title calls Cofnas a “race scientist,” but his relevant peer-reviewed article is principally a philosophical defense of free inquiry. It does not report a new genome-wide association study, representative IQ survey, adoption experiment, or causal decomposition of a racial gap. Cofnas explicitly says in the first interview that population genetics is not his specialty and that he cannot supply the requested genetic details (approximately 79:24–79:40).
It is therefore more accurate to distinguish:
original contribution: an ethical and epistemological case for permitting research on population differences;
synthesis: his interpretation of psychometrics, behavioral genetics, adoption studies, intervention research, and evolutionary theory;
political theory: his equality-thesis model of wokeness and his proposed “hereditarian revolution”;
empirical claim not originally demonstrated by him: that most of a broad racial IQ gap is caused by inherited genetic differences.
His own 2020 paper says researchers still debate genes versus environment and that the question had not been definitively settled. That is narrower and more cautious than the confidence expressed in these 2026 interviews and in his post-2023 political essays: Cofnas, “Research on Group Differences in Intelligence: A Defense of Free Inquiry”.
The race-and-IQ argument
1. What he means by hereditarianism
At approximately 37:48–38:39 in the TRIGGERnometry interview, Cofnas defines group-difference hereditarianism as the view that inherited genetic differences make a nontrivial contribution to average differences in socially important traits such as intelligence. When asked for a magnitude, he says he thinks the genetic contribution should be a majority of the difference.
That claim must be kept separate from three weaker propositions:
genes contribute to individual differences in intelligence within a population;
allele frequencies and ancestry patterns differ among populations;
observed group means differ in some samples.
All three weaker propositions can be true without establishing the causal genetic share of a between-group mean difference. A high within-group heritability is a ratio of variances under particular conditions; it does not identify why two groups have different means. Cofnas himself acknowledges that heritability depends on a particular population and environment (approximately 39:01–39:11).
The cumulative indirect case
Cofnas’s strongest reconstruction
Cofnas does not rest the group-gap conclusion on within-group heritability alone, and he accepts that current genomics has not directly decomposed broad racialized IQ gaps. He thinks the combined indirect evidence—adoption studies, regression patterns, cross-national results, intervention fadeout, migration selection, evolutionary considerations, and population structure—nevertheless warrants a hereditarian conclusion. The intended logic is convergent and abductive: each line is individually defeasible, but he argues that their conjunction fits a partly genetic model better than an environment-only model.
Line of indirect evidence
How it functions in Cofnas’s case
What it can reasonably support
Principal limitation and status
**[[#The Minnesota study
Adoption studies]]**
If children raised in comparatively advantaged adoptive homes later move toward the average score associated with their racialized birth population, shared rearing alone appears insufficient.
Transracial adoption can test simple claims that the adoptive household fully determines later cognitive scores. The Minnesota pattern is compatible with some inherited contribution.
**[[#Regression toward a group mean
Regression patterns]]**
Children of extreme-scoring parents are expected to be less extreme; Cofnas emphasizes reports that high-scoring Black and White parents’ children regress toward different observed group means.
Different regression points can be consistent with groups having different underlying distributions, including a model with inherited differences.
Cross-national and migrant-group results
Recurring score rankings across countries, generations, or migrant settings are treated as difficult to explain through one country’s discrimination, schools, or history.
Repetition across genuinely different environments can challenge a narrow explanation tied to one local institution and can motivate a broader causal model.
Tests, samples, languages, schooling, health, age structures, migration streams, and racial categories are rarely equivalent. Migrants are often positively or negatively selected, and countries do not provide interchangeable environments. No single specified cross-national dataset in the interviews carries the general claim. Potentially informative, but asserted too broadly.
**[[#The intervention-fadeout claim
Intervention fadeout]]**
When an early program raises measured IQ and the gain later shrinks, Cofnas interprets the reversion as evidence that environmental enrichment does not permanently eliminate an underlying difference.
Fadeout rebuts the strongest claim that any early score gap can be durably erased by a short intervention. It shows that immediate test gains need not equal permanent changes in a latent trait.
Migration selection
Positively selected migrants may score well above the source population; their children are predicted to regress toward an assumed ancestral mean as the parents’ nonheritable advantage dissipates. Cofnas uses Nigerian migration as an illustration in “A Guide for the Hereditarian Revolution”.
Regression after selecting unusually high-scoring parents is a real quantitative expectation, and multigenerational data could in principle test competing inheritance models.
In Cofnas’s numerical example, the ancestral mean and heritability are assumptions used to generate the predicted result. Without parental genotypes, representative source data, direct versus indirect genetic effects, assortative mating, and comparable environments, migration regression is not an independent estimate of a racial genetic gap. It substantially overlaps the preceding regression argument. Illustrative model, not a completed causal test.
**[[#What the 2026 Reich-lab paper shows
Evolutionary considerations]]**
Intelligence-related traits are heritable and polygenic; human evolution continued after populations dispersed; different ecologies and subsistence systems could therefore have imposed different cognitive selection pressures.
Recent selection and population-specific adaptation make a categorical claim of biological impossibility untenable. The 2026 ancient-DNA study supports selection on combinations of alleles associated in modern samples with cognitive performance within West Eurasia.
**[[#4. His population concept
Population structure]]**
Because ancestry can be inferred from many loci and allele frequencies vary geographically, Cofnas argues that fuzzy racial boundaries do not make all population-level genetic differences impossible.
It establishes a necessary background condition: human populations have real, overlapping, ancestry-correlated genetic structure, so population differences in complex traits cannot be ruled out a priori.
Why seven lines do not automatically equal seven independent confirmations
Combining evidence is legitimate when the observations are at least partly independent, when a hypothesis predicts them better than its competitors, and when the same hidden assumption is not doing the work repeatedly. Cofnas is right that the failure of any single decisive experiment does not force agnosticism if several weaker indicators genuinely converge. A cumulative case can rationally change confidence before a complete molecular decomposition exists.
The difficulty is that these seven lines are not fully independent:
adoption outcomes, parent–child regression, and migration selection all rely substantially on regression across generations and can inherit the same assumptions about baseline group means;
cross-national recurrence and migration results reuse observed group rankings whose comparability is itself disputed;
evolutionary selection and population structure establish possibility and background mechanism, not the trait-specific effect needed for the modern gap;
intervention fadeout concerns the durability of particular treatments, not the original causal partition between genes and environments;
environmental and mixed models can predict several of the same observations through selection, measurement, social inheritance, health, schooling, cultural transmission, or gene–environment correlation.
No interview or popular essay assigns likelihoods to the competing models, estimates how correlated the seven indicators are, or shows that their conjunction yields a reproducible genetic percentage. “Totality of the evidence” is therefore a defensible description of Cofnas’s reasoning, but not yet a formal cumulative model.
Calibrated verdict on the combined case
The bundle is stronger than the invalid syllogism IQ is heritable + groups differ → the gap is genetic. It makes a partly genetic contribution a scientifically permissible and biologically plausible hypothesis, and some readers may reasonably treat the circumstantial pattern as increasing its probability. It does not currently identify a unique genetic pathway, establish that the contribution is nonzero, or justify Cofnas’s stronger claim that inherited differences probably explain a majority of broad racialized IQ gaps. Several evidentiary lines are compatible with hereditarianism without being uniquely predicted by it. Direct, representative, cross-ancestry and within-family designs would be needed to separate direct genetic effects from population structure, genetic nurture, selection, and measured environments.
Murray and Cofnas: different strengths of conclusion
In their contemporary defense of The Bell Curve, Murray and Herrnstein’s 1994 position was that a completely zero-genetic contribution appeared improbable while the relative genetic and environmental mixture remained unknown; they described themselves as “resolutely agnostic” about the proportions. Cofnas advances the stronger claim that the cumulative indirect case already supports a nontrivial and probably majority genetic contribution, despite agreeing that present polygenic evidence has not directly solved the question. The distinction is between a probabilistic suspicion with no assigned magnitude and a politically load-bearing causal conclusion.
That causal difference should be kept separate from their contemporary disagreement about human worth. In their September 2026 debate, Murray distinguishes the usefulness and predictive power of intelligence from a person’s fundamental worth and rejects globally ranking people by IQ. Cofnas’s reply argues instead that intelligence is one component of worth and that differences in intelligence can partly make some people more valuable than others.
Murray’s worth position does not resolve the empirical cause of group-score gaps, and Cofnas’s worth position does not follow from psychometric reliability, heritability, or predictive validity. A scientific description of cognitive variation supplies no premise ranking legal rights, dignity, or the global value of persons. The updated Farina–Cofnas–Murray source audit keeps the causal and normative disputes separate.
2. His view of IQ tests
Cofnas rejects the idea that IQ tests are merely tests of elite cultural trivia. He emphasizes pattern recognition, spatial rotation, and abstract reasoning, and he argues that scores correlate with consequential life outcomes (approximately 40:18–43:40).
He also makes an important concession: a culture-free test is impossible. Test familiarity, schooling, background concepts, language, and a culture of formal classification affect performance. His narrower claim is that, among people exposed to modern mass schooling, scores retain meaningful individual-level predictive validity.
This is broadly consistent with the mainstream psychometric view that cognitive test scores are reliable, show a hierarchical structure including a general factor, and predict educational, occupational, social, and health outcomes. The correlation does not imply that IQ alone causes each outcome, nor does it mean scores measure total human worth: Deary, Cox, and Hill, “Genetic variation, brain, and intelligence differences”.
3. The group means he reports
In the first interview, Cofnas states approximately:
Population label used in the interview
Mean he reports
Evidence status in this note
African American
82–85
Attributed estimate, not a timeless population constant
White American
100
Norm-referenced comparison point
East Asian
A few points above 100
Broad and sample-dependent generalization
Hispanic
Intermediate, in the 90s
Heterogeneous umbrella category
Ashkenazi Jewish
About 110–112
Contested estimate; he rejects still higher small-sample claims
The durable fact is that average standardized-test differences appear in many datasets. The precise size depends on the test, norming year, age, sampling frame, schooling, immigration selection, language, socioeconomic context, and category definition. Cofnas’s assertion that a roughly 15-point Black–White IQ gap has not narrowed is too categorical. Direct analyses have reached conflicting estimates, while several standardization studies report meaningful twentieth-century narrowing. A prominent 2006 analysis estimated a 4–7 point relative gain by Black Americans between 1972 and 2002: Dickens and Flynn, “Black Americans Reduce the Racial IQ Gap”. A later cohort analysis by Charles Murray found earlier narrowing but possible stalling or slight reversal in later cohorts, illustrating why one fixed number should not be treated as a biological constant: “The magnitude and components of change in the Black–White IQ difference”.
4. His population concept
Cofnas explicitly rejects homogeneous, genetically pure, sharply bounded races. He distinguishes African immigrants in Britain from African Americans and Appalachians from White residents of Massachusetts. His language analogy is that populations can be distinguishable even though they overlap, blend, and admit mixed cases (approximately 36:44–37:35 and 88:28–90:56).
This avoids the crude claim that folk races are discrete subspecies. It does not solve the inferential problem. The National Academies recommends that geneticists not use race as a proxy for genetic variation, not assign ancestry labels from race, and avoid assumptions of group homogeneity, hierarchy, or stability: Using Population Descriptors in Genetics and Genomics Research. For a cognitive claim, the appropriate unit must be specified by sampling and genetic ancestry rather than imported from a broad census label.
Evidence Cofnas invokes—and what it can bear
Cofnas’s evidentiary move
What supports it
What the interview does not establish
Audit
Intelligence is heritable.
Twin, family, adoption, and molecular studies find substantial heritability of individual differences in many populations; estimates often rise with age.
The cause of a mean difference between racialized groups.
Strong premise; invalid if used alone for the group-gap conclusion.
IQ predicts life outcomes.
Large literatures connect cognitive scores with education, work, health, and mortality.
That every association is causal or that IQ exhausts merit, character, creativity, or dignity.
Strong descriptive premise with causal and normative limits.
The same group ranking appears across settings.
Some cross-national and migrant data show recurring score patterns.
Comparable sampling, measurement invariance, environmental equivalence, or absence of immigration selection.
Requires a defined dataset; asserted too generally in the interview.
Early IQ gains fade.
Some early-childhood programs show fadeout on measured IQ.
That environmental intervention has no durable cognitive or life-course effects.
Partly true, but the universal negative is false or unproved.
Transracial adoption favors hereditarianism.
Adoptee scores changed with age and differed by social classification in the Minnesota study.
A clean genetic estimate: placement age, pre-adoption experience, selective placement, attrition, identity treatment, and small samples complicate inference.
Ambiguous, not decisive.
Regression to racial group means proves biology.
Children of extreme-scoring parents tend to be less extreme.
A genetic racial cause; regression occurs statistically and can reflect measurement error, assortative mating, selection, and persistent group environments.
Not a stand-alone genetic test.
Polygenic evidence will eventually decide the question.
Intelligence is highly polygenic, and within-ancestry scores predict some individual variance.
A valid cross-ancestry comparison of average genetic cognitive potential.
A 2026 ancient-DNA study found selection in West Eurasians on combinations of alleles that now predict cognitive performance.
That modern continental racial IQ gaps are genetic, or that the past selected phenotype was “IQ” as currently measured.
Real study; politically overextended.
Winter and farming selected for higher intelligence.
Planning demands can generate plausible evolutionary hypotheses.
Direct causal evidence tying climate or subsistence mode to present racial test-score gaps.
Speculation, which Cofnas partly acknowledges.
Heritability and age
Cofnas is on firm ground that intelligence is substantially heritable in many studied populations and that estimates can rise from childhood into adulthood. A major review summarizes estimates increasing from roughly 20% in infancy to as high as 80% in later adulthood, while also emphasizing that environmental influence remains because heritability is never 100%: Plomin and Deary, “Genetics and intelligence differences”.
But heritability is not biological immutability. It can rise in a relatively uniform environment because environmental variance shrinks; it can also incorporate gene–environment correlation, where people partly select or evoke environments correlated with their traits. It is not a percentage of an individual’s intelligence “caused by genes,” and it does not automatically transport between populations.
The intervention-fadeout claim
Cofnas says intensive early interventions raise childhood IQ but that scores return toward a group mean by early adulthood, then states that no intervention currently known has lasting effects (approximately 50:21–52:27 and again 67:01–67:13 because the transcript repeats this segment).
The randomized Abecedarian Project reported long-lasting cognitive and academic effects, although magnitudes and interpretations remain debated: Campbell and Ramey, follow-up study.
Even when an initial IQ effect fades, educational attainment, health, employment, crime, or other life outcomes can remain improved. Fadeout on one test is not equivalent to “the environment did nothing.”
The scientifically safer conclusion is: some measured IQ gains fade, some educational exposures have durable effects, and intervention results do not by themselves identify the genetic share of a racial mean gap.
The Minnesota study
The interview inaccurately calls the Minnesota Transracial Adoption Study a study of “twins reared apart.” It was a follow-up of transracial adoptees. Cofnas interprets the fall in adoptee scores toward racialized population means as evidence consistent with hereditarianism.
The study’s authors did not regard it as a clean genetic demonstration. Their abstract says rearing children in the culture of the tests and schools benefited performance, and their later published response described the results as providing little or no conclusive evidence for genetic causes of racial differences. Cofnas’s own 2020 paper records that conclusion and the study’s confounds: Weinberg, Scarr, and Waldman, “The Minnesota Transracial Adoption Study”.
The study is relevant evidence, but its design cannot isolate genes from correlated pre-adoption experience, placement timing, visible-race treatment, adoptive-family effects, attrition, and classification. Presenting it as decisive is stronger than its design permits.
Regression toward a group mean
Cofnas argues that high-scoring Black parents’ children regress toward a Black mean and high-scoring White parents’ children regress toward a White mean, and he presents this as “impossible” under an environmental account (approximately 74:14–76:41).
That is an overclaim. Regression toward the mean is expected whenever parent scores include transient environmental effects and measurement error and parent–child correlation is below one. Different observed group means can generate different empirical regression points without proving why those means differ. A causal hereditarian test would require genetically informative models that separately measure ancestry, parental phenotype, direct genetic effects, indirect parental-genetic effects, assortative mating, and environments. The regression pattern can be evidence in a larger model; it is not proof by itself.
Polygenic scores: his most important concession
When the hosts say that no identified variants explain racial IQ gaps and that European-derived polygenic scores do not reliably transfer across populations, Cofnas agrees that cross-ancestry prediction often becomes less valid (approximately 76:45–78:28). He then says brain-expressed variants differ in frequency among populations but cannot identify the relevant details and calls the conclusion uncontroversial.
The concession matters more than the assertion. Variants can differ in frequency and be expressed in the brain without producing a known average cognitive effect. Current polygenic scores face ancestry portability problems because of linkage disequilibrium, allele frequencies, population structure, social context, indirect genetic effects, and gene–environment relations. The PRIMED consortium review treats race, ethnicity, and ancestry as noninterchangeable and explains why a score trained in one population cannot simply rank another: Kachuri et al., “Principles and methods for transferring polygenic risk scores”.
Therefore:
genes affect individual cognitive variationdoes not yet yielda valid genetic decomposition of broad racial mean differences.
What the 2026 Reich-lab paper shows
Cofnas accurately identifies a recent paper from a David Reich-led team. Using 15,836 ancient West Eurasian genomes, it reports directional selection over the last ten millennia, including changes in combinations of alleles that in modern industrial samples predict measures of cognitive performance.
The paper is evidence that human evolution did not stop in the remote past. It is not a test of African–European or Black–White IQ gaps. The authors explicitly warn that the modern association measures come from industrialized societies and that their relationship to phenotypes favored in the past remains unclear: Akbari et al., “Ancient DNA reveals pervasive directional selection across West Eurasia”.
Intergenerational deprivation without “biological memory”
When pressed about intergenerational deprivation, Cofnas answers that there is no biological memory of what happened several generations earlier and regards long-lasting epigenetic explanations skeptically (approximately 72:54–73:58).
The narrow epigenetic caution is reasonable: dramatic claims of many-generation human epigenetic inheritance often outrun evidence. But intergenerational environmental persistence does not require a biological memory. Wealth, neighborhood, school quality, exposure to toxins and violence, parental education, social networks, expectations, health, and institutional treatment can be reproduced socially and ecologically. A PNAS perspective by Lala and Feldman explains how cultural and ecological inheritance can stabilize differences without genetic transmission and reiterates why within-group heritability does not identify between-group causes: “Genes, culture, and scientific racism”.
Cofnas’s theory of wokeness
The causal chain
flowchart LR
A["Equality thesis<br/>same innate group distributions"] --> C["Persistent group disparities"]
B["Moral premise<br/>people should realize their potential"] --> D["Disparity becomes a moral emergency"]
C --> D
D --> E["Search for environmental injustice"]
E --> F["Systemic-racism explanations"]
F --> G["DEI, affirmative action,<br/>speech and institutional controls"]
H["Cofnas's intervention:<br/>elite acceptance of hereditarianism"] -.-> A
H --> I["More meritocratic selection<br/>and institutional counter-capture"]
Across the interviews, Cofnas says wokeness follows from the denial of inherited group differences. The Saad interview sharpens the point: he says it is not legislated and not a conspiracy; elites sincerely regard persistent disparities as an emergency because they believe groups are innately alike (approximately 32:15–32:54). In the Coleman Hughes interview, he makes the corresponding communication claim explicit: the hereditarian conclusion need not dominate cable news or elementary education, but it must be understood by the “cognitive elites” who set cultural and institutional agendas (approximately 36:46–40:14 and 88:28–89:23).
This is a real theoretical advantage over crude “secret cabal” accounts. Cofnas does not need coordinated bad faith. His mechanism is decentralized moral reasoning:
assume equal innate distributions;
observe unequal outcomes;
infer correctable environmental causes;
treat failure to correct them as injustice;
expand institutional interventions.
What the model explains well
Why disputes about the causes of disparities become morally charged.
Why some institutions treat an unexplained disparity as presumptive evidence of bias.
Why legal color-blindness does not necessarily end demographic monitoring or outcome equalization.
Why hereditarian claims are perceived not as isolated statistics but as threats to an entire political-moral framework.
Why anti-woke reforms may be unstable if leaders remain committed to demographic parity.
Why DEI can persist without corruption, conspiracy, or conscious hostility toward merit: sincere actors may believe differential treatment is needed to neutralize unequal starting conditions.
Where the model is too compressed
The false binary
Cofnas repeatedly presents two choices: accept substantial genetic group differences, or interpret disparities as pervasive injustice. That excludes mixed, local, and uncertain explanations. A disparity can reflect some combination of selection, migration, age structure, class, education, family conditions, institutions, discrimination, culture, measurement, and biology. It can justify investigation without serving as proof of any one cause.
The model also conflates three questions:
logical implication: what follows if someone literally believes identical innate group distributions;
historical causation: why modern DEI institutions arose when and where they did;
normative policy: what governments and organizations ought to do under causal uncertainty.
Even if Cofnas’s syllogism describes some activists, it does not show that the equality thesis is the primary historical cause of civil-rights law, professional HR bureaucracies, litigation risk, university politics, demographic coalitions, social-media incentives, or organizational imitation. This is where The Origins of Woke - Richard Hanania supplies a competing institutional mechanism: Hanania emphasizes law and incentives, while Cofnas makes elite beliefs causally prior. Both mechanisms can operate together.
Equality of worth is not equality of traits
Equal citizenship, equal protection, and moral dignity do not require the empirical premise that every ancestry or sex group has an identical distribution of every trait. Conversely, a biological contribution to a group average would not determine a person’s rights or worth. Cofnas sometimes recognizes this separation when he advocates individual treatment, but his theory often treats egalitarian morality as if it logically commits one to identical group potentials.
This produces an is–ought gap:
different average distributionsdoes not entailunequal rights, group-based treatment, or indifference to remediable disadvantage.
The Coleman Hughes debate: truth, social peace, and elite knowledge
The Coleman Hughes interview, published 27 August 2026, supplies the most sustained moral and political challenge to Cofnas’s project. Their disagreement is not simply over whether group differences exist. Hughes grants the legitimacy of population-genetic inquiry and defends Cofnas’s right to research and speak. The dispute concerns a second-order question:
The governing question
When a proposition may be true but is easily converted into stereotypes, resentment, or unequal treatment, who should discuss it, how publicly, and for what political purpose?
This separates three issues that are often collapsed:
freedom of inquiry: may researchers investigate the proposition without professional punishment?
epistemic warrant: how strong is the evidence, and with what confidence should the proposition be believed?
ethics of communication: even if warranted, should it be actively publicized, quietly known by specialists, taught broadly, or surrounded by norms of restraint?
Hughes and Cofnas largely agree on the first. Their sharp disagreement is about the second and especially the third.
Cofnas’s two-tier communication norm
Cofnas explicitly rejects the idea that every uncomfortable truth deserves publicity merely because it is true. He says some questions should be researchable without being broadcast; race realism need not be taught in elementary schools or debated on cable news. His intended audience is the “cognitive elites” who set cultural agendas, staff institutions, and influence policy. Near the end of the exchange, he accepts that a mild taboo—a norm that makes public discussion impolite without making research punishable—could encourage civic peace (approximately 36:46–37:46, 88:28–89:23, and 93:26–94:15).
His preferred arrangement is therefore:
open technical inquiry ↓acceptance among informed decision-makers ↓policy and institutional expectations adjusted to group disparities ↓public norm of individual treatment and restrained group talk
This is more cautious than the description “mainstream race realism” suggests. Cofnas is advocating elite epistemic conversion, not continuous mass racial broadcasting.
It also introduces a tension. His Substack essays and podcast appearances are publicly accessible; the political program requires enough advocacy to change elite opinion; and elite beliefs that shape admissions, hiring, and public policy cannot be fully insulated from democratic scrutiny. Hughes asks how elite knowledge could be prevented from “spilling out.” Cofnas offers an etiquette norm rather than an institutional mechanism. The resulting position combines free inquiry with a form of epistemic paternalism: specialists and decision-makers may need to believe consequential claims that ordinary citizens should be discouraged from emphasizing.
The elite-knowledge problem
Cofnas identifies a plausible difference between permission to know and prudence in broadcasting. But an elite-only settlement raises its own moral questions: Who counts as sufficiently informed? Who decides when the evidence has crossed the threshold for policy relevance? How can citizens evaluate policies whose empirical rationale is supposed to remain socially muted? And how does a “mild taboo” avoid becoming the stronger professional censorship he opposes?
Hughes’s false-binary objection
Hughes rejects Cofnas’s recurring suggestion that Western societies face a forced choice between hereditarianism and woke disparity-as-injustice reasoning. He identifies at least two additional possibilities: cultural or mixed explanations, and a colorblind civic settlement that emphasizes fair processes while deliberately reducing the political salience of group outcomes.
Possible settlement
Core premise
Political risk
Woke disparity model
Persistent group disparity presumptively indicates remediable injustice.
Perpetual grievance, demographic balancing, and collective blame.
Cofnas’s elite hereditarianism
Elites accept that inherited average differences substantially help cause disparities while citizens are treated individually.
Stereotyping, elite paternalism, and possible erosion of equal citizenship.
Culture or plural causation
Outcomes can differ because of norms, history, selection, institutions, class, discrimination, and other interacting causes without requiring one master explanation.
Cultural blame can itself become essentializing, and causal uncertainty may not satisfy political demands.
Colorblind process liberalism
Teach equal citizenship, police discrimination, build fair procedures, and avoid treating demographic parity as the test of justice.
May become unstable if large disparities persist and people continue to demand causal and moral explanations.
Cofnas replies that conservative institutions have promoted culture-first explanations for decades without winning over intellectual elites. When Hughes asks for evidence that thoughtful people overwhelmingly reject the cultural account, Cofnas concedes that he has no polling data and relies on his experience of universities and public discourse (approximately 45:45–50:59). That weakens the claim that the failure of the cultural strategy is an established empirical fact. It may accurately describe elite discourse; the interview does not demonstrate why the discourse took that form or whether a plural-causal, process-focused liberalism is inherently unstable.
Competing models of social harm
The deepest disagreement is consequentialist. Each thinks the other’s communicative settlement predictably produces racial conflict:
Hughes’s predicted harm from public hereditarianism
Cofnas’s predicted harm from suppressing hereditarianism
People convert overlapping distributions into categorical stereotypes.
People interpret persistent disparities through environmental wrongdoing.
Children and adults use group averages to humiliate or prejudge individuals.
Historically advantaged groups become objects of inherited blame, while disadvantaged groups acquire grievance narratives.
Tribal cognition overwhelms the logical distinction between an average and an individual.
DEI expands because every remaining disparity becomes evidence that the environment is still unjust.
Historical acceptance of racial hierarchy suggests that empirical hierarchy can erode political equality.
Woke racial narratives already intensify resentment, conflict, and differential treatment.
Cofnas makes a significant concession: the schoolyard and stereotyping harms Hughes describes are real possibilities and count as reasons against broad dissemination. He argues that they must be balanced against the harms of teaching that environmental injustice explains group outcomes (approximately 57:35–60:14). This turns the dispute into a comparative-risk problem rather than a conflict between truth and cowardice.
Neither side supplies the evidence necessary to resolve that comparison. A serious test would need to distinguish the effects of:
private specialist inquiry;
elite consensus without public campaigns;
neutral education about distributions and individual prediction;
inflammatory racial messaging;
disparity-focused antiracist instruction;
colorblind civic instruction;
institutional confidence that selection procedures are fair.
The harms may also be asymmetric across contexts. Technical discussion among geneticists is not equivalent to partisan broadcasting; elementary education is not graduate education; and an admissions committee has different informational duties from a television pundit.
Average differences and equal citizenship
Cofnas’s normative defense is liberal-individualist: accepting average group differences does not logically justify denying rights because every group contains a wide distribution and the full range of human ability. He notes that societies recognize large individual differences in intelligence without making disenfranchisement of below-average citizens a mainstream position. A racial mean cannot determine the ability or rights of a particular person (approximately 75:23–76:41).
Hughes accepts the logic but questions the psychology. Humans are not ideal statisticians; tribalism magnifies cognitive bias; and historical societies often paired beliefs about inherited group hierarchy with unequal citizenship. He therefore treats foreseeable misuse as part of the ethics of public communication. Cofnas responds that the historical association requires more careful study and points to multiethnic empires and civic incorporation, while conceding that a responsible communicator must consider how subtle statistical claims will be simplified (approximately 76:45–82:03).
Two different questions
Logical question: Do average differences entail unequal rights or justify judging an individual by race? No. Cofnas is correct on this point.
Empirical-political question: Will large numbers of people nevertheless use such claims to rationalize stereotyping or unequal treatment? Possible and historically important, but not quantified in the interview. Hughes’s concern cannot be dismissed merely by restating the logic.
Is colorblindness a stable equilibrium?
Hughes defends the late-twentieth-century colorblind civic norm: teach equal treatment, reduce racial fixation, enforce fair processes, and allow outcomes to differ without converting every disparity into a grievance. He invokes transparent competition in sports, music, and comedy as cases where people tolerate demographic imbalance because they regard selection as comparatively meritocratic (approximately 60:49–64:28). This is less an environmental theory of every disparity than a proposal to lower the political salience of causal decomposition once procedures are sufficiently fair.
Cofnas replies that the postracial settlement was temporary because many supporters expected nondiscrimination to produce convergence. When legal equality, outreach, and affirmative action did not deliver equal outcomes, people who retained the equality thesis intensified their search for hidden injustice. In his account, the Great Awokening was the delayed crisis of colorblind liberalism’s unmet empirical expectation (approximately 64:30–67:13 and 69:57–71:05).
Hughes offers a competing historical mechanism: smartphones and algorithmic social media amplified emotionally divisive police videos and racial narratives around 2012–2013. Cofnas accepts social media as a catalyst but argues that the underlying moral concern and disparity-focused ideology long predated it. These explanations are compatible:
long-standing egalitarian expectations +persistent visible disparities +legal and bureaucratic incentives +social-media amplification ↓the Great Awokening
The interview therefore improves the earlier audit. Cofnas’s equality thesis may explain why a disparity narrative was morally resonant, while social media helps explain why it intensified at a particular historical moment. Law, organizations, and partisan sorting remain additional mechanisms.
Moral and political verdict
What the Hughes exchange establishes
The interview narrows the disagreement. Cofnas and Hughes both defend research freedom, equal citizenship, individual rather than racial judgment, and some restraint in public communication. Cofnas thinks elite acceptance of hereditarianism is necessary to prevent environmental explanations from generating collective blame and DEI. Hughes thinks that normalizing inherited racial hierarchy—even among elites—risks leaking into stereotypes, tribal conflict, and unequal treatment, while a fair-process colorblind settlement remains possible.
Cofnas wins the narrow logical point that unequal group averages do not entail unequal rights. Hughes identifies the harder political problem: logical compatibility does not guarantee psychological or institutional stability. Cofnas, in turn, identifies the symmetrical danger that suppressing one causal hypothesis does not produce neutrality; it can privilege an injustice narrative with its own collective harms. The unresolved question is comparative and empirical: which communication and institutional regime best protects truth, equal citizenship, social trust, and fair treatment under causal uncertainty?
His Locke genealogy
In the Saad interview, Cofnas traces the equality thesis and denial of inherited individual or group differences to John Locke in 1690 (approximately 37:00–37:45). Locke is an important ancestor of empiricism and theories of education, but calling him the origin of “race denial” is anachronistic and monocausal. Early modern natural law, Christian universalism, abolitionism, liberal citizenship, Enlightenment moral psychology, anthropology, civil-rights politics, and twentieth-century social science all contributed different elements. The interview offers a genealogy, not a demonstrated intellectual history.
DEI, meritocracy, and institutional politics
What he means by DEI
Cofnas distinguishes benign outreach from DEI. In his vocabulary, searching broadly for qualified applicants is not the problem; DEI means applying different standards or demographic preferences under the premise that disparities are products of injustice. This is a stipulative political definition, narrower and more condemnatory than the range of practices commonly grouped under “DEI.”
That distinction is analytically useful because “DEI” can refer to unlike things:
nondiscrimination compliance;
outreach and recruitment;
accessibility and accommodation;
demographic preferences or targets;
disparate-impact monitoring;
ideological training;
bureaucratic offices and reporting systems.
A valid evaluation must identify the mechanism rather than infer it from the label.
Meritocracy versus representation
The most revealing political exchange occurs at approximately 94:33–101:23. Cofnas says his lodestar is individual merit and that society benefits when the best-qualified person receives a job. But he also recognizes representation and visible participation as legitimate values. He says a balance may be required and expresses openness to limited quotas, offering a hypothetical 3% Black share at Harvard rather than proportional representation.
This creates a substantive tension:
Stated position
Practical implication
“No DEI”
Reject the injustice narrative and large demographic equalization programs.
Some representation has independent value
Permit a limited identity-based thumb on the scale.
Treat people as individuals
Avoid inferring a candidate’s ability from a group average.
Elite institutions would show extreme disparities under meritocracy
Prepare the public to accept unequal demographic outcomes.
The hosts correctly observe that limited representation preferences resemble a milder form of what many institutions call DEI. Cofnas’s reply is primarily semantic and justificatory: he would not call it DEI because it is not grounded in a claim of racial injustice. The practical difference is therefore one of magnitude, rationale, and limits, not necessarily the complete absence of identity-conscious selection.
“Retaking” institutions
Cofnas rejects retreat into a separate dissident ecosystem. In the Saad interview he says anti-woke scholars should maintain a physical presence in important institutions, “muscle” their way back in, and not cede universities to left-wing activists (approximately 30:26–32:15). Cultural change must precede durable institutional reform, but personnel and power still matter.
His strategy has four parts:
protect inquiry into taboo empirical questions;
persuade a critical mass of cognitive elites;
replace or disempower ideological gatekeepers;
align admissions, hiring, and public narratives with a hereditarian understanding of disparities.
This is not a theory of neutral institutions. It is a theory of counter-elite capture after an epistemic conversion. The tension is that Cofnas condemns ideological personnel control by the left while endorsing politically organized personnel change by the anti-woke right. His defense is that one project suppresses truth and the other restores it; that defense succeeds only to the extent that his disputed empirical premise is actually established and the new gatekeepers retain open inquiry.
The “right’s stupidity problem”
Cofnas argues that intellectual elites lean left, that people identifying as very liberal score higher than very conservative White respondents on the General Social Survey’s abbreviated vocabulary measure, and that the right has failed to build or maintain serious universities and cultural institutions (approximately 108:40–113:29).
He is appropriately cautious at one point: he says no definitive study proves the whole thesis. The supporting evidence is suggestive, not dispositive:
WORDSUM is a short vocabulary measure, not a comprehensive IQ battery.
“Left” and “right” bundle social, religious, economic, populist, and institutional dimensions.
occupational selection may send cognitively able conservatives into business, engineering, law, or medicine rather than academia.
university ideological imbalance can reflect preferences, networks, discrimination, geography, credential pipelines, or institutional culture as well as average ability.
His deeper political claim is more interesting than the insult: mainstream conservatism accepts moral equality and usually avoids hereditarian group claims, so it lacks an explanation of persistent disparities that can compete with the woke account among highly educated elites. Whether hereditarianism is the necessary remedy depends on the scientific and moral objections developed above.
Rejection of an ethnic-conspiracy explanation
The Saad interview places Cofnas’s anti-woke theory beside his critique of Kevin MacDonald. Cofnas argues that Jewish overrepresentation in left-wing movements can be explained by general factors—high representation in cognitively demanding fields, political selection away from openly antisemitic movements, and selective attention to left-wing Jews. He denies evidence for a coordinated Jewish evolutionary strategy to undermine non-Jewish society (approximately 39:38–44:34).
This matters for his theory of wokeness. He offers a belief-and-incentive model, not a Jewish conspiracy model. Elite overrepresentation, sincere egalitarian belief, and institutional diffusion can explain ideological convergence without ethnic coordination. See the fuller source audit in Nathan Cofnas > The Kevin MacDonald debate.
Argument audit
Strongest parts
IQ nihilism is not a serious answer. Cognitive tests measure something real, reliable, and predictively useful.
Biology is relevant to human behavior. It is not scientific to exempt human psychology from all evolutionary and genetic inquiry.
High heritability and environmental malleability coexist. Cofnas acknowledges both, even though he later understates the environmental evidence.
Population boundaries can be fuzzy and real. Admixture and clines do not make all population structure imaginary.
Suppression is not refutation. Moral anxiety about a hypothesis does not determine its truth.
Disparity-to-discrimination reasoning is a genuine political mechanism. Some DEI arguments do move too quickly from unequal outcomes to systemic bias.
Conspiracy is unnecessary. Sincere beliefs and institutional incentives can coordinate behavior without a secret plan.
He states the tradeoff. Unlike many polemicists, Cofnas admits that representation and merit can conflict and that he values both.
He distinguishes inquiry from indiscriminate publicity. In the Hughes interview, he accepts that foreseeable stereotyping is morally relevant, that not every truth should be broadcast, and that a mild norm of public restraint may be legitimate.
Weakest parts
The central causal magnitude is asserted, not measured. No direct molecular evidence in the interviews shows that genes explain most of a broad racial IQ gap.
He slides between individual and group inference. Individual heritability, group means, and ancestry structure are repeatedly combined as if their conjunction identified the cause of a mean gap.
His categories change scale. African Americans, sub-Saharan Africans, African immigrants, Hispanics, East Asians, Ashkenazi Jews, and local White populations are treated at different levels of demographic resolution.
He overstates intervention failure. IQ fadeout in some programs does not negate durable educational or life-course effects.
He overreads ambiguous designs. The Minnesota adoption study and regression-to-the-mean patterns do not uniquely support a genetic explanation.
He invokes genomic evidence outside its scope. Selection within ancient West Eurasians is not evidence for the genetic cause of modern Black–White differences.
His evolutionary story is speculative. Winter and farming explanations lack the direct evidence needed for his political confidence.
The wokeness model is monocausal. It explains a line of reasoning, not the full historical rise, institutional form, or variation of DEI.
His policy conclusion exceeds the science. Meritocracy, limited quotas, and institutional purges require independent moral and institutional arguments.
The interview ends at the falsification question. At 115:00 the hosts ask what evidence would convince him that group differences are environmental rather than partly genetic; the supplied transcript contains no answer. That leaves the most important test of his epistemic openness unresolved.
His elite-only solution is under-theorized. He wants decision-makers to absorb a politically consequential doctrine while mass discussion remains muted, but does not explain how this arrangement would remain transparent, democratically accountable, or contained.
He treats the failure of culture-first conservatism as established without measuring it. When Hughes requests evidence that thoughtful elites reject the cultural account, Cofnas supplies no polling data. The claim may describe his experience, but it is not demonstrated in the interview.
A recurring epistemic asymmetry
When evidence is consistent with hereditarianism, Cofnas often treats it as affirmative support; when evidence is environmentally suggestive, he emphasizes confounds, fadeout, or the absence of a single decisive experiment. The standards should be symmetrical. Adoption data, group rankings, polygenic scores, historical score changes, and interventions all require the same questions:
What observation was predicted in advance?
What competing model predicts it too?
What confounders were measured?
What result would count against the preferred hypothesis?
How much of the gap—not merely whether “some” genetic effect exists—is estimated?
Without those answers, “totality of the evidence” can become a label for an intuition rather than a reproducible causal model.
What evidence would move the debate?
Evidence that would strengthen Cofnas’s causal claim
large, representative, cross-ancestry cognitive datasets using measurement-invariant tests;
within-family genomic designs that separate direct genetic effects from population stratification and genetic nurture;
polygenic scores validated within multiple ancestries and environments before between-population comparison;
admixture and local-ancestry studies with preregistered predictions and direct measures of socioeconomic, educational, health, and discrimination exposures;
replicated estimates of a group-gap genetic share that remain stable across tests, ages, cohorts, and environments;
prospective evolutionary predictions rather than post hoc climate or subsistence stories.
Evidence that would weaken it
large shifts in group gaps over time or context that are too rapid for genetic change;
durable gap reduction following improvements in schooling, health, nutrition, safety, pollution, or family resources;
cross-ancestry genetic predictors that fail to reproduce the alleged ordering after portability problems are solved;
within-family direct-effect estimates substantially smaller than population-level associations;
admixture studies showing little or no ancestry-linked cognitive effect after well-measured environments and selection are addressed;
failure of hereditarian models to make better out-of-sample predictions than environmental or mixed models.
The key unresolved question
Cofnas defines hereditarianism as a nontrivial genetic contribution but says he believes the contribution is probably a majority. Those are different burdens of proof. What evidence fixes the magnitude above 50%, and what result would lower his estimate below that threshold? The interviews never supply a quantitative answer.
Practical conclusions without accepting his full thesis
One need not accept Cofnas’s majority-genetic conclusion to preserve several useful lessons:
Judge individuals by relevant evidence, not group averages.
Do not treat demographic disparity as automatic proof of discrimination.
Investigate discrimination with causal and institutional evidence.
Distinguish race, ethnicity, ancestry, culture, nationality, and migration selection.
Protect legitimate inquiry while demanding ordinary methodological standards.
Separate descriptive group statistics from equal legal status and moral dignity.
Evaluate each DEI mechanism separately—outreach, accommodation, preferences, training, and compliance are not interchangeable.
Recognize both the costs of false accusations of bias and the costs of ignoring genuine exclusion.
Require anti-DEI theories to be falsifiable, not merely contrarian.
Require “meritocracy” to specify the institution’s purpose, the relevant competencies, and the acceptable tradeoffs—not simply equate merit with one test score.
Distinguish the right to investigate a proposition from a duty to promote it; research freedom, evidentiary confidence, public pedagogy, and policy relevance have different thresholds.
Evaluate the harms of both racial stereotyping and collective injustice narratives rather than treating only one direction of error as socially dangerous.
Require policies based on specialized or controversial knowledge to remain publicly explainable; elite expertise does not eliminate democratic accountability.
Transcript map
TRIGGERnometry — The Controversial Race Scientist Who Exposed Jason Arday
Intergenerational environment, epigenetics, adoption, regression to means
76:45–80:32
Absence of identified gap genes, polygenic portability, ancient DNA
80:34–84:30
Evolutionary adaptation, migration selection, winter and farming hypotheses
84:31–90:56
Class differences, within-race heterogeneity, fuzzy population categories
91:01–105:29
Policy, meritocracy, representation, limited quotas, persistence of wokeness
108:21–115:07
Hereditarianism, the right’s intellectual weakness, unanswered falsifier
Transcript defect
The supplied TRIGGERnometry transcript repeats a substantial block after an advertisement—roughly the IQ-test, population-mean, environmental, and intervention discussion. The duplicated text was synthesized once rather than counted as separate support.
Gad Saad — Was It Right to Alert the World About Jason Arday?
Approx. time
Substantive topic retained in this note
20:37–22:10
Forbidden knowledge and the place of biology in social science
30:26–33:00
Retaking institutions; wokeness as elite belief rather than conspiracy
33:03–39:10
“Biophobia,” blank-slate genealogy, and human evolutionary explanation
39:12–44:34
Rejection of MacDonald’s Jewish group-strategy explanation
Coleman Hughes — Nathan Cofnas on Race, IQ, and the Jason Arday Scandal That Made Him Famous
Approx. time
Substantive topic retained in this note
0:00–7:09
Hughes’s opening distinction between research freedom and the prudence of mainstreaming race realism
23:33–30:14
Responsibility, journalism, institutional incentives, proportionality, and Cofnas’s view of Arday as partly a victim
30:30–40:14
Two meanings of race realism; elite knowledge; why Cofnas thinks the equality thesis produces wokeism
40:17–53:57
Hughes’s false-binary and culture objections; Cofnas’s claim that culture-first conservatism failed
54:06–60:14
Social consequences of dissemination; truth, taboo, stereotyping, collective guilt, and comparative harm
60:49–71:34
Colorblind process liberalism, fair procedures, the stability question, and social media as a competing cause of the Great Awokening
72:28–89:23
Rawlsian comparison, historical political equality, overlapping bell curves, tribal cognition, and Cofnas’s elite-only strategy
89:26–94:15
Liberal–conservative IQ claims, political polarization, and Cofnas’s acceptance of a mild public taboo
94:17–101:28
University of Austin, heterodoxy, scientific depth, and the difference between free speech and curricular selection
Focused interview assessment
Cofnas’s interviews are valuable because they reveal the full architecture of his position more clearly than a single label such as “race realist” or “anti-DEI.” He is not saying merely that genes matter, nor merely that universities suppress controversy. He believes a scientific conclusion about average inherited group differences is the indispensable premise of a political realignment. In his account, changing law or replacing DEI offices cannot endure until elites abandon the equality thesis.
That architecture also exposes the vulnerability of the project. If the causal genetic magnitude is unresolved, then the political foundation is unresolved. The interviews demonstrate real knowledge of psychometrics and philosophy of biology, important concessions about culture and population heterogeneity, and a defensible case for inquiry. They do not demonstrate the key percentage on which the “hereditarian revolution” depends. The safest verdict is therefore:
Calibrated verdict
Cofnas identifies real phenomena and asks permissible scientific questions, but he presents an unsettled causal inference as a sufficiently secure basis for sweeping institutional and political change. His equality-thesis model is worth retaining as one explanation of DEI’s moral logic. It should be combined with legal, bureaucratic, economic, demographic, and cultural explanations—and it should not be treated as proof that modern disparities are mostly genetic.
In “Are Smart People Superior?”, Cofnas argues that intelligence contributes to human excellence and some defensible conceptions of worth. He favors a “natural aristocracy” of cognitive and cultural excellence.
This is a normative philosophical claim. Empirical correlations between intelligence and achievement do not prove a hierarchy of moral worth. One can value excellence and select for task competence while preserving equal legal status and refusing to rank the value of persons.
The right’s “stupidity problem”
Cofnas uses General Social Survey WORDSUM results to argue that the contemporary right attracts lower-scoring constituencies, while highly educated and cognitively able people disproportionately enter the left’s institutions. He also says current self-identified race realists tend to score worse than environmentalists and are vulnerable to antisemitism and conspiracy thinking.
Caution: WORDSUM is a short vocabulary measure, not a complete IQ test; self-placement on ideology scales changes over time; education, age, religion, region, and elite sorting confound simple causal stories. The descriptive asymmetry is worth studying, but calling one side “stupid” is polemical compression.
Anti-populism
In “Podcast Bros and Brain Rot,” “MAGA Communism and the End of America,” and related essays, Cofnas attacks protectionism, conspiracism, alternative-media epistemology, and the collapse of expert gatekeeping. He supports markets and argues that competent institutions—sometimes with paternalistic controls—are necessary to filter information.
This makes him a critic of both progressive institutional ideology and the populist right’s alternative information ecosystem.
Academic capture and counter-power
In “How to Take Back Academia,” “Four Reasons Heterodox Academy Failed,” “Philosophy Has Been Hijacked by Woke Twitter,” and “Philosophers Tried to Smear One of Their Own for Gender Heresy,” Cofnas argues that procedural neutrality is insufficient once institutions are ideologically homogeneous. He favors changing personnel and using lawful political power to remove ideological gatekeepers.
The strongest evidence concerns concrete episodes—petitions, attempted retraction, investigations, and institutional statements. The weakest move is treating those episodes as proof that whole disciplines have no meaningful pluralism.
News and criticism audit
Representative coverage makes three different kinds of argument that should not be conflated. Cambridge’s student paper documented the text of Cofnas’s 2024 proposals and quoted academics calling his presence analogous to “a flat earther on the physics faculty”; later reports documented 58 formal complaints, a petition with more than 1,200 signatures, student testimony about harm, and the separation between university and college action: initial Varsity report, complaints and legal-action follow-up. The scholarly critique by Larsen et al. argues that Cofnas mishandles the biological race concept and presents a disputed hereditarian expectation as science. Recent reporting around Jason Arday makes a third argument: that even documentable plagiarism concerns were selected and amplified as part of Cofnas’s racial anti-DEI campaign; The New Yorker, for example, describes Arday as a “hook” for confirming Cofnas’s prior beliefs while still noting that the broader University of Cambridge found no regulatory breach in Cofnas’s own case: “The Real Meaning of the Jason Arday Scandal”.
Each contains a testable core. The first accurately reports his own proposals but sometimes turns “offensive and scientifically disputed” into “therefore no real research question exists.” The second raises genuine methodological objections but partly attacks a discrete-race position Cofnas rejects. The third fairly scrutinizes selection, motive, and proportionality but cannot make independently documented textual overlap disappear. The audit below separates those layers.
Claim in coverage or criticism
What the primary record says
Verdict
“Cofnas is a white supremacist.”
He calls himself a race realist, predicts racial hierarchy under meritocracy, and endorses barriers that preserve racial distinctions. But he explicitly rejects white nationalism, racial hatred, and a state project of white domination.
Not established; usually a polemical label. His actual views are radical enough to criticize precisely.
“He believes Black people are, on average, genetically less intelligent than white people.”
His post-2023 popular essays say this directly at the level of group averages. He does not claim every Black person scores below every white person.
Substantially true, if stated as an average-distribution claim.
“His research proved the racial IQ gap is genetic.”
The 2020 paper is a defense of inquiry and an argument from plausibility/indirect evidence; it presents no new molecular demonstration of the causal share of the gap.
False. This exaggerates what he published.
“He simply denies environment matters.”
The academic paper explicitly discusses environmental hypotheses and the within- versus between-group distinction. Popular essays nevertheless dismiss culture-first accounts with excessive confidence.
False as a description of the paper; partly fair as a criticism of his public rhetoric.
He defines race as ancestry-correlated populations with fuzzy boundaries and accepts clines and admixture. Critics are right that folk labels are heterogeneous and that classification success does not validate cognitive generalization.
Mixed. “Discrete pure races” is a straw man; broad racial extrapolation remains a real problem.
“Scientists have completely debunked all hereditarian questions.”
Individual intelligence is heritable and population structure is real. The unresolved question is the causal decomposition of average group differences.
Too sweeping. Rejection of Cofnas’s certainty is not proof of a zero-genetic causal share.
“There is no censorship because hereditarian papers get published.”
Cofnas himself notes publication on both sides; he alleges relative funding, reputation, and career pressures. His own publication history and Cambridge’s formal inquiry show both access and pressure.
The no-publication rebuttal attacks a stronger claim than he makes. The magnitude of suppression remains unmeasured.
“He advocated apartheid or legal segregation in the 2020 paper.”
The paper defended research and did not propose apartheid. His 2024 political essay later endorsed social barriers preserving races and more homogeneous communities, while rejecting white nationalism.
False about the paper; later writing gives a narrower factual basis for concern about separatism.
“Cambridge fired him because his claims were unlawful or violated university rules.”
Cambridge’s inquiry found no legal or regulatory breach. Emmanuel separately ended an affiliation; the time-limited university fellowship later expired.
Misleading shorthand. Institutional actions must be separated.
“Criticism is only polemic because opponents cannot challenge the science.”
Larsen and colleagues challenge population concepts, causal inference, and the leap from possibility to scientific fact. Current genomic-method papers document portability and confounding problems.
False. Some attacks are polemical, but substantive methodological criticism exists.
The Larsen et al. critique
In “More than Provocative, Less than Scientific” (2020), Rasmus Rosenberg Larsen and colleagues criticize the journal’s decision to publish Cofnas’s defense of intelligence research. Their strongest points are that:
genetic variation is clinal and admixed, making socially broad race labels dangerous inferential shortcuts;
Cofnas moves from the possibility of genetic differences to an expectation that science may confirm them;
he does not adequately treat the social history and misuse of racial intelligence claims.
Their weakest point is the suggestion that “racial realism” necessarily presupposes discrete, immutable, pure races. Cofnas explicitly uses an ancestry-population account that allows fuzzy boundaries. But classification by ancestry and causal explanation of cognitive averages are separate problems, so refuting the discrete-race caricature does not establish Cofnas’s positive hereditarian conclusion. See the published critique.
Were the attempted retractions merely ideological?
Some critics and petitioners did misstate the 2020 paper as advocating segregation or apartheid and called for institutional condemnation before engaging its actual argument. Those are genuine cases of polemical substitution for criticism. The resignation of an editor and attempts to retract a philosophical paper because of its subject also support Cofnas’s narrower claim that reputational pressure affects inquiry.
However, it does not follow that every objection is censorship. Editors can legitimately criticize poor inference, conceptual slippage, one-sided citation, or inadequate treatment of foreseeable harm. The correct audit is claim by claim, not “criticism proves suppression” or “publication proves freedom.”
The Jason Arday episode
What Cofnas alleged
In “DEI Fraud and Cover-Up at Cambridge” (2026), Cofnas accused Cambridge professor Jason Arday of plagiarism, misrepresented qualifications and honorary appointments, and questionable biographical claims. Cofnas framed the case as evidence that DEI standards protected an unqualified Black academic.
What independent evidence supports
Other journalists reported identical or near-identical passages in Arday’s work.
Journals had issued or considered corrections in response to overlap allegations.
Those facts mean the allegations were not fabricated solely because Cofnas had a racial or anti-DEI agenda.
What remains unproved or overextended
There was no completed public adjudication establishing the full scope of research misconduct before Arday’s death.
The plagiarism evidence does not prove that his appointment resulted from racial preference or that similarly situated non-Black academics would have been rejected.
Questions about childhood development, athletics, charity totals, and other biography items require separate evidence; an error or exaggeration in one category does not validate every allegation.
Cofnas’s racial framing and comments about removing Black people from high-status positions make motive and proportionality legitimate subjects of scrutiny, even when some underlying documentary allegations are accurate.
Arday died after an intense media campaign. His family and supporters connected the campaign to severe distress, but the public evidence does not establish a single person’s causal responsibility for his death.
Moral responsibility and proportionality in the Coleman Hughes interview
The Coleman Hughes exchange adds moral nuance that is missing from simple accusations and defenses. Hughes argues that investigative journalism cannot operate under a rule that misconduct must remain unreported whenever exposure might cause severe distress or self-harm. He therefore defends Cofnas’s right to publish documented allegations and rejects an automatic inference that the reporter caused the subject’s death (approximately 23:33–28:20).
Cofnas does not celebrate the outcome or portray Arday as the principal villain. He describes Arday as manipulative and dishonest but says his wrongdoing was comparatively small, that he responded to incentives created by more powerful actors, that the consequences were “totally disproportionate,” and that Arday was to some extent a victim. Cofnas relocates the center of moral responsibility to the institutional system: in his account, DEI created incentives, gatekeepers failed to apply normal scrutiny, and later suppression allowed the eventual exposure to become more catastrophic than an earlier proportionate correction might have been (approximately 28:30–30:14).
Causal and evidentiary boundary
These are Hughes’s and Cofnas’s moral interpretations, preserved from an automated transcript—not adjudicated findings about Arday’s intentions, institutional motive, or the cause of his death. Public evidence supports neither the claim that Cofnas alone caused the death nor the claim that institutional DEI was its singular cause.
A better responsibility map separates several questions:
Question
Appropriate standard
Was the copied or inaccurate material real?
Document-by-document comparison and formal investigation.
Was publication justified?
Public interest, verification, proportionality, opportunity to respond, and avoidance of gratuitous humiliation.
Did Cambridge and journals respond appropriately before publication?
Timeliness, due process, consistency, and availability of corrective measures short of public catastrophe.
Did media amplification become excessive?
Repetition, framing, tone, independent verification, and foreseeable harassment.
Who caused Arday’s death?
Requires evidence about a complex individual event; temporal sequence and moral outrage do not establish singular causation.
The morally strongest synthesis is that reporting substantiated concerns can be justified while the scale, tone, timing, and institutional handling of the episode remain open to criticism. Accountability for scholarship and compassion toward a distressed person are not mutually exclusive. Nor does identifying institutional failure erase Arday’s agency or transfer total responsibility to the person who made the evidence public.
Overall assessment
Where Cofnas is strongest
Close source criticism of Kevin MacDonald. He identifies cherry-picking, unfalsifiability, base-rate neglect, and slippage from overrepresentation to group coordination.
Conceptual analysis in evolutionary biology. His work on selection, innateness, mismatch, incest norms, and moral evolution makes precise, contestable contributions.
Identifying ideological pressure. Some retraction and denunciation campaigns plainly attributed positions to him that the targeted paper did not contain.
Criticizing his own side. His rejection of antisemitic race realism, conspiracism, protectionism, and populist epistemology complicates simplistic classifications.
Auditing specific claims. His challenges to Sowell’s undocumented Dunbar IQ number and to Jason Arday’s textual overlap deserve examination independently of his broader politics.
Acknowledging communication tradeoffs. He does not defend provocation for its own sake; he accepts that truth, audience, foreseeable misuse, and social peace must be considered together.
Where he is weakest
Treating indirect evidence as closure. He turns a researchable hereditarian hypothesis into “scientifically correct” doctrine without direct causal genomic evidence for the group gap.
Moving from groups to institutions. Harvard admissions models and average test scores cannot support a prediction that nearly all Black people would vanish from elite occupations.
Collapsing descriptive and normative claims. Intelligence predicts some outcomes; that does not establish moral superiority, political entitlement, or the design of a just social order.
Monocausal political explanation. His equality-thesis model of wokism underweights law, incentives, religion, coalition politics, organizational imitation, and moral commitments not based on empirical sameness.
Polemical overreach. Terms such as “fraud,” “stupidity,” “brain rot,” and “woke religion” create memorable arguments but often conceal uncertainty and heterogeneity.
Using audit success as support for a worldview. Correctly finding a bad citation, plagiarized passage, or false accusation does not validate hereditarianism or a political revolution.
Elite epistemic paternalism without institutional design. His proposed combination of elite acceptance and mass restraint leaves unanswered who certifies the knowledge, how dissenting experts participate, how policies are publicly justified, and how a mild etiquette norm avoids becoming censorship.
Final verdict on the media dispute
The premise that critics “cannot attack the accuracy of his research” is not supported. Some coverage is careless: it conflates his Cambridge affiliations, says his research proved claims it merely defended investigating, or substitutes labels such as “white supremacist” for exact description. But the central methodological objection survives fact-checking: Cofnas’s peer-reviewed work does not establish the causal genetic share of racial intelligence gaps, while his popular political writing treats the hereditarian conclusion as settled. Both things can be true at once—he has been smeared in some instances, and he overstates the science in others.
Complete peer-reviewed paper inventory
The following inventory covers the 21 items listed on Cofnas’s papers page as reviewed on 23 August 2026.
Year
Paper
Field
What it contributes
Evidence status
2024
Natural Selection Requires No Teleology in Addition to Heritable Variation in Fitness
Philosophy of biology
Adds a non-teleology condition to distinguish natural selection from guided processes
Conceptual argument
2023
Still No Evidence for a Jewish Group Evolutionary Strategy
Evolutionary psychology
Replies to MacDonald and argues the coordinated-strategy evidence remains inadequate
Source critique and synthesis
2023
How Gene–Culture Coevolution Can—but Probably Did Not—Track Mind-Independent Moral Truth
Metaethics/evolution
Tests whether cultural selection can be truth-tracking
Philosophical model
2022
The Golden Rule: A Naturalistic Perspective
Moral psychology/religion
Explains reciprocal norms naturalistically and limits the rule’s impartial meaning
Comparative conceptual argument
2021
The Anti-Jewish Narrative
Philosophy/political psychology
Audits three MacDonald claims about ethnocentrism, liberalism, and antisemitism
Source critique
2020
Are Moral Norms Rooted in Instincts? The Sibling Incest Taboo as a Case Study
Moral psychology
Challenges a direct instinct-to-norm account
Conceptual and empirical synthesis
2020
A Debunking Explanation for Moral Progress
Metaethics
Gives non-truth-tracking explanations for liberal moral convergence
Philosophical argument
2020
Coercive Paternalism and the Intelligence Continuum
Applied ethics
Connects decision competence to possible targeted paternalism
Normative argument
2020
Research on Group Differences in Intelligence: A Defense of Free Inquiry
Philosophy of science
Defends research and presents indirect hereditarian considerations
Methodological/ethical argument, not proof of genetic gap
2020
Should Dietary Guidelines Recommend Low Red Meat Intake? (with Frédéric Leroy)
Nutrition policy
Challenges certainty of categorical low-red-meat guidance
Evidence review
2019
Is Kevin MacDonald’s Theory of Judaism “Plausible”? A Response to Dutton (2018)
Evolutionary psychology
Continues the MacDonald source and mechanism critique
Scholarly reply
2019
Is Vegetarianism Healthy for Children?
Nutrition/ethics
Argues pediatric safety claims were more confident than evidence
Evidence review; time-sensitive
2018
Power in Cultural Evolution and the Spread of Prosocial Norms
Cultural evolution
Gives powerful coalitions a deliberate role in norm formation
Theoretical model
2018
Judaism as a Group Evolutionary Strategy: A Critical Analysis of Kevin MacDonald’s Theory
Evolutionary psychology
Introduces default hypothesis and documents source problems
Major source critique
2018
Religious Authority and the Transmission of Abstract God Concepts
Cognitive science of religion
Gives authority and reasoning a role beyond minimal counterintuitiveness
Theoretical/empirical synthesis
2018
Larregue’s Critique of Cofnas et al.: A Rejoinder (with Noah Carl and Michael A. Woodley)
Intelligence/history of science
Defends an earlier historical interpretation
Scholarly reply
2018
Does Activism in Social Science Explain Conservatives’ Distrust of Scientists?
Sociology of science
Links ideological activism to declining conservative trust
Observational inference; causality not isolated
2017
Innateness as Genetic Adaptation: Lorenz Redivivus (and Revised)
Philosophy of biology
Defines innate traits as genetic adaptations
Conceptual proposal
2016
A Teleofunctional Account of Evolutionary Mismatch
Philosophy of biology
Clarifies mismatch by selected function and environment
Conceptual taxonomy
2016
Scientific Literacy, Optimism about Science and Conservatism
Political psychology
Separates social and economic conservatism in science attitudes
Empirical analysis
2016
Science Is Not Always “Self-Correcting”: Fact–Value Conflation and the Study of Intelligence
Philosophy/history of science
Uses intelligence controversies to criticize automatic-correction narratives
Historical and conceptual argument
Complete popular-article inventory
The following covers all 29 entries listed on Cofnas’s popular articles page as reviewed on 23 August 2026. These are arguments and interventions, not “research findings” merely because the author is an academic.
Approx. date order
Article
Main argument
Audit note
1
DEI Fraud and Cover-Up at Cambridge
Jason Arday allegations reveal DEI-protected failure
Documentary overlap concerns are real; generalized DEI conclusion exceeds adjudication
Consensus and authority sometimes substitute for evidence
Correctly demands source scrutiny; funding insinuations can become overgeneralized distrust
28
When Accusations of Antisemitism Lose Their Bite
Politicized accusations reduce credibility
Legitimate warning; must not become a self-sealing reason to dismiss evidence of antisemitism
29
Are College Admissions Officers Equipped to Judge Who Is Ethical?
Requiring applicants to document virtue rewards performative morality and asks officers to judge character beyond their competence
The original National Review page was inaccessible; the title, URL, and central argument were corroborated through the author’s CV and quotations in later commentary, so do not quote beyond those excerpts without the archived primary text
Practical reading order
Start: “Judaism as a Group Evolutionary Strategy” (2018).
Then: “The Anti-Jewish Narrative” (2021) and “Still No Evidence…” (2023).
For his academic race/IQ position: “Research on Group Differences in Intelligence” (2020).
For the shift from inquiry to political certainty: “A Guide for the Hereditarian Revolution” (2024).
For breadth beyond race: the natural-selection, innateness, moral-progress, cultural-power, and incest papers.
For institutional controversy: Cambridge’s official statements, Larsen et al.’s critique, and Cofnas’s retraction/free-speech essays.
Questions to keep open
What direct molecular evidence would Cofnas accept as falsifying a substantial genetic explanation of a group gap?
How much of the measured gap changes across age, cohort, country, socioeconomic context, test, and migration history?
Can a stable race category support prediction after modeling specific ancestry, family environment, geography, and social treatment?
How would his political program protect individuals from group-statistical discrimination?
Does replacing “woke” gatekeepers with hereditarian gatekeepers improve open inquiry or merely reverse institutional partisanship?
Can his power-centered cultural-evolution model explain his own proposed elite revolution as interest selection rather than truth tracking?
Which parts of the Arday case would remain newsworthy if the accused and accuser’s races and politics were reversed?
Handwritten Notebook Pages
Source
Open Nathan Cofnas notebook pages — start at notebook p. 192 (file not included in this release)
Open Nathan Cofnas notebook pages — start at notebook p. 200 (file not included in this release)