Before we call someone anxious, unmotivated, distrustful, or unable to cope, how much do we know about the conditions under which that person has had to live? What has repeatedly demanded their attention? What has interrupted their sleep? What has taught them that a mistake may cost more than they can afford? And when we describe the person without describing those conditions, what exactly are we claiming to understand?

A life is experienced from the inside, but it is not produced entirely from the inside. We encounter ourselves through moods, habits, bodily sensations, expectations, and the stories we use to connect them. Over time, these can become so familiar that we stop asking where they came from. “This is simply who I am” may be an honest description of what we experience. It may also be an incomplete explanation of how we arrived there.

The first essay approached this problem through epigenetics and the possibility that environments designed around fear and attention could have consequences beyond the screen. The next step requires greater precision. How does an experience become biologically consequential? Which effects remain within one lifetime, which are recreated through relationships and institutions, and which might pass through reproduction? Where does evidence end and interpretation begin?

These distinctions matter because a warning about environmental influence can quietly become a new doctrine of inevitability. If we tell people that their circumstances have entered their biology, will they hear an invitation to change those circumstances—or a declaration that change has already become impossible?

Epigenetics gives us reasons to investigate that question. It does not answer it with a single verdict.

Our cells contain more than a DNA sequence. They also contain machinery that helps regulate how that sequence is used. Epigenetic research examines mechanisms involved in the regulation and maintenance of cellular states, including DNA methylation and modifications of the proteins around which DNA is organized. Some of these states can persist through cell division. The relationship between genetic variation, environmental exposure, epigenetic variation, and observable traits is complex; identifying a molecular difference is not the same as demonstrating what caused it or what it causes. (Amy K. Webster and Patrick C. Phillips, “Epigenetics and Individuality: From Concepts to Causality Across Timescales,” 2025.)

The familiar metaphor of DNA as hardware and the epigenome as software can help introduce the subject, but it becomes misleading if taken literally. There is no single settings menu through which experience switches a person from resistant to obedient. Nor does every response to the environment qualify as an epigenetic change.

An altered mood, a learned habit, a hormonal response, a change in gene expression, and a persistent epigenetic modification are different things. They may be connected, but the connections must be investigated. Otherwise, “epigenetic” becomes an impressive word placed wherever the explanation is missing.

Precision does not make the subject less consequential. It makes the consequences harder to dismiss.

Consider stress. A response that helps an organism meet an immediate challenge can become costly when repeatedly activated, poorly regulated, or insufficiently followed by recovery. Bruce McEwen’s work on allostatic load examined this tension: the physiological systems that help maintain stability under changing demands can also contribute to wear and tear. The relevant picture involves interacting systems and patterns over time, rather than a single substance acting as a universal toxin. Cortisol cannot adequately be understood as a chemical label for everything harmful about modern life. (Bruce S. McEwen, “Protective and Damaging Effects of Stress Mediators,” 1998.)

This suggests a more useful question than “Is stress bad?” What kind of demand is being placed on a person, how often, with what degree of control, and with what opportunity to recover?

A demanding task with a clear endpoint is different from a threat that cannot be located or resolved. A difficult day followed by safety is different from a situation in which safety itself remains uncertain. An account of experience that records only intensity may miss duration, predictability, support, and the possibility of action.

Imagine two people receiving the same instruction: “You need to manage your stress.” One can reduce commitments, sleep in a quiet room, and take time away without losing income. The other works unpredictable hours, shares an overcrowded home, and supports relatives who depend on every payment. The instruction is identical. The available choices are not.

At what point does advice about self-regulation become a way of avoiding a discussion about the distribution of demands?

This is where the biological question becomes inseparable from the social one. We should investigate how conditions affect people, but we should also ask who has the power to modify those conditions. A workplace may measure an employee’s errors while leaving its own scheduling practices outside the evaluation. A school may assess a child’s concentration without asking what kind of night preceded the school day. A platform may count engagement without counting what that engagement displaced.

The resulting measurements may be accurate within their narrow frame. The frame can still omit part of the cause.

The Dutch Hunger Winter provides a concrete example of why early conditions deserve serious attention. In a study published in 2008, Bastiaan Heijmans and colleagues compared people exposed to famine around conception with their unexposed siblings of the same sex. Roughly six decades later, they found differences in DNA methylation at the IGF2 locus. The finding linked an early exposure with a persistent molecular difference. It did not, by itself, establish that this difference caused every later health outcome associated with famine exposure. (Bastiaan T. Heijmans and colleagues, “Persistent Epigenetic Differences Associated with Prenatal Exposure to Famine in Humans,” 2008.)

The distinction between a trace and a mechanism is essential. A mark can be associated with an exposure without accounting for all its consequences. It might participate in a causal pathway, reflect another process, or coexist with other changes. To discover which explanation applies, we need more than the fact that a difference can be measured.

Yet even this carefully bounded result changes the kinds of questions we can ask. When does an event end? When the immediate shortage is over? When the person leaves the dangerous place? When a government declares the emergency finished? Biological investigation can reveal that the calendar of exposure and the calendar of consequences need not coincide.

That possibility should expand our understanding of responsibility without allowing us to invent consequences that have not been demonstrated.

The same discipline is necessary when discussing inherited effects.

In the widely discussed mouse study by Brian Dias and Kerry Ressler, male mice underwent conditioning that paired a particular odor with mild foot shocks before reproduction. Subsequent generations showed increased sensitivity to the conditioned odor, alongside changes in the relevant olfactory pathway; the researchers also reported differences in sperm DNA methylation. These findings concern a specific experimental system and particular behavioral, anatomical, and molecular measurements. They do not show that descendants inherited a conscious recollection of an ancestor receiving a shock, nor do they establish that human political obedience can be transmitted in this way. (Brian G. Dias and Kerry J. Ressler, “Parental Olfactory Experience Influences Behavior and Neural Structure in Subsequent Generations,” 2014.)

What crosses a generational boundary, if something does, need not be a memory in the everyday sense. A change in sensitivity is not a narrated experience. An altered response to a stimulus is not an inherited explanation of what happened. If we erase those distinctions, we turn a demanding scientific question into a story whose apparent clarity comes from saying more than the experiment measured.

The human evidence requires particular care. Rachel Yehuda and colleagues examined FKBP5 methylation in a small sample of Holocaust survivors and their adult offspring, alongside comparison groups. The study involved 32 survivors and 22 offspring, with smaller parental and offspring control groups. It reported differences associated with parental Holocaust exposure. It did not follow children from birth to demonstrate that they were born with a specific inherited molecular state, and its design could not by itself establish transmission through reproductive cells as the explanation. (Rachel Yehuda and colleagues, “Holocaust Exposure Induced Intergenerational Effects on FKBP5 Methylation,” 2016.)

This limitation does not make the experience of descendants unimportant. It clarifies the question of how that experience is produced.

A child may be affected by conditions during pregnancy, by the emotional availability of caregivers, by family stories and silences, by material deprivation, by discrimination, or by dangers that never ceased. These are different proposed routes of influence. Identifying one does not automatically eliminate the others, and describing an outcome as “intergenerational” does not specify its mechanism.

There is also a basic problem of exposure. During a pregnancy, the pregnant individual, the fetus, and the fetal germ cells that may contribute to a later generation can all be within the directly exposed biological system. In such a design, effects in children and grandchildren do not alone establish transmission to an unexposed generation. Researchers therefore distinguish intergenerational effects from transgenerational inheritance, with the relevant generation depending on the exposure route. Mammalian development also involves extensive epigenetic reprogramming, making persistence across generations a question to demonstrate, not a default assumption. (Maximilian H. Fitz-James and Giacomo Cavalli, “Molecular Mechanisms of Transgenerational Epigenetic Inheritance,” 2022.)

These distinctions may sound technical, but they protect a much larger conversation from confusion. Without them, several generations living under related conditions can be mistaken for proof that a molecular message traveled unchanged through reproduction.

Are descendants carrying an ancestral burden, or are they being placed under a similar burden again? Could both processes operate in a particular case? What evidence would separate them?

The answer matters for what we decide to change. If a neighborhood remains unsafe, an explanation centered exclusively on ancestral trauma may turn our attention away from present violence. If insecurity is reproduced through employment, housing, or exclusion, describing it only as biological inheritance may conceal the institutions still participating in its production.

The past can matter without relieving the present of responsibility.

We should also be cautious about whose responsibility becomes visible. Discussions of pregnancy, parenting, and inherited effects can place extraordinary moral weight on parents, especially mothers. A concern about children’s development can become an accusation that every difficult emotion or imperfect circumstance has damaged the next generation.

But what choices were actually available? Who controlled the working hours, the living conditions, the access to care, or the threat of violence? How much of what is described as a parental failure was a failure to support the parent?

A serious account of influence must follow the chain of conditions far enough to avoid locating every cause in the person closest to the child.

Care itself deserves attention in this discussion. In a 2004 rat study, Ian Weaver and colleagues linked differences in maternal care with differences in hippocampal regulation involving a glucocorticoid receptor gene. Cross-fostering helped distinguish the rearing environment from biological parentage, and a pharmacological intervention altered some of the measured differences. This was evidence from a particular animal model, not a formula for human parenting or a promise that any difficulty can be reversed. It nevertheless showed why the relationship between experience and biology should be investigated in terms of mechanisms and possible change, rather than treated as a one-way inscription of fate. (Ian C. G. Weaver and colleagues, “Epigenetic Programming by Maternal Behavior,” 2004.)

The broader question is worth carrying forward. If we take seriously the possibility that harmful conditions matter, why would we refuse to investigate whether supportive conditions matter too?

The answer cannot simply be that everything is reversible. Some effects may persist, some interventions may fail, and timing may matter. But permanence also requires evidence. A biological explanation does not become stronger by making the future disappear.

This brings us back to the digital environment.

A screen can carry a message, emit light, interrupt an activity, replace another activity, or connect a person with support. These are different features of use. Treating them as a single exposure called “technology” obscures the mechanisms we need to understand.

In a controlled study comparing evening reading on a light-emitting electronic device with reading a printed book, Anne-Marie Chang and colleagues found differences in sleep onset, melatonin secretion, circadian timing, and alertness the following morning. The result concerns a specific exposure under experimental conditions. It provides evidence for a pathway involving evening light and sleep; it does not demonstrate that every form of screen use has the same effect or that the effect is epigenetic. (Anne-Marie Chang, Daniel Aeschbach, Jeanne F. Duffy, and Charles A. Czeisler, “Evening Use of Light-Emitting eReaders Negatively Affects Sleep, Circadian Timing, and Next-Morning Alertness,” 2015.)

A separate question concerns the experience of using a platform. In a randomized experiment, Hunt Allcott and colleagues studied four weeks of Facebook deactivation before the 2018 United States midterm election. Deactivation improved subjective well-being on average and changed how participants spent their time; it also reduced factual news knowledge. The study provides causal evidence about outcomes in a defined setting, including tradeoffs. It does not establish a universal effect for all users, all platforms, or all periods, and it did not demonstrate an epigenetic mechanism. (Hunt Allcott, Luca Braghieri, Sarah Eichmeyer, and Matthew Gentzkow, “The Welfare Effects of Social Media,” 2020.)

Why insist on that last sentence? Because evidence at one level should not be promoted into evidence at every level.

A change in well-being does not automatically establish a hormonal pathway. A hormonal change does not automatically establish an epigenetic pathway. An epigenetic difference does not automatically establish transmission to descendants. And none of these findings, alone, establishes deliberate social control.

Each transition asks a new question. Each requires its own evidence.

This does not leave us unable to criticize harmful design. It gives the criticism a clearer target.

Imagine a hypothetical service rewarded primarily for keeping people engaged. Suppose a particular design increases usage while users experience costs elsewhere in their lives. The service’s success measure and the user’s well-being may diverge. We can investigate that divergence without first proving that the designers intended every consequence.

Intention, incentive, knowledge, and responsibility are related but distinct. An effect may begin as unanticipated, become observable, and later be tolerated because changing it would reduce revenue. Alternatively, an apparent harm may have another cause, or a design change may produce benefits for some people and costs for others. These possibilities should direct investigation, not be settled in advance by a preferred story.

The important questions become concrete. What is being optimized? What is being measured? Which costs appear in the evaluation, and which are transferred outside it? When evidence of harm emerges, who has the authority to respond, and who has a reason not to?

A system need not possess a unified intention for its incentives to produce a recurring pattern. Equally, a recurring pattern does not by itself prove a hidden plan.

This is a more demanding account of power than the image of a single actor controlling everyone. It asks how decisions, procedures, commercial goals, and ordinary habits can combine. It also leaves room for responsibility at each point where someone could reasonably know, choose, or intervene.

Consider the phrase “the user chose to stay.” It may describe a real action. Does it provide a complete account of the conditions under which the action occurred?

To answer, we would need to examine what the person understood, what alternatives they perceived, what obligations kept them connected, and how the service was organized. A formal option to leave does not tell us everything about the practical cost of leaving. A worker, caregiver, student, and socially isolated person may encounter the same interface through very different dependencies.

This does not erase choice. It asks what kind of choice exists.

Freedom becomes a more substantial question when we examine the conditions needed to exercise it. Having an option is one thing; having the time, security, attention, and resources to evaluate and use it is another. If we discuss only the first, we can congratulate ourselves on providing freedom while leaving the second unequally distributed.

What would it mean to assess a technology by the quality of the choices it makes possible, rather than only by how often people choose it?

The same question applies beyond technology. An institution may offer a wellness program while leaving unpredictable schedules untouched. It may teach resilience while rewarding constant availability. It may ask people to become better at recovering without examining why recovery time is repeatedly removed.

There may be value in the program. The unresolved question is whether it addresses the source of the demand or helps people endure that demand longer.

When does resilience expand a person’s freedom, and when does it become an expectation that the person absorb costs someone else has decided not to change?

This is not a reason to reject individual skills. It is a reason to evaluate them alongside the conditions in which they are used. A useful coping strategy and an unjust arrangement can coexist. Helping someone survive the arrangement does not settle whether the arrangement should continue.

The biological language of adaptation needs similar care. A response may be understandable in relation to a threatening environment without being desirable in every other environment. The word “adaptive” does not mean painless, morally good, or permanently appropriate. It points us toward a relationship between a response and the conditions under which that response makes sense.

Suppose a person has learned to check repeatedly because overlooking something once carried serious consequences. In a safer situation, others may see only the checking and call the person controlling. Suppose another person avoids disagreement because disagreement previously brought punishment. Others may call that person passive. These are illustrative possibilities, not diagnoses. They remind us to ask about the history of a behavior before treating it as a complete explanation of character.

What looks like a defect from outside may require a history to become intelligible.

Understanding that history does not excuse every action or remove responsibility for harm. It changes the questions we ask about how change might become possible. Blame alone cannot tell us what a behavior is protecting, what sustains it, or what alternative the person could realistically learn to trust.

Here the boundary between self and environment becomes difficult to draw. We do not experience a habit as an external policy. We experience it as the way we reach, hesitate, react, anticipate, or withdraw. Once a pattern becomes familiar, its origin can disappear from the story we tell about it.

The result may be a mistaken permanence: “I have always been this way” becomes “I can only be this way.”

But a history is not the same as an essence. Explaining how something developed does not establish that it could never develop differently. Nor does the possibility of change tell us that change will be easy, equally available, or entirely under individual control.

This distinction matters especially when biological terms enter self-understanding. “There may be a biological component” can offer relief to someone whose suffering has been dismissed. It can also be heard as “this is fixed.” The same vocabulary can validate an experience or close its future, depending on what we allow it to imply.

A measured biological difference is not a moral judgment. It is not a complete description of a person. And it is not, without further evidence, a prediction of everything that person can become.

We should also question why suffering is sometimes treated as more legitimate once a molecular explanation is proposed. Does exhaustion become worthy of concern only after it can be attached to a biomarker? Does an unsafe workplace become a problem only after a particular biological mechanism has been established?

Scientific explanation is valuable. Moral recognition should not require a person to produce a molecular certificate of what they have endured.

There is a danger in allowing the authority of biology to replace attention to experience. We may appear to be taking people more seriously while listening to them less, waiting for a test to certify what deserves our concern. We may also allow uncertain biological claims to carry ethical arguments that could have been made honestly without them.

A person’s need for rest, safety, dignity, and meaningful participation does not become important only when epigenetics is mentioned.

This is why the strongest continuation of the original concern cannot be a claim that screens have permanently rewritten humanity. It must be an investigation of how particular environments affect particular people through particular pathways—and of who can change those environments.

The first task is to separate the pathways. Is the proposed effect physiological, behavioral, relational, developmental, or reproductive? Is the evidence observational or experimental? Was the outcome directly measured, inferred through a proxy, or assumed? Could the same result arise through another route?

The second task is to locate the opportunities for intervention. If interrupted sleep is part of the problem, what produces the interruption? If economic insecurity is central, what would a change in security test? If a platform feature is suspected, can that feature be altered while other conditions are compared? If a pattern is recreated through caregiving, what support would make a different pattern possible?

These are research questions as well as practical ones. They prevent “change your life” from becoming an instruction detached from the conditions needed to carry it out.

The third task is to evaluate improvement at the level we actually care about. A molecular measure may be informative, but it should not automatically replace questions about how someone sleeps, functions, relates to others, or experiences their life. A change in a marker and a meaningful improvement are not interchangeable claims; their relationship must be established.

Otherwise, we risk creating a new version of the same problem: measuring a convenient variable and treating the number as the person.

At a collective level, this suggests a different standard for evaluating environments. We can ask whether they permit recovery, make expectations understandable, provide meaningful control, and allow people to seek correction without disproportionate cost. These are proposed criteria for judgment, not promises that one institutional design will produce the same outcome everywhere.

The point is to make conditions part of the evaluation rather than treating them as a fixed background against which individuals succeed or fail.

For the individual, the corresponding inquiry can begin without a theory of inherited damage. Which reactions do I repeatedly experience, and in what settings? Which demands can I influence? Which require help, cooperation, or institutional change? What happens when a condition changes? What would count as evidence against the story I currently tell about myself?

These questions do not require us to deny the reality of a limitation. They ask whether we understand its boundaries.

A person may discover that some difficulties persist across settings and need sustained support. Another may find that a supposedly personal inadequacy diminishes when a specific demand is removed. Many lives may involve combinations that resist a simple explanation. The purpose of inquiry is to remain open to those differences.

This is also where the discussion returns to the self-fulfilling prophecy. An explanation of how conditions shape people can itself become one of those conditions. Tell someone that their future has been permanently written into their biology, and the explanation may influence what they attempt, what support they seek, and what change they consider worth demanding.

The warning can begin to help produce the resignation it describes.

We should therefore ask this essay the same questions it asks of other forms of influence. What expectation does it create? Does it help the reader distinguish evidence from possibility, or merely replace one frightening certainty with another? Does it make responsibility more visible, or distribute blame so widely that no one remains answerable? Does it open a question about change, or imply that the most important decisions have already been made?

The answer should remain subject to revision.

Epigenetics can enrich our understanding of how biological regulation relates to experience. It cannot carry, by itself, a complete theory of personality, a diagnosis of society, an explanation of political submission, and a verdict on future generations. Asking it to do all of these things weakens the very inquiry it could help strengthen.

The more consequential possibility is also the more demanding one: some of what we treat as private limitation may have a history that extends beyond the individual, while the ways to address it may require both personal and collective action.

That possibility does not tell us in advance which explanation is correct in every case. It tells us where to look more carefully.

Before we call a child incapable, an employee unmotivated, a parent inadequate, or a population apathetic, what have we learned about the conditions under which their capacities are being tested? What opportunities have been withheld? What costs have been normalized? What would we discover if we changed part of the situation before concluding that we had discovered the nature of the person?

And when the person being judged is ourselves, can we allow the same questions?

We may never be able to separate completely what we brought into the world from what the world has made of us. But we can become more careful about treating the present arrangement as the final answer.

Before we ask people to become different, have we asked what in their world must change so that being different becomes possible?