Critical windows: why early development locks in so many traits

The Short Version

Many traits people assume are fixed by genes are actually set by what happens during short windows early in life, and once those windows close, the result is hard to undo. Neuroscientist David Linden shows that nearsightedness, heat tolerance and speech accent all depend on early experience, not DNA [1]. Erich Jarvis explains that speech has an especially strong critical period, which is why a child picks up a language and its accent effortlessly while an adult struggles, and he offers a reason: the brain locks in what it learns so it doesn't keep overwriting itself [2]. Andrew Huberman's tour of sexual development shows the same logic in hormones: the right molecule at the right developmental moment builds the body and wires the brain in ways that last a lifetime [3]. That is why early exposures, good ones like sunlight and language, and bad ones like maternal infection, herbicides, alcohol or cannabis during puberty, carry so much weight [1][3].

Table of contents

What They Said

David Linden: traits that look genetic but are set by early experience

David Linden, a Johns Hopkins neuroscientist and author of Unique, a book about human individuality, makes the case that a surprising number of traits are fixed early in life by experience rather than by genes [1]. As he puts it, "there are a number of situations where you would guess that something is genetic, but it isn't. It's actually dependent on early life experience" [1].

Nearsightedness. His first example is the eye. "Kids that don't get outside are much more likely to be myopic," he says, referring to roughly the first five years of life [1]. The mechanism is physical: "light seems to stimulate the expression of a class of molecules called trophic factors" that change the shape of the eyeball [1]. It isn't the retina, lens or cornea that differs, but "the actual degree of elongation of the eyeball," which "seems to be light dependent early in life" [1]. Huberman, true to form, calls it "news you can use. You should get your kids outside" [1].

Heat tolerance in Japanese soldiers. Linden's favorite example comes from World War II. Japanese soldiers from cold northern Hokkaido suffered heat stroke far more than soldiers from semi-tropical Kyushu [1]. Both groups had the same number of eccrine sweat glands, but in southerners more of those glands were innervated by nerves, so they could actually be switched on [1]. The decisive evidence came from families who moved. Children of northern families raised in the south developed the southern pattern, and the reverse was also true: when a well-established Kyushu family moved to Hokkaido and had a child, "that child developed the northern sweat gland innervation pattern," with "less nerve innervation of those sweat glands" and so lower heat tolerance [1]. The trait was set by early experience, not genes [1].

Linden sees this as a feature, not a flaw. "This gives an advantage that you can't get through evolution, and that it can happen right away, in one generation," he says [1]. "Evolutionary change is slow... But when you have a phenomenon that is set by early life experience, well, then you can benefit from that early life experience within your own life" [1].

Seasonal fur in voles. The same principle works even before birth. Voles gestated under artificially lengthening days are born with thin summer fur, while those gestated under shortening days get thick winter fur, work associated with Irv Zucker, a mentor to both Linden and Huberman [1]. The body reads a signal about the world it is about to enter and builds itself accordingly.

Perfect pitch. Linden says perfect pitch shows up in maybe one in ten trained musicians [1]. Twin studies suggest it is roughly 30 to 40% heritable, and early ear training greatly increases the odds of developing it [1]. So it is a trait where genes and an early window interact.

Speech accent: zero percent heritable, fixed in childhood. Linden uses accent to anchor the far end of the heritability scale. Earwax type is 100% heritable, but "speech accent is 0% heritable. It is entirely dependent upon the speech that you experience in your childhood. And interestingly, it's the speech of your peers more than the speech of your family, which is why the children of immigrants sound like the place where they wound up, not like their parents" [1]. He is careful to separate accent from voice quality, such as whether a voice is high or low or nasal, which depends on the physical vocal tract and is in part heritable [1]. Accent is the clearest example of a trait that is entirely learned and yet, once set in childhood, notoriously hard to change.

The womb as a critical window. Linden is skeptical of claims that trauma or famine is passed through epigenetic marks across generations in mammals; he says "you may well inherit your grandma's or great grandma's trauma, but you're probably doing it socially, not through marks on your DNA" [1]. What he considers well established is single-generation effects during pregnancy [1]. Male children who were in the womb during the 1918 flu pandemic were a millimeter or two shorter on average, and rates of schizophrenia rose about fourfold, from about 1% to 4%, with what we'd now call autism also rising [1]. Gloria Choi's lab at MIT modeled this in mice: a simulated viral infection in pregnancy, interacting with gut bacteria, triggers the immune molecule interleukin-17, which crosses the placenta at a sensitive window and disorganizes the layers of the developing cortex [1]. Blocking IL-17 prevented the effect, and similar clumps of cells appear in some autistic brain tissue [1]. Linden notes that we don't yet know the effects of COVID during pregnancy [1].

Environment shapes how much genes can show. Height is about 85% heritable in the United States, but only about 50% in places like rural Bolivia or India, where poor nutrition and infections stop people from reaching their genetic potential [1]. IQ is about 60 to 70% heritable in affluent populations and less so among people facing poverty or discrimination, for the same reason [1]. His takeaway is that everyone needs basic nutrition, medical care and schooling to reach their potential [1].

Erich Jarvis: speech has one of the strongest critical periods

Erich Jarvis, a Rockefeller University neuroscientist who studies how brains learn to speak, explains why childhood is the prime time for language, drawing on both humans and songbirds [2].

Humans and birds share the same windows. Researchers going back to the 1950s found that songbirds, parrots and hummingbirds, "just three of them out of the 40-something bird groups," can imitate sounds like humans do [2]. Then they found "even more similarities, these critical periods" [2]. Jarvis describes the grim human evidence: "if you remove a child, this unfortunately happens where a child is feral, and it's not raised with human and goes to their puberty phase of growth, becomes hard for them to learn a language as an adult. So there's this critical period where you learn best. And even later on, when you're in regular society, it's hard to learn. Well, the birds undergo these same thing" [2]. Another shared feature: when humans or vocal-learning birds go deaf, their vocalizations deteriorate, while a deaf chicken or monkey changes very little [2].

Birds learn best from their own kind, but can learn others. Following Peter Marler's work, young songbirds learn their own species' song best but can learn another species' song less well, a tendency Marler called "the innate predisposition to learn," roughly the equivalent of universal grammar in linguistics [2]. A zebra finch raised with a canary sings a hybrid song [2]. Given a choice between an adult of its own species and a Bengalese finch, a young zebra finch prefers its own kind, but if that neighbor is removed, it learns the Bengalese finch's song very well, which points to social bonding as part of the learning [2].

Children merge languages in ways adults can't. On pidgin languages, Jarvis says "cultural evolution remarkably tracks genetic evolution" [2]. A child raised by parents who speak, say, Chinese and English will merge phonemes and words during the critical period in a way adults cannot, and the shared sounds become the ones used most [2].

Huberman's niece and the adult learner. Huberman describes his 14-year-old niece, who speaks Spanish and English "incredibly well," while "I've struggled with Spanish my whole life... I've tried to learn Spanish as an adult. It's really challenging" [2]. He asks whether it is easier to learn languages without an accent early in life, and whether early multilinguals find new languages easier later. Jarvis answers: "So the answer to both of those questions is yes" [2].

The whole brain has critical periods, but speech's is stronger. "The entire brain is undergoing a critical period development, not just the speech pathways," Jarvis explains, "And so it's easier to learn how to play a piano. It's easier to learn how to ride a bike for the first time and so forth as a young child than it is later in life" [2]. He specifies he means learning from first principles, "the very first time" [2]. But "speech behavior I think has a stronger critical period" than other circuits [2].

Why the window closes. Jarvis gives a clear reason: "the brain can only hold so much information. And if you are undergoing rapid learning to acquire new knowledge you also have to dump stuff" [2]. On top of that, "for survival you don't want to keep forgetting things. And so the brain is designed I believe to undergo this critical period and solidify the circuits with what you learned as a child and you use that for the rest of your life" [2].

Humans stay more open than other animals. There is a partial escape hatch. "We humans stay even more plastic in our brain functions controlled by a gene called SRGAP2," Jarvis says; "We have an extra copy of it" that keeps speech circuits and other brain regions "in a more immature state throughout life compared to other animals" [2]. Or, as the summary quotes him, "We're still juvenile like compared to other animals" [2]. And people who grow up with several languages find extra languages easier, not because their brains are more plastic, but because they have kept a wider set of phonemes they can recombine [2]. Childhood, in other words, doesn't just teach one language; it stocks a toolbox that makes later learning easier.

Adult change is harder, not impossible. Jarvis also notes that adults can improve a stutter with behavioral therapy, such as speaking more slowly or tapping out a rhythm, which he thinks works through sensory-motor integration [2]. And his own life shows early talent carrying forward: he began winning dance contests as a teenager, trained at the Alvin Ailey and Joffrey schools, and never stopped dancing after choosing science [2]. Huberman wonders whether that discovered ability reflects something "built up in childhood and or that perhaps we do all have different genetic leanings for different motor functions" [2]. Jarvis says his 23andMe results show variants linked to fast-twitch muscles and one linked to difficulty singing on pitch, though he stresses he does not know whether they explain anything [2].

Huberman on hormones: timing builds the body and wires the brain

In a solo episode, Huberman walks through how sex hormones shape development from embryo through puberty, and the timing theme runs throughout [3].

Sex hormones make long-lasting changes. Sex steroid hormones like testosterone and estrogen are lipophilic, so they pass through cell and nuclear membranes and interact directly with DNA [3]. As Huberman explains, "they can change the sorts of things that cells will become and they can change the way that cells function in a long-term way" [3]. That is the biological basis for effects that stick.

A fork in the road, early. The Y chromosome carries genes that steer development toward testes, including one for Müllerian inhibiting hormone, which prevents formation of the female Müllerian ducts, and others like SRY that promote testes [3]. Huberman calls this transition "a fork in the road that happens very early in development while fetuses are still in the embryo" [3]. From there he traces chromosomal sex to gonadal, hormonal and morphological sex, and notes "it's a long distance from chromosomes to gender identity and gender identity has a lot of social influences and roles" [3].

Guevedoces: the wrong molecule at the wrong time. Huberman's most vivid example shows how a missed early window plays out. "It's not testosterone that's responsible for the development of the penis" in the fetus; it is dihydrotestosterone (DHT), made from testosterone by the enzyme 5-alpha reductase [3]. In a mutation first identified in the Dominican Republic and published in Science in the 1970s, children lacking this enzyme are born looking female, with "very little or no external penis" [3]. After being raised as girls, "around the age of 11 or 12 or 13," they "start to sprout a penis," because pubertal testosterone drives secondary growth [3]. The name translates roughly as "penis at 12" [3]. The case shows two separate windows: DHT for primary development in the womb, testosterone for secondary changes at puberty [3].

Estrogen masculinizes the brain. A second surprise: "the masculinization of the brain is not accomplished by testosterone. It is accomplished by estrogen" [3]. Neurons contain the enzyme aromatase, which converts testosterone into estrogen [3]. Citing his Stanford colleague Nirao Shah, Huberman summarizes: "Estrogen sets up the masculine circuitry in the brain. And testosterone is then what controls the display of those behaviors later in life" [3]. This is critical-window logic in its clearest form: one molecule early builds the circuit, another later determines whether it is used.

Receptors matter too. In androgen insensitivity syndrome, people are XY, have internal testes and make testosterone, but their androgen receptors don't work, so their bodies can't respond; they look completely female and generally report feeling like girls and later women [3]. A hormone only acts through a working receptor [3].

Early exposures that interfere. Because these pathways are so sensitive, outside chemicals can disrupt them [3]. Huberman credits Tyrone Hayes at UC Berkeley, who identified the common herbicide atrazine as causing severe testicular malformations in frogs; at sites in the western and midwestern U.S., 10 to 92% of male frogs had testicular abnormalities [3]. In humans, average sperm density in the United States and Western Europe fell from 113 million per milliliter in 1940 to 66 million in 1990, and semen volume dropped about 20% [3]. Between 1981 and 1991, the proportion of normal sperm production fell from 56.4% to 26.9% [3]. Huberman attributes this largely to widely used herbicides and says herbicides disrupting estrogen "might explain why puberty is happening so much earlier in young girls these days" [3]. He links it directly to development: "you need testosterone to get you to die hydro testosterone for primary sexual characteristics. You need estrogen that's come from testosterone" to masculinize the brain [3].

Puberty is another window. Huberman reports that many studies point to THC or other compounds in cannabis increasing aromatase activity, raising circulating estrogen, which may explain a higher incidence of breast tissue development in male users and can counteract testosterone and DHT's effects [3]. "certainly during puberty it can have profound effects on these hormonal systems," he says [3]. Alcohol during pregnancy causes fetal alcohol syndrome, and drinking during puberty is probably also a bad idea because alcohol, especially beer and other grain alcohols, can increase estrogenic activity in both boys and girls [3]. He reminds listeners that puberty has a beginning, middle and end, and that development really spans the whole lifespan [3].

Where They Agree, and Where They Don't

Broad agreement: timing matters as much as the signal. All three sources describe the same pattern. A signal arriving during a specific window has a lasting effect that the same signal later would not: light before about age five shapes the eyeball [1]; climate in childhood fixes sweat gland innervation [1]; day length during gestation sets a vole's coat [1]; language heard before puberty sets speech and accent [1][2]; DHT in the fetus builds the penis and estrogen wires the male brain [3]; an infection at a sensitive point in pregnancy disorganizes the fetal cortex [1].

Agreement that "not genetic" doesn't mean "easy to change." Linden's accent example and Jarvis's language learning point the same way: accent is 0% heritable [1], yet once childhood sets it, adults struggle to shift it [2]. Heat tolerance is also set by experience, not genes, and yet it is fixed within the individual [1]. The sources jointly overturn the assumption that only genetic traits are hard to change.

Agreement that the window is adaptive. Linden sees early-life setting as a way to adapt "in one generation" to the world a child is born into [1]. Jarvis sees closing the window as a way to keep a limited-capacity brain from constantly overwriting and forgetting what it needs to survive [2]. Neither treats critical periods as a defect.

Some difference in emphasis on how firmly things lock. Jarvis stresses that humans remain unusually plastic because of the extra SRGAP2 copy, that early multilinguals keep a larger toolbox, and that adults can still improve a stutter [2]. Huberman notes that peripheral nerves "regenerate and can remodel themselves extensively in ways that neurons within the brain and spinal cord tend to remodel less, especially as one gets older out of the so-called critical period" [1], suggesting the lock is tighter for the brain than for the body. Linden adds that exercise is about as good an antidepressant as SSRIs and that relief from any psychiatric condition involves some form of plasticity [1], so adult change remains real. The sources agree windows exist; they differ slightly on how much room remains after them.

A limit on the idea: inheritance across generations. Huberman raises epigenetic inheritance of acquired traits, but also says "simply going to the gym and getting fit does not make your children more fit, because the germ line as far as we know is not modified in a direct way" [1]. Linden goes further, saying "I remain unconvinced of transgenerational epigenetic inheritance in mammals," while adding that "just because it hasn't been shown convincingly now, it doesn't mean that it won't be" [1]. So early windows are powerful within one life, but the sources don't support claims that they are passed down biologically.

Uneven strength of evidence on exposures. The atrazine and sperm count data are presented as textbook and peer-reviewed findings [3]. The cannabis link to breast tissue is described as observed anecdotally, backed by studies Huberman could identify [3]. And on herbicides, he declines to tell people what to do: "Now does this mean that you have to run around and neuratically avoid anything that includes things like atrazine... I don't know. That's up to you" [3].

Why It Works

The sources give several mechanisms for why early windows lock in outcomes.

Growth signals act on tissues while they are being built. In the eye, light triggers trophic factors that control how much the eyeball elongates, which is only possible while the eye is growing [1]. In sweat glands, the number of glands is the same, but the nerve connections to them are set early [1]. Once the structure is built, the chance to shape it has largely passed.

Hormones change gene expression permanently. Sex steroids enter the cell nucleus and act on DNA, changing what cells become and how they function "in a long-term way" [3]. A hormone present during the fork in the road decides which path a tissue takes [3]. Missing the molecule at that point, as with the Guevedoces, leads to a completely different early outcome [3].

Organizing versus activating. Estrogen made from testosterone sets up male brain circuitry early, and testosterone later controls whether those behaviors are shown [3]. The early signal builds the hardware; the later one runs it.

The brain trades flexibility for stability. Jarvis's explanation for language windows is capacity and survival: the brain can only hold so much, rapid learning requires discarding, and a stable adult brain shouldn't keep forgetting [2]. So after a period of rapid acquisition, the circuits solidify [2]. SRGAP2 partly offsets this in humans by keeping circuits more juvenile [2].

Sensitive immune windows in pregnancy. Choi's work shows a specific pathway: infection plus gut bacteria trigger IL-17, which crosses the placenta at a sensitive window and disrupts cortical layers; blocking IL-17 prevented it [1].

Social learning drives what gets locked in. Accent comes from peers more than parents [1], and young zebra finches learn from their own kind when they can, but learn another species' song well when they can't, pointing to social bonding [2]. What a child is surrounded by during the window determines what gets set.

Randomness adds a layer on top. Linden adds that the genome is "a rather vague recipe": identical twins can differ by 30% in organ size at birth, and inbred mice behave differently [1]. So early development is shaped by genes, experience and chance, which he summarizes as heritability interacting with experience filtered through the random nature of development [1].

Where It Applies, and Where It Doesn't

It applies most to structures and skills built early. Eye shape, sweat gland innervation, genital and brain development, and speech and accent are the clearest cases [1][2][3]. Linden also includes perfect pitch, where early training greatly raises the odds [1].

It applies to the womb and to puberty, not only early childhood. Pregnancy is a window for infection, alcohol and endocrine disruptors [1][3], and puberty is another for cannabis and alcohol effects on hormones [3].

It doesn't mean adults can't learn. Jarvis says learning from scratch is easier early, but people keep learning throughout life, humans stay unusually plastic, early multilinguals learn new languages more easily, and stuttering improves with therapy in adulthood [2]. Linden notes that relief from psychiatric conditions involves plasticity and that exercise works about as well as SSRIs [1].

It doesn't mean every trait is set early. Many traits are substantially genetic: height is about 85% heritable in the United States and Big Five personality traits about 50% [1]. Beard density and baldness patterns depend on the genetically set density of DHT receptors [3]. And Linden says family upbringing explains almost none of the non-genetic part of personality, pointing instead to chance [1]. So parents shouldn't assume they can shape everything in the early years.

It doesn't mean effects are passed to grandchildren. Both Huberman and Linden reject the idea that fitness or trauma is passed down biologically in mammals, based on the current evidence [1].

Some evidence is still thin. Linden notes we don't yet know the effects of COVID during pregnancy [1], and Huberman frames the cell phone and gonad data as coming from studies of varying quality, beginning with a 2013 rat study [3].

For pregnancy, puberty and children's development, it is worth talking to a doctor or pediatrician before making decisions based on any of these findings.

What To Do With It

1. Get young children outdoors every day. Kids who spend less time outside in about their first five years are much more likely to become nearsighted, because light drives the growth factors that control eyeball length [1]. Huberman adds getting sunlight in the eyes even on cloudy days [1].

2. Expose children to languages early, ideally through peers. Speech has a stronger critical period than other skills [2], and accent comes mostly from the speech of peers rather than parents [1]. Children raised with more than one language keep a wider set of phonemes that makes later languages easier [2].

3. Start skills like music early when you can. Early ear training greatly increases the odds of perfect pitch [1], and learning piano or riding a bike from scratch is easier for a young child than an adult [2].

4. Take illness in pregnancy seriously. The 1918 flu and the IL-17 research show that maternal infection during a sensitive window can affect fetal brain development [1]. Work with your doctor on preventing and managing infections during pregnancy.

5. Avoid alcohol in pregnancy and discourage it in puberty. Alcohol in pregnancy causes fetal alcohol syndrome, and during puberty it can raise estrogenic activity in both boys and girls [3].

6. Keep cannabis away from adolescents. Cannabis appears to increase aromatase and circulating estrogen, and Huberman says during puberty it can have profound effects on these hormonal systems [3].

7. Consider limiting exposure to herbicides such as atrazine, especially around pregnancy. Huberman ties them to testicular abnormalities in frogs, falling human sperm counts and possibly earlier puberty in girls, while leaving the degree of avoidance to each person [3].

8. Provide the basics that let genetic potential show. Height and IQ are less heritable where nutrition, medical care and schooling are lacking, because people can't reach their potential [1].

9. Don't give up on adult change. Adults can still learn languages, improve a stutter with slow speech or rhythm tapping, and use exercise, which Linden calls about as good an antidepressant as SSRIs [1][2]. Jarvis keeps dancing and says movement helps keep cognition intact into old age [2].

10. Be skeptical of claims about inherited trauma or fitness. Current evidence suggests you pass these on socially, through how you raise children, not through marks on DNA [1].

Sources

1. Dr. David Linden: Life, Death & the Neuroscience of Your Unique Experience · Huberman Lab
2. Dr. Erich Jarvis: The Neuroscience of Speech, Language & Music · Huberman Lab
3. How Hormones Shape Sexual Development · Huberman Lab