I trained as an exercise physiologist, which means I grew up on one idea: the body adapts to what it is asked to do. Ask a muscle for more and it changes. Stop asking and it changes back.
So when people began telling me that stress, trauma and even grief can leave chemical marks on our genes, and that the marks can be undone, I had two reactions at once. The first was recognition. The second was suspicion, because it is exactly the kind of idea that gets sold before it is understood.
This essay is my attempt to hold both. What is epigenetics? What are researchers actually looking at? How may it affect us? And how much of it can we change? It is a companion to What the Body Carries, which touches the inheritance question briefly, and to The Load That Was Never Yours, which asks what we do with what we carry.
What It Is
Same recipe book, different dishes.
Almost every cell in your body carries the same DNA, yet a heart cell and a neuron behave nothing alike. The difference is which genes are active. Epigenetics is the study of how cells control gene activity without changing the DNA sequence.
Two mechanisms get most of the attention. In DNA methylation, small chemical groups are added to DNA. Depending on where those groups attach, they can influence how actively a gene is expressed. Methylation near a gene's regulatory regions often reduces its activity, but the relationship is not always that simple. In histone modification, DNA is wound around proteins called histones, and chemical groups added or removed change how tightly it is wound, which affects whether nearby genes can be read. Together these marks make up the epigenome. It varies among individuals, among tissues, and even among cells within a tissue, and environmental factors such as diet and pollutants can affect it.
One distinction runs through everything below. Marks that change within one person's lifetime are one thing. Evidence that such changes pass through reproductive cells to later generations is a separate claim, and a much thinner one.
A useful picture, and only a picture: the genome is the recipe book, and the epigenome is a set of dimmer switches that decide which recipes are turned up. Most of the dimmers are set during development, when a cell becomes a heart cell or a neuron, and that has little to do with how your childhood went.
Two cautions to carry through the rest of this essay. First, a mark is a measurement, not a meaning. A difference in methylation does not by itself tell us how a person will feel, behave or live. Second, where the sample comes from matters. Most human studies measure blood, saliva or cheek cells, which may not reflect what is happening in the brain.
What Researchers Are Looking At
Five questions, and how far each has got.
| Question | Best evidence | What it shows | Main limit |
|---|---|---|---|
| Can early care set the dial? | Rats: pups of mothers who licked and groomed more differed in methylation at a stress-receptor gene, a difference set in the first week of life (Weaver 2004) | Maternal behavior can shape the epigenome in rats | Rats. It shows a mechanism is possible, not that it works the same way in people. |
| Does early life leave marks in people? | Postmortem hippocampus from suicide victims with a history of childhood abuse: less glucocorticoid-receptor expression and more methylation of its promoter (McGowan 2009) | An association between childhood abuse and a mark in the brain | Postmortem tissue, a single snapshot. It cannot show which came first. |
| Does a hard environment before birth leave marks? | Dutch Hunger Winter: people exposed in the first weeks after conception had lower methylation of the IGF2 gene six decades later than their unexposed siblings | A lasting effect on the exposed person | Whether the mark affects health is not established. |
| Can marks pass to the next generation? | Mice conditioned to fear an odor; Holocaust-offspring studies in people | Shown in mice. Associations in small human samples. | Preliminary in humans. A 2023 review called the support preliminary, a 2026 review found no independent replication, and critics note that mammals erase most methylation in the germline. Details are in the companion essay. |
| Can a methylation pattern measure how fast we age? | Epigenetic clocks: methylation patterns that track age and are used to estimate "biological age" | Clocks help predict age and risk across populations | They may be consequences of aging, not causes. A clock is a snapshot, and a better score does not by itself show slower aging (2025 critique). |
How It May Affect Us
A set of possibilities, each with a string attached.
This is where the popular version runs ahead of the evidence, so I want to be careful. What the research supports is a short list of possibilities.
- A stress system tuned differently. The most-studied genes, NR3C1 and FKBP5, are involved in how the body responds to stress hormones. Differences in their methylation have been linked to early adversity, trauma and treatment response in small studies. The direction is not simple. In the Holocaust-offspring replication, lower methylation at one FKBP5 site went with lower self-reported anxiety.
- Health risks that start early. The Hunger Winter shows that an early environment can leave a measurable mark decades later. Whether such marks change anyone's health is a separate question the research has not settled.
- Aging. Clock scores are linked to age-related risk across populations, but they may reflect aging more than drive it.
- Adaptation as well as damage. Some changes may prepare a person for the environment they were born into. A 2024 review of prenatal stress in humans raises this adaptive view, and notes that robust, replicated findings are still scarce.
What none of this says: that a feeling can be inherited as a feeling, that a mark equals a diagnosis, or that your childhood fixed your epigenome for life. Marks respond to experience by their nature. That is the whole point of them.
Can We Change It?
The marks can change. Whether changing them improves our lives is a harder question.
If marks respond to experience, then the real questions are how much, how fast, and whether it matters. Here is the evidence, from the most controlled setting to the least.
| Approach | Best evidence | What it showed | Main limit |
|---|---|---|---|
| Early care, and a drug, in rats | Weaver 2004: differences set by maternal behavior were reversed by cross-fostering in infancy, and a histone deacetylase inhibitor infused into the brain removed the group differences in methylation and stress responses | In rats, an early-set mark was reversible | Rats. This is a laboratory tool, not a treatment for people. |
| Psychotherapy | Yehuda 2013: 16 combat veterans with PTSD, 12 weeks of prolonged exposure therapy. FKBP5 methylation fell in responders and rose in non-responders, mainly between the end of treatment and follow-up. | A mark moved along with symptom change | Sixteen people, and responders differed from non-responders at baseline. The authors call it preliminary. A 2024 family-therapy trial, as described in a 2026 review, saw cortisol fall but no epigenome-wide differences survive correction. |
| Diet, sleep, movement and relaxation | Fitzgerald 2021 pilot trial: 43 healthy men aged 50 to 72, eight weeks of diet, sleep, exercise and relaxation guidance plus probiotics and phytonutrients. The treatment group scored 3.23 years younger on the Horvath clock than controls (p = 0.018). | A bundle moved a clock score | A small pilot of men only, measured in saliva with one clock. The bundle cannot separate its parts. The paper was corrected twice over how within-group results were reported, and the between-group result did not change. A clock score is not a health outcome. |
| Physical activity | A meta-analysis of 44 studies and 145,465 people: higher activity was generally linked to lower methylation age. Per standard deviation of weekly activity, GrimAge acceleration was lower by 0.09 and Horvath by 0.03, with no significant link for two other clocks. | Small associations | Mostly cross-sectional, so direction is unknown. The authors call for trials. |
| Mind-body practices | A 2017 review of 18 studies and 846 people, mostly observational, with fewer than a third randomized | Preliminary signals in gene activity | Gene expression is not the same as changing DNA. Mindfulness trials also show publication bias: 87% of published trials reported a positive outcome, against roughly 66% expected. |
Read down the table and the pattern is consistent. The mechanism is shown in animals. In people there are promising small studies, almost all measuring marks in blood or saliva, and few measuring anything that matters to health. Among the studies I reviewed, I found no study showing that changing a mark changes a person's life.
I think about one client more than any study. He came to me not long after a serious diagnosis and wanted an exercise routine. We started modestly. Before long we were doing long bike rides, and he was trying parts of fitness he had never tried in his life. In the last year he was alive, he got into the best shape of his adult life, until the illness took over.
None of that changed the diagnosis. His fitness improved while the larger situation of his life stayed hard, and I would not tell you his cells were healed, because I never measured anything that deep. I cannot say whether any epigenetic mark moved. What I can say is that something real changed for him, week after week, because of what he did.
That is why I care about the line between a mark that changes and a life that changes. They are not the same thing. The second can be real, and worth having, without the first being proven.
So here is what I would do with all this. It is partly inference, and I have labeled it so.
- Do the things for their own sake. Sleep, movement, food, relationships and, where needed, therapy have benefits that do not depend on epigenetics being the mechanism. Epigenetics may explain some of it. It is not the reason to do them.
- Aim at the source, not the mark. If a stress system is tuned high, the reasonable bet (my inference) is to lower what keeps it running: chronic sleep loss, isolation, unsafe or chaotic conditions, and unresolved trauma.
- Get help with the heavy parts. The best evidence for therapy is that it helps symptoms. Its effects on marks are early. Bring someone steady for the trauma.
- Be wary of anyone selling to reprogram your DNA. A biological-age score is a snapshot that may not track cause, and a change in gene activity after a retreat or a course is not a change in your DNA.
Where This Goes Wrong
- Fatalism. "It's in my genes" turns a tendency into a verdict. The evidence does not support that.
- Blame. "I should have fixed my marks" turns an interesting hypothesis into a new load. Nobody is at fault for their epigenome.
- Hype. This field is young, and its results are moving. Several details here come from abstracts and secondary summaries, so check the originals before relying on any one number.
A dimmer switch is not a verdict. It is a setting, and settings may be adjusted, slowly, under the right conditions. That is a smaller promise than the one on the supplement label, and I find it easier to believe.
We may never know how much of the change we make in ourselves reaches all the way down to the molecules that regulate our genes. Perhaps some of it does. Perhaps much less than we hope.
But we do not have to wait for that answer.
We can sleep better, move more, build relationships, ask for help, and change the conditions in which we live. Not because we have learned to control our biology, but because we have learned that our biology is part of a life still being lived.
The dimmer switches are interesting. What we do with the light matters more.