Intelligence does not come predominantly from either the mother or the father. It is polygenic, meaning thousands of gene variants across the whole genome each contribute a tiny amount. The X chromosome carries some cognition-linked genes, and some studies find slightly stronger maternal effects, but the evidence is not strong enough to rank parents. Most of the heritable signal is spread across both parents.

The idea that intelligence is inherited mainly from one parent is one of the most persistent claims in popular science. It travels well because it sounds mechanistic: a chromosome here, a protein there, a tidy story about a "smart gene." The real picture is messier and far more interesting. Intelligence is one of the most heavily studied human traits, and the research spans twin studies, adoption studies, genome-wide association studies (GWAS), and molecular work on individual chromosomes. Each design answers a different question. Stacking them together shows why no serious geneticist assigns intelligence to the mother or the father alone.

How polygenic inheritance spreads the signal across both parents

The mechanism matters more than the metaphor. Intelligence is not a trait that a single gene switches on. It is quantitative, varying continuously across a population, and it emerges from the combined, small effects of very many genetic variants. This is polygenic architecture, and it is the key fact that demotes the "mother versus father" framing.

Each variant contributes a minuscule nudge. Some variants sit in genes involved in building neuronal synapses. Others influence myelination, the wrapping of nerve fibers that speeds signal transmission. Still others shape early brain growth or the availability of neurotransmitters. None of these acts alone. A human embryo inherits roughly half its variants from each parent, and because the variants are distributed across all 23 chromosome pairs, the aggregate propensity is a blend. That blend is why siblings raised in the same home differ so much in cognitive ability while still resembling their parents.

Monozygotic twins share essentially 100 percent of their segregating DNA, while dizygotic twins share about 50 percent, the same as ordinary siblings. Twin studies exploit that contrast. When identical twins resemble each other more on cognitive tests than fraternal twins do, the difference is attributed to genetics, because both types of twins share a rearing environment. This design cannot tell you which parent supplied the DNA. It can only show how much inheritance matters overall.

One complication is that the DNA we inherit is not the only DNA that influences development. Epigenetic marks, chemical tags on DNA that turn genes up or down, and maternal RNA delivered through the egg, shape early development before any gene acts. This is a real biological asymmetry: the egg is a large cell stocked with molecules and machinery, while the sperm is small. That asymmetry feeds maternal-effect theories, but it does not mean cognitive genes are maternal. It means the developmental environment of the earliest embryo is maternally provisioned.

Genomic imprinting is the strongest scientific version of the "one parent matters more" claim. Imprinting is a form of parent-of-origin gene regulation in which a gene is active only when inherited from the mother, or only when inherited from the father. Imprinted genes are real and can have dramatic effects. The known cases cluster around growth and brain development rather than general intelligence. Imprinting affects a small subset of genes, so it cannot carry the whole story of a highly polygenic trait.

A plain-language version: intelligence is less like an inheritance from one parent and more like a recipe you were given from both, rewritten many times in the kitchen.

What the numbers say about heritability and parental contributions

The word heritability trips up most readers. It does not mean "the fraction of your intelligence that is genetic." It means the share of variation in a trait across a specific population, in a specific environment, at a specific time, that is attributable to genetic differences. Heritability is a property of populations, not of individuals. This distinction explains why estimates climb as people age.

Twin studies in affluent countries typically put the heritability of cognitive ability around 50 to 70 percent in adulthood, lower in early childhood, and rising through adolescence and young adulthood. A large meta-analysis of twin studies (Polderman et al., 2015, Nature Genetics) covering thousands of traits estimated cognitive ability heritability in the mid-50s to 60s range across the lifespan. The rise with age is counterintuitive: environments seem to matter less as people get older. The usual explanation is gene-environment correlation. As children gain autonomy, they select, evoke, and build environments that match their genetic inclinations, so genes and environments become more entangled over time.

GWAS goes further, scanning millions of variants across huge samples. Heritability estimates from common single-nucleotide variants, called SNP heritability, are lower than twin estimates. Most published figures sit in the 20 to 40 percent range for cognitive test performance, depending on sample size and the exact phenotype. The gap between twin-based and GWAS-based estimates is called missing heritability. Part of it is rare variants, part is interactions between genes, part is measurement error, and part is that common variants are individually too weak to detect at current sample sizes. These numbers do not identify a parent.

The shared environment, meaning the home and family conditions shared by siblings, contributes a notable but smaller share to cognitive variation, typically in the range of 20 to 30 percent of variance in childhood studies. Yet the unique environment, the distinct experiences each child has even within one family, accounts for a large share that grows with age. The practical upshot: the parents contribute genes from both sides and a shared setting, and life beyond the home supplies much of the rest.

No credible quantitative estimate assigns intelligence to "mother" or "father" as a source. The heritable component is not partitioned by parent in any published model.

The real evidence for maternal or paternal skew

Some research does find small parent-specific effects, and it is worth understanding what those studies actually show. The most cited line of work involves the X chromosome. Females carry two X chromosomes; males carry one X and one Y. If cognition-related genes enriched on the X chromosome were the dominant force, maternal-line inheritance should look slightly stronger. Some candidate-gene studies reported X-linked genes associated with cognitive disability, which is consistent with an X-linked vulnerability in males. But for typical intelligence variation, X-chromosome effects are modest.

A more substantial body of work examined imprinted genes and brain function. GWAS of cognitive traits have found signals near known imprinted loci, but parent-of-origin testing is technically hard and occasional reports of maternal or paternal skew in particular gene clusters have not replicated robustly. The honest summary is that a small number of individual genes show parent-of-origin effects, and no evidence supports a general maternal or paternal dominance for intelligence.

There is also a methodological trap. Studies sometimes compare children to mothers versus fathers on cognitive tests and report that the correlation with one parent is stronger. Correlations are easily confounded by assortative mating, meaning people tend to partner with people of similar cognitive ability, so both parents' genes correlate with the child's. A stronger mother-child correlation could reflect better measurement, more time spent together, greater test-taking comfort, or statistical artifact. It does not isolate genetic transmission.

Family studies with adoption designs offer a cleaner test. Adopted children resemble their biological parents on cognitive measures more than their adoptive parents in adulthood, while the shared home effect fades. But those studies compare biological mothers and fathers separately only rarely. The general pattern is that both biological parents contribute, and the shared environment matters early and then matters less. When a study reports one parent "driving" the effect, the personal caveat is to check whether the design can separate genetics from the parenting environment that also tracks with abilities.

A counterintuitive detail: the more carefully you measure cognition and the larger your sample, the smaller and more diffuse every individual genetic effect becomes. The single-gene story loses power as the data scale up. The parent-specific story shrinks for the same reason.

Alternatives compared: genetic evidence designs

Different study designs answer different questions. The table below compares the main approaches.

Study design What it measures Approximate intelligence estimate What it cannot tell you
Twin studies Genetic vs shared vs unique environment in a population Heritability roughly 50-70% in adults Which parent contributed any given gene
Adoption studies Adopted child vs biological vs adoptive parents Biological family resemblance persists into adulthood Parent-specific genetic split when both are compared rarely
GWAS Individual common variants across the genome SNP heritability roughly 20-40% Effects of rare variants or parent of origin
Polygenic score analyses Aggregate weighted effect of many variants Explains a small fraction of variance in any single individual Parent-specific or causal contributions
Imprinting studies Parent-of-origin gene activity in a small gene subset Real but localized effects General intelligence architecture
Family/trio sequencing Rare variants and de novo mutations within families Identifies rare high-impact variants Population-wide heritability or parent ranking

Two cautions run through this table. First, heritability and SNP heritability are population statistics and should not be read as predictions for one child. A trait with 60 percent heritability says nothing definitive about whether a specific person's cognitive score is 60 percent genetic. Second, an aggregate score derived from thousands of variants explains a small share of variance in any given individual, which is why consumer polygenic scores for complex traits are poor individual predictors and should not guide personal decisions.

Misconceptions that keep the mother-father story alive

Several specific claims circulate widely and deserve direct correction. These are the ones that most often hijack the question.

"Intelligence is carried on the X chromosome, so mothers matter more." Some cognition-linked genes sit on the X chromosome, but the contribution is small, and the X chromosome is one of 23 pairs. The claim overstates a minor effect. "Smart people inherit it from their mothers only." No credible study assigns most cognitive heritability to one parent. Parent-child resemblance is confounded by assortative mating and shared environment. "The father's genes are silenced for brain traits." This confuses a few known imprinted genes with the entire genome. Imprinting affects a small number of genes, mostly related to growth and specific disorders. "A polygenic score tells you how smart a person will be." Scores explain a small fraction of variance in any individual and are poor predictors outside the populations in which they were built. "Heritability of 60 percent means 60 percent of intelligence is genetic." Heritability describes variation in a population, not the composition of an individual's intelligence. "Since intelligence is genetic, environment does little." Heritability estimates are environment-dependent, and in conditions of deprivation or enrichment, environmental effects can change substantially.

The parent-of-origin framing also misreads adoption data. When an adopted child's cognitive outcomes resemble the biological mother more than the biological father, that is not necessarily a maternal genetic signal. It can reflect the sample studied, the ages at which families were evaluated, or attrition in which one parent was harder to study.

A plain-language correction: genes for cognitive ability are spread all over both parents' genomes, and the environment is a shifting backdrop that changes what those genes even mean.

Practical actions for reading intelligence research

You do not need a genetics degree to evaluate these claims, but you need a checklist that separates signal from story.

Check whether a study measured heritability in a population or made an individual prediction. If a headline promises your child's IQ, the underlying paper likely did not. Ask whether the design can separate genes from environment. Adoption and twin designs can; parent-child correlations usually cannot. Look for assortative mating. If both parents have similar cognitive profiles, any maternal or paternal "effect" is confounded. Treat polygenic scores as group statistics. A score built in one ancestry group predicts worse in another, which makes them unreliable for personal use. Reconcile thresholds yourself. If one paper says heritability is 50 percent and another says SNP heritability is 25 percent, these are measuring different things, not contradicting each other. The roughly 30 to 40 percentage-point gap is the missing heritability problem. Distinguish "genes matter" from "genes from one parent matter." The first is well supported; the second is not.

The practical threshold for acting on any genetic claim about cognition is high. Nothing in current science supports planning around a parent's "share" of a child's intelligence. What the evidence supports is a more diffuse model: both parents supply variants, shared family conditions shape early development, and a large portion of what a person becomes is individual and environmental.

Limits of the evidence and the honest bottom line

Three limits constrain what can be concluded. First, most large cognitive GWAS samples are heavily of European ancestry. Estimates may not transfer cleanly to other ancestry groups, which is a serious problem for extending polygenic scores to diverse populations. Second, intelligence tests capture particular skills in particular cultural contexts; a test score is not a pure measure of an underlying capacity. Third, parent-of-origin studies are technically demanding, and the sample sizes for imprinting effects on typical cognition are small. Claims of maternal or paternal dominance for intelligence should be labeled as unproven until replicated in very large, diverse samples.

Two reported percentages that often seem to conflict are worth reconciling in one place. Twin studies say roughly 50 to 70 percent of adult cognitive variation is heritable. GWAS says roughly 20 to 40 percent of variation is attributable to common SNPs. These are not two competing facts. Twin heritability includes all genetic effects, including rare variants, gene-gene interactions, and any parent-of-origin effects, plus assumptions about identical environments. SNP heritability counts only the common variants GWAS can measure. The difference is the missing heritability, and it is an open research problem, not a reason to prefer one estimate as "the" answer.

What would change the picture? A large, well-powered study that partitions heritability by parent of origin across typical-range cognition and replicates across ancestries. No such study currently settles the question, and the strong prior from polygenic architecture is that any parent-of-origin effect will be small relative to the total.

So the direct answer stands: intelligence inheritance is essentially shared across both parents, with a minor, real but unquantified role for parent-of-origin effects on a small set of genes. The "mother versus father" question gets the science wrong at the level of architecture. Cognition is not a single package handed down a line. It is a vast set of tiny probabilities drawn from both sides and reshaped by everything that happens after conception.