15 Epigenetics in Public Health

2.1. “Ignoring Biology”: The Historical Rise of Social Epidemiology

According to Olga Amsterdamska, a historian of epidemiology, “Throughout all its complicated history, it seems epidemiologists expressed a perennial worry that epidemiology is not quite ‘up to snuff’ as a science because of its anomalous position as a nonlaboratory (and not a clinical medical) discipline and because of its continuous involvement with the politics and practice of public health” (Amsterdamska 2005, 18). While this concern can be traced back to the bacteriological revolution (Parascandola 1998), this paper examines how efforts to defend the “scientificity” of epidemiology have shaped the discipline and, in turn, extended its social and political ambitions.

In the latter half of the twentieth century, epidemiology changed its focus from stopping the spread of contagion, and became more concerned with the prevention of chronic diseases.6 Epidemiologists exploring these new types of epidemics, of completely unknown origin, returned to observation. They resorted to descriptive studies of disease distribution, proceeded to exploratory sweeps for potential risk factors, and developed new tools like case-control and cohort. A new conception of disease causation emerged, relying on complex statistical analysis that enabled epidemiologists to relate environmental/behavioral exposure to health outcomes. Such an etiology had two main advantages: numerous risk factors could be connected, within a web of causation (MacMahon and Pugh 1970), in the absence of any prior understanding of pathogenesis. Cancer studies illustrate this approach particularly well: For years, the correlation established by epidemiology between lung cancer and smoking,7 or lung cancer and arsenic exposure in industrial settings (Landrigan 1983), lacked an animal model of carcinogenicity.

Because multifactorial epidemiology relied on a model that intended to calculate risks via correlations, with no search for evidence of underlying mechanisms (Peto 1984), the black box metaphor was borrowed from engineering (Cauer 1954) to describe the paradigm. A black box is a device, system, or object that can be viewed in terms of its input and output characteristics, without any knowledge of its internal workings. Applied to epidemiology, the paradigm may be summarized as follows: “Black box thinking labels the methodologic approach that ignores biology and thus treats all levels of the structure below that of the individual as one large opaque box not to be opened” (Weed 1998).

Although dimensions of social life fail to display the sort of universality and uniformity required by earlier models to qualify them as causes for diseases (Broadbent 2009), the acceptance of the black-box paradigm enabled social epidemiology to overlook strictly biological concerns and focus instead on the social distribution of states of health in populations (Susser 1973). It revived the historical mission of the discipline—to improve the living conditions of the poverty-stricken—through an innovative scientific agenda meant to understand the structural patterns of social inequalities of health.

However, a question remains as to how causal inference is assessed when the only evidence provided is correlation. The identification of causal connections between “exposure” and “outcomes” is de facto resolved by the calculation of risk factors. In the black-box model, a study showing that a risk factor is unlikely to be due to chance, and eliminating cofounding variables, can build a good case that the risk factor is causal (Broadbent 2015). To put it simply, the causal sequence that links exposure and disease is postulated, not proven. The shift in perspective, from one model of causality to the other, definitely derives from the complex nature of the social factors studied.

2.2. The Black-Box Controversy: Tension in Epidemiology over a Social Mission and a Quest for Scientific Legitimacy

The black-box paradigm became the prevailing model of chronic disease epidemiology until the late 1980s. At that time, some epidemiologists noted that if their discipline were restricted to a pure quest for correlation, devoid of the ambition to propose general explanations, it would be exposed to a serious threat in the competitive landscape of the medical and life sciences. A lively controversy ensued between proponents and opponents of the black-box paradigm. Debates involved radically different conceptions about decisive issues such as the methods used to prove causation in epidemiology and the status of biological evidence regarding other scientific results. Epidemiologists who defended the integration of molecular studies within epidemiology—that is, those who wanted to move beyond the black-box model and to involve the living machinery of the bodies—typically developed three types of arguments.

2.3. The Role of Epigenetics in Providing Biological Evidence to Support Social Claims

The charge against black boxing, as well as the epistemic, strategic, and pragmatic benefits to be gained in developing molecular studies within epidemiology frameworks, explains epidemiologists’ interest in epigenetics. This branch of genomic research developed in the early 1990s to examine the molecular pathways between environments and biological processes.9 The position of epigeneticists, who share experimental methods, analytic technologies, and laboratory practices with molecular biologists but address theoretical questions that may inform any investigation of body-environment interactions, is of particular relevance for the academic needs of epidemiology.

Practically, the integration of epigenetics into epidemiology allows for hypotheses developed via statistical correlations to be simulated and lab tested. In the spirit of the Hill’s criteria for causation, which stipulate the minimal conditions necessary to provide adequate evidence of a causal relationship between an incidence and a consequence (Hill 1965), summoning biological evidence could be interpreted as a way to strengthen epidemiology’s case for causality. Epigenetic evidence adds biological plausibility to an epidemiological correlation by ensuring the temporality and specificity of the causal relation; arguing on the coherence between epidemiological and laboratory findings; and presenting experimental evidence from animal and human studies.

Interestingly, when epidemiologists adopt epigenetics approaches, they change their vocabulary and start speaking of environment-disease “pathways” instead of “associations” (Relton and Davey Smith 2012). The semantical change reflects profound shifts in the sort of knowledge epidemiology seeks. In physiology, the term pathways is applied to the routes formed by a chain of nerve cells; in biochemistry, pathways are the chains of chemical reactions. In both cases, the metaphor of the path emphasizes the continuity of the causal sequence and the materiality of the route. In epigenetics, pathways connect isolated elements, chosen for their significance relative to a problem, and the act of tracing them constitutes a path. This representation entails two consequences.

• To follow the molecular pathway leading from social behaviors to DNA components enables researchers to move from one proximal cause to the next, without concern for breaks (such as threshold effects). It obviates the need to account for the heterogeneity of the entities involved and their specific type of agency.
• The materiality of the path also ensures that the connection between exposure and health outcomes is perceived as substantial. In the conception called “biological embedding,” all entities, intra- or extracorporeal, belong to the same network of molecules and interact with each other, so that a certain environment may be incorporated and a certain experience may pass under the skin.

The pathway metaphor is all the more striking because it allows epidemiological discourse to shift from the alleging correlations to an exploration of the material reality. The subsequent paragraph illustrates how, boosted by the explanatory power of molecular biology, epidemiological claims gain substantial credibility to support public health interventions.

New insights from epigenetic research might influence policy and practice—for example by enabling stratification of populations (. . .); or by demonstrating mechanisms by which environmental and social factors influence disease, thus providing stronger evidence for policy makers and for civil society. Indeed, showing how epigenetic changes are linked to specific exposures could be more powerful than statistical associations between risk factors and NCDs (noncommunicable diseases), and might lead to innovative legislation. (Vineis et al. 2014)

But how is the new focus on molecular mechanisms consistent with the conception of the “social” developed during the black-box era? Put differently: If the black box allowed the rise of social epidemiology as a discipline analyzing the structural aspects of societies, how will the new focus on biological mechanisms influence epidemiology’s conception of “the social”?

This epistemic turning point asks two sets of questions:

1. The black box emerged as a model emphasizing the social dimension of risk factors. Within the new model, therefore, the conception of the “social” is rearranged. Hence, social epidemiologists who integrate epigenetic studies in their framework must align two conceptions of “the social”:
   • “the social” operationalized as a structural risk factor in statistical operations and
   • “the social” tested as an environment in experimental settings.
2. The epistemic operation whereby a scientific theory is explained by a theory from a different branch of science is termed “reduction” (Schaffner 2001). According to this definition, the use of biology to support epidemiological claims deserves to be called a reduction. The effects of social macro-structures on the distribution of states of health are indeed beginning to be explained by biological embedding, which is a theory of epigenetics. But reductions are delicate operations, because they result in informing our understanding of the behavior of entities pertaining to a specific field by our understanding of the behavior of the components of these entities.10 The main problem for us is that for the scientific reduction to work, the social would need to be apprehended under physical-chemical laws, or to put it bluntly, human behavior would have to be conceptualized through the same rules of physics that explain the behavior of molecules. But are the conceptions of agency and causation engaged at these different levels compatible?

With epigenetics, epidemiology seems ready to count social causes as risk factors for biological events and to show how biological processes contribute to explaining social behaviors. The way the causality engaged in biology is articulated with causality involved in the social sciences deserves to be questioned.

3.1. How Does “the Social” Become an Empiric Variable in Social Epidemiology?

Classical epidemiology treats the “social” as a layer of variables. Data are recorded through questionnaires, stocked in databases, and made available for quantitative analysis. The term social is applied to any trait that is not (patho-)physiological. In practice, though, the “social” may even designate a generic context of difficult-to-grasp variables associated with physiological or molecular mechanisms (Bauer 2011). The lack of definition explains why the use of proxy variables culminates when considering social features. In classical epidemiology, because social variables are perceived to lack specificity compared to physiological variables, they are considered to be so similar to one another as to be almost interchangeable. Proxy variables make it possible to infer the value of variables that could be relevant, but that are missing. For instance, data concerning income are currently used in place of data concerning education, etc.

Social epidemiology, however, brought precision to the definition of the “social,” in regard to three main aspects:

• The social as a cause: In direct contrast to classical epidemiology, in which “the social” is generally treated as “noise,” in need of adjustment to eliminate its potential biasing effects, social epidemiology considers the influence of the “social” on health as a determinant of health per se, and the hypothesis to be tested.
• The social as a structure: Whereas the classical conception of “the social” tends to focus on individual behaviors (diet, lifestyle, smoking), social epidemiologists consider the social on the structural level: that is, as aspects of social organization that inevitably disadvantage a proportion of the population and negatively affect their health; or, conversely, confer advantage and benefit health. Such aspects include the distribution of income and wealth; the way the production and distribution of life necessities are organized, housing quality, and the provision of medical care (Blane et al. 2013). The conceptualization of “the social” as structural pervades the design and terminology of the investigation:
  • Social epidemiologists develop composite categories meant to capture the complexity of social life, whose putative environmental mediators are correlated with one another. For instance, the “socioeconomic status” (SES) combines a person’s work experience and an individual’s or family’s economic and social position in relation to others—based on income, education, and occupation.
  • Social epidemiologists test the relation between economic inequality and the health of the population to determine if SES is a factor in health inequity. This correlation suggests that it is not only the poor who tend to be sick when everyone else is healthy, but that there is a continual gradient, from the top to the bottom of the socio-economic ladder, relating status to health. This phenomenon is often called the “SES Gradient.”11 Gradients are produced to reflect the relative social position of a population subgroup in terms of access to resources (Marmot 2003).
  • Results are interpreted according to a classification of the collectives differentiated by the prior operations (socioeconomic variables; gradient) and come to be expressed in terms of “social class,” “social position,” or “social status.” In the absence of any collectively elaborated framework attesting to the degree to which epidemiology is informed by social theory, the structuralist stance that emerges from the concepts used by social epidemiologists should be interpreted as pragmatic, based on the need to work with clear-cut indicators and to connect a scientific investigation with the social mission of the discipline.
• The social is reflexive: Operative concepts such as social position, gender, or health are regularly used with respect to their reflexive dimension.
  • The concept of “perceived social status” shows that the inhabitants’ interpretation of their social and material environment plays a key role in human development, alongside the role played by this environment itself.
  • The concept of “perceived health” emphasizes the importance of the subjective and social dimensions to categorize and treat experiential events such as growing up, aging, or being sick or disabled.
  • Using the category of “gender” instead of “sex” allows for the inclusion of sex-based social structures and identity dimension within the studies. Epidemiologists, therefore, increasingly replace the discrete concept of “sex” with complex formulations, such as Nancy Krieger’s notions of “biological expression of gender” and “gendered expressions of biology” (Krieger 2003).

In summary, the social turn of the discipline invites researchers to cultivate a broad perspective of “the social” and to document both its structural and reflexive dimension via the collection of institutional facts—that is, facts that exist because we make sense of them in our social life and because they reflect status-functions (Searle and Willis 1995). The structure of social reality according to social epidemiology is thus understood as a hierarchy in which social facts are determined by a collective intentionality. This representation carries limitations, both internal to the methodologies of epidemiology (a); and regarding its articulation with other scientific frameworks (b).

1. To fit quantitative research, “the social” must be defined as a factor for potential outcomes. The operation that allows for “the social” to be framed into a variable has two consequences: The thickness of social events is domesticated to fit within categories; and some aspects of social life that escape notice may fail to be captured as health determinants (Murphy 2004).
2. Institutional facts are highly specific. They are facts by human agreement, and are ontologically subjective, because they do not exist unless they are agreed upon. There is no physical-chemical formula to describe them. It does not necessarily follow from this conception of social variables as institutional facts that their existence never relates to other kinds of things. But the institutional nature of such facts has to be borne in mind when considering a rapprochement with biology. The variable of “income,” for instance, is ontologically different from the physiological features of an organism. Therefore, it would seem to be impossible to simulate this variable in an animal study. This, however, is precisely what the articulation of social epidemiology and epigenetics requires.

5.1. Epidemiology

For epidemiology, tragedies such as wars and famines are excellent sources of data because they are extreme enough to leave their mark on the bodies of those who experienced them. Data collected from individuals who were in utero during the Dutch hunger winter of 1944 have shown that the famine had long-term impacts on their health (Heijmans et al. 2008). Similar reasoning linking prenatal nutrition and adult chronic disease has been in circulation since the 1980s, under the name of the “Barker hypothesis.” The theory provides one possible explanation for the poorer health of poor people in both Western and developing countries. Barker and his colleagues showed a correlation between lower weight at birth and during childhood and greater risk for coronary heart disease later in life. This implies that normal variations in the transfer of food from mothers to babies have profound long-term implications for the health of the next generation. Later studies showed that low birth weight is associated with an increased risk of hypertension, stroke, and type 2 diabetes. This led to the “Fetal Origins Hypothesis,” which suggests that coronary heart disease and the diseases related to it originate with responses to undernutrition during fetal life and infancy. These responses permanently change the body’s structure, physiology, and metabolism. Because low SES is associated with poorer access to nutritious food, this theory would provide an explanation as to why generational patterns of disease match patterns of SES.

5.2. Biology

The experimental settings that test the impact of prenatal nutrition treat food as an exposure: According to the type and quantity of food eaten by pregnant mice, the constitution of the pups’ organisms is explored at the molecular level and prenatal nutrition outcomes are tracked on the pups’ bodies. Nutritional epigenetics shows that nutrients influence epigenetic phenomena either by directly inhibiting enzymes that catalyze DNA methylation or histone modifications, or by altering the availability of substrates necessary for those enzymatic reactions. All these reactions may modify the expression of genes critical to certain physiological and pathological processes, including embryonic development, aging, and carcinogenesis.

Experimental work in nutritional epigenetics is mainly conducted with inbred mouse populations. The ability to use diet to manipulate gene expression, and thus phenotype, is the most notable feature of this mouse model system. Comparative studies are made: The offspring of pregnant mice fed with methyl donors and vitamin cofactors are compared to the offspring of pregnant mice subjected to normal or methyl-donor-deficient diets. Methylation levels of specific parts of the pups are then measured. The offspring of the pregnant mice fed with methyl donors show higher methylation levels than the others, even though their own diet is normal (Waterland and Jirtle 2003). Thus, the patterns of methylation set in utero or in infancy seem constant even if the diet is changed later in life.

5.3. Articulation

The articulation of epidemiology and biology on the subject suggests that, in humans, large-scale social changes can translate into population-wide, heritable physiological changes due to diet—and that these changes can be tracked and mapped. Therefore, this research implies the existence of a mechanism whereby one generation’s experiences affect the metabolic systems of their offspring.

The connection between the two sciences requires further examination, however. In the two epidemiological examples, two different “social” contexts—war and childhood poverty—were viewed through the same lens: the quality of prenatal nutrition. But, in both cases, the emphasis on prenatal nutrition as an accurate indicator of social reality is questionable:

• In the first case, researchers limit causality to prenatal food deprivation—when wartime obviously entails many other stress factors (fear, bombings, the absence of men, the prospect of defeat, etc.).
• Concerning childhood poverty, mediating factors are also nested within broader contexts that may impact on food availability and quality of nutrition, such as the differences between rural and urban poverty (Evans 2004).

Because of the statistical nature of epidemiology, accessible information is limited to the collective experience of populations. Individuals are treated as immersed in their historical conditions or as a group defined solely by SES. Yet in reality, social contexts are complex, composed of many specific features. In these studies, diet is framed as an immersive environment—homogeneous for an entire group or time.

The connection between epidemiology and biology is possible only because social information is reduced to a pure question of nutrition. Through food, “the social” becomes embedded, not only in the bodies of those who eat, but in their capacity for replicating their own conditions of production. This form of naturalization of “the social” is particularly sensitive as it “recasts social suffering as molecularly heritable” (Landecker 2011).

6.1. Epidemiology

Because health behaviors and access to health care have failed to explain the SES gradient in health (Adler et al. 1993), psychosocial stress has emerged as one of the most plausible pathways whereby lower socioeconomic position can get “under the skin” and damage an individual’s health. It is postulated that individuals of low socioeconomic position are more likely to experience both chronic and acute stressors in their lives, from their physical environment, financial/occupational environment, and their sociocultural environment.

6.2 Biology

Crucial to the theory that stress makes poorer people sick is epigenetic research linking the physiological stress response to impaired immune function and other health risk factors. These are precisely the results that animal studies are designed to support. But how can stress induced in a laboratory rat reflect the human experience of people who are struggling economically? Diverse stress-induction procedures can be evaluated with regard to their construct validity—that is, the model’s ability to re-create with mice the etiological processes that cause stress-related diseases in humans.

• Because behavioral tests like the forced swim and tail suspension tests subject rodents only to acute (short-term) stress, they fail to mimic the stochastic and chronic environmental exposures that are the lot of people with low SES.
• Subjecting rodents to a series of repeated physical stressors (e.g., restraint, foot shock, and cold) over a period of weeks or longer does not reflect the psychosocial aspects of stress experienced by an underprivileged human.
• Chronic social defeat stress involves subjecting rodents to repeated bouts of social subordination. This design is all the more relevant as it involves a test of “resilience” if a subset of mice, subjected to the same stress, fails to develop behavioral and metabolic disturbances.

6.3. Articulation

The connection between epidemiology and biology depends on the validity of animal modeling for bodily reactions to social stress. Interestingly, the model of “social defeat stress” is favored by biologists because of its “social nature,” although they estimate “the intensity of the stress (. . .) more severe than that seen in most humans” (Nestler and Hyman 2010). But even establishing this comparison may be questionable, as it assumes that humans and mice share a comparable experience of social exclusion—a particularly bold assumption, in the absence of access to the rodent’s perception of the situation. All that is accessible to the scientists is the animal behavior (Crist 1999). Why is this phenomenological barrier so important? Because, lacking any shared perceptual apparatus, common language, or similar behavior, we are devoid of a reference system for interpreting the experience of the mice and comparing it to ours. We have to state that the human and the animal experiences of the world are incommensurable. They cannot be compared.

The collective intentionality of animal groups in general also eludes our understanding. The experience of having low SES, for a human being, and how it is experienced as a social class, can hardly be compared to chronic social defeat experienced by mice.

Bracketing out subjectivity has significant consequences for the way scientists represent the animal world, but it also affects the way they represent the animal in relationship to the human.12 The experiment described above presents a very flat picture of human stress—a representation that contradicts the interest initiated by social epidemiology for the reflexive dimension of the social.

• Findings of animal studies cannot be extrapolated to epidemiological studies that assume the reflexive dimension of social facts (based on categories such as “perceived social status” or “gender”).
• The “molecular biology of social position” assumes that socially differentiated collectives leave their specific epigenetic mark on bodies during windows of development, but this sense of belonging to a social group as tested in animals cannot be extrapolated to humans.

The incommensurability of “the social” experienced by animals and humans (Bimbenet 2011) fundamentally impedes the connection between social epidemiological research and animal studies.

7.1 Epidemiology

In social epidemiology, exposure to adverse childhood experiences (like trauma, abuse, or neglect) has been linked both “indirectly” (through tobacco and alcohol use) and “directly” to alterations in brain structure and in neurobiological stress-response systems, which in turn have consequences for health and emotional well-being (Reading 2006). The physiological embedding of stress induced by adversity in childhood would therefore have effects on the health of an individual over an entire life course (Delpierre and Kelly-Irving 2011; Kelly-Irving et al. 2013a, 2013b).

7.2. Biology

Rodent models have been used to demonstrate a link between adverse events early in life and health outcomes (Antoni et al. 2006; Lutgendorf et al. 2010). To provoke stress on pups, experimenters orchestrated regular phases of separation from the mother just after birth. The stress is proved to induce lifelong behavioral and neuroendocrine abnormalities in the pups.

In this specific type of experimental setting, the “natural” relationship between the mother mouse and her pups is impeded, and the separation procedure is supposed to simulate a “social” environment deprived of parental care and attachment. The experimental protocol includes refinements such as pups being confined in environments intended to minimize any additional stress during isolation (e.g., because of temperature change or a duration that would trouble their feeding). The point is to limit the effects evaluated to the lack of maternal care alone. But the most interesting experimental device concerns the selection of the mouse strain that is the most relevant to modeling parental care and the layers of interpretation it involves.

• First, except for a few strains that are not currently used in animal studies, mice are not monogamous and the care of the pups is assigned exclusively to mothers. Because the model of parental care is maternal behavior, an interspecies demonstration inevitably induces a focus on mothers. If epidemiologists intend to assess the role of both paternal and maternal behaviors (Sharp et al. 2018), they simply cannot apply animal studies.
• Second, among the different strains of mice available in laboratories, experimenters select the strain whose maternal behavior is genetically the most developed. This step requires explanation. The mouse has been an organism of choice for modeling human disease for ninety years. But, as Rheinberger contends, “model organisms are not to be found ready-made in nature; their conditioning takes effort” (Rheinberger 2010, 124). The production of laboratory mice entails breeding and transgenics (Haraway 1997; Rader 2004)—two practices that allow for the selection of very specific traits and for the generation of clone-like individuals. This genetic selection provides greater control of interindividual variability and results in highly specific breeding producing strains with physiological features or distinctive behaviors that may be more relevant to human modeling. Over 450 inbred strains of mice have thus been described, providing a wealth of different genotypes and phenotypes for animal studies (Beck et al. 2000). Maternal behavior is one of the phenotypic traits described and scored. Strain comparisons suggest that genetic factors influence maternal behavior, but studies show behavioral variability in the same strains of mice across laboratories (Crabbe et al. 1999; Wahlsten et al. 2003). Such findings, therefore, challenge the assumption that model organisms are determined solely by their genetics. This discovery should not surprise researchers carrying out epigenetic studies, because they aim to prove precisely that individuals differ biologically despite a similar gene sequence. They should therefore admit “the potential of model organisms to produce quality findings is contingent on and unfolds across their development, including within the experiment itself” (Friese 2013). But how can environment–gene interaction be explored if neither the biological nor the social can be strictly identified or derived from a single locus?

7.3. Articulation

To study mouse maternal behavior as a biological fact, an experimental device, and a model for human behavior, layers of interpretations and assumptions (somewhat contradictory to the purpose of the research) are necessary. The extent to which animal parental care can be compared to human behavior appears to be quite limited. The following mouse behaviors are used to score maternal care: nursing posture, pup licking, pup grooming/handling, nest building, and contact rest (Shoji and Kato 2006). How these observations could serve as a basis for human modeling is quite an issue.

With mice being used as models to project the effects of parental care on infant health, it is not surprising to see the term maternal warmth emerge. When social epidemiology, in association with animal studies, indicates that individuals lower in SES are at higher risk of poor mental and physical health, findings hint that children who grow up low in SES but who have “warm” maternal figures may be less likely to develop symptoms associated with a number of mental and physical illnesses later in life: “In the animal literature, high levels of maternal licking/grooming and arched-back nursing early in life buffer rats from elevated physiological responses to stressors later in life as adults. These findings suggest that early life maternal care may serve as one type of support factor that protects against detrimental biological responses in adult life. In the present study, we hypothesized that among adults who faced adverse circumstances early in life (low SES), those who experienced high levels of maternal warmth in childhood would show reduced indications of pro-inflammatory signaling compared to those who experienced low maternal warmth” (Chen et al. 2011, 2).

In ordinary language, “maternal warmth” refers not to temperature, but to particularly supportive and attentive people: It is therefore a metaphor for a social fact. But as researchers go back and forth from animals to humans in their demonstration, the metaphor used to describe caring human mothers echoes the tangible physical experience of mother–infant proximity shared by the majority of newborn mammals. The ambivalence of the term, which can be interpreted literally or metaphorically as a biological condition or as a social norm, shows how the use of animal models exerts an insidious influence on our conceptions of maternal care and may distort our perception of “the social.”

In all three cases, experimenters sought to simulate human social environments using laboratory animals, in order to collect biological information. The biological correspondence between human and animal bodies is understood through the agential properties of genes and proteins (Maleszka et al. 1998). It is because at the molecular level a gene has a certain function, being either animal or human, that the comparisons are possible. As Rose puts it, molecular biology enabled biological materials “to be regarded in many respects, as manipulable, and transferable elements or units, which can be delocalized—moved from space to space, from organism to organism” (Rose 2007, 15). But epigenetics is precisely the science that puts the context back into the picture. In epigenetics, “the biological” needs to be understood within context (typically intracellular, intercellular, or organismic) and even includes higher levels that integrate ecosystems and sociality (van Speybroeck 2000). Therefore, animal modeling cannot rely solely on comparison between single functional units. It must involve a correspondence between interactional dyads (composed of functional units in context)—hence the need to question exactly what is being modeled at the molecular, genetic, cellular, organism, or ecological scale. In this sense, the precise nature and extent of the comparison being drawn between animals and humans, in reference to the influence of biological and social contexts, has yet to be determined.

The theory of biological embedding states that there is a cellular experience of the social. Animal studies are therefore conducted to test how functional entities such as neurons, genes, and proteins react to social stimuli. Interestingly, “the person” is absent from this theory, although the person is (a) the product of “the social”; (b) the subject that experiences “the social”; and (c) the model of agency in society.

1. The person designates a human being with reference to his/her relationships (family, friends, profession, etc.) and to his/her behavioral patterns as conditioned by the culture.
2. The person is the experiencing subject. Experiences such as embodiment, perception, suffering, or well-being, that are at the core of epidemiological investigations, rely on phenomenological interpretations of what it means for a person to experience his/her world.
3. The person is the social agent: It is the person who is self-conscious and rational, and who can be called responsible for his/her action in society.

The person is a complex social construct that clearly differentiates human beings from animals or objects. Removing “the person” from the study of social-to-biological pathways may be convenient for animal modeling, but it creates two important problems.

• First, as we emphasized, the phenomenological incommensurability of “the social” experienced either by animals or by humans impedes any extrapolation from animal studies to the human realm. In this sense, we approve Nagel’s statement that “we will have as much difficulty understanding our own experience properly if we approach it from another point of view as we would if we tried to understand the experience of another species without taking up its point of view” (Nagel 1974).
• Second, that which may be grasped in terms of agential properties for biological functional units can hardly apply for social influence. The “cell-to-society” link may amount to neglecting the role of human agency and instantiating behaviors as biological functions.

Animal modeling generates a partial representation of human experience that may not reflect what scientists believe to be central to the human condition. Instead, it is based on what is considered scientifically “doable” at a certain time, in a certain scientific context (Nelson 2012). However, the approach has consequences on the production of knowledge and on knowledge itself.

7.4. Testing the Solidity of Social-to-Biological Causal Chains

Both epidemiology and environmental epigenetics are interested in “the social” as a cause for health and disease, but it is not clear how the two models of the same phenomenon may articulate. Is “the social” as a cause for health and disease to be apprehended as a biological mechanism, or as a social pattern? Are these models even compatible along the same causal chain?

8.1. Focus on the Environment: “the Social” as a Cause for the Biological

In environmental epigenetics, “the social” is causal in the sense that it is an immersing molecularized environment that reacts with molecularized bodies (Niewöhner 2011). In this model, causation is mechanistic (Birney et al. 2007). The four main types of epigenetic mechanisms are histone modifications, DNA methylation, small and non-coding RNAs, and chromatin architecture. They all refer to chemical reactions that occur within cells in response to environmental stimuli and cause certain base pairs of DNA or RNA to be “turned off” or “turned on” to regulate gene activity. If epigenetic modifications are mechanisms, they are reducible to mechanical principles, which means that (1) they are to be explained in terms of motion and collision of matter, and (2) they are determined/predictable.

According to these two criteria, “the social” can hardly be considered a mechanistic cause. First, the power of “the social” lies not in its material properties, but in a collective intentionality. Second, social patterns do not obey physical determinism; they are arrangements shaped by past events and structures, and shaping current practices and structures. Also, importantly, social patterns condition our very perceptions of them. Habitus is an example of the effort by social scientists to grasp socialized norms or tendencies that guide behavior and thinking. According to this concept, power is socially, culturally, and symbolically created, but also constantly re-legitimized through an interplay of agency and structure (Bourdieu 1984). The process that creates a habitus is not individual but social, because it leads to patterns that are enduring and transferrable from one context to another. A person is not an automaton, so a habitus contains some flexibility, but within a certain habitus, at any one time, the choices are not limitless. Habitus is one of the many concepts social science has constructed to capture the force of social patterns; it describes how human behavior is neither a result of free will nor determined by structures. Instead, it is created by an interplay between the two over time.

8.2. Focus on Behavior: “the Social” Caused by and Causing the Biological

In a different perspective from environmental epigenetics, behavioral epigenetics examines the role of epigenetics in shaping animal (including human) behavior. It is an experimental science that seeks to explain how nurture shapes nature, where nature refers to biological heredity and nurture refers to virtually everything that occurs during the life span (e.g., social experiences, diet and nutrition, and exposure to toxins). Behavioral epigenetics attempts to understand how gene expression is influenced by experiences and the environment to produce individual differences in behavior, cognition, personality, and mental health.

Behavior stands as “the social” in the transition from biological to sociological; it is an effect of “the biological.” But here again, there are two different layers of explanations for the question of how human behavior is regulated.

• If biological behaviors are grasped as byproducts of genetics, influenced by the endocrine and nervous systems, they may be explained in terms of reactions to stimuli.
• If social behaviors are conceived as dispositions, the focus of inquiry should be on an interpretation of social power, an exploration of social patterns, and biographical differences.

In both perspectives, human action follows regular patterns, making a behavior a relevant object for science. Here, the question is whether the regulation of social behavior by biological mechanisms makes sense of the social patterns. Considering the specificity of human agency, a molecular explanation is bound to fail to capture the relevant factors in the social situation. This argument is close to the anti-reductionists’ demonstration that micro-explanations can only explain the processes according to which macro-explanations are operating, but cannot serve as a substitute for them (Garfinkel 1981; Kitcher 1984).

8.3. Scientific Pluralism

The specificity of human agency, therefore, rules out considering the social and biological along the same causal chain. In our demonstration, “cell-to-society” approaches are prey to two forms of disunity: one methodological and one ontological.

• From a methodological point of view, depending of the field of science considered, differences in the methods and research strategies employed result in different practices. As we saw earlier, the aim of epidemiology would be to register and formulate controlled statistical regularities, whereas the goal of experimental science would be to determine measured outcomes.
• Ontological disunity is a more radical form of disunity, according to which there is no overall unity across the various aspects of reality. If experimental science and social science seem to be so different in what they posit, this is because the aspects of the world the relevant theories describe are fundamentally different (Cartwright 1999). The latter discussion relies on a metaphysical demonstration of the existence of different levels of causality in nature (Fodor 1974, 1997).

We would like to advance the discussion on methodological disunity by approaching biological mechanisms and social patterns as scientific constructs that inherit from different “styles of reasoning” (Hacking 1992). Styles of reasoning are patterns of inferential relations that are used to select, interpret, and support evidence for certain results. They point to the broad frameworks that govern a certain way of investigating the world. In this perspective, the deterministic approach of epigenetic mechanism may apply to explain behaviors because of a certain set of laboratory practices. In his discussion of what he calls the “laboratory style,” Hacking notes that it is characterized by “the building of apparatus in order to produce phenomena to which hypothetical modeling may be true or false, but using another layer of modeling, namely models of how the apparatus and instruments themselves work” (Hacking 1992). Crucial to the laboratory style are therefore the practices of building an apparatus, as well as hypothetical and instrumental (or animal) modeling. Hypothetical modeling is concerned with accommodating relevant phenomena, whereas instrumental modeling focuses on figuring out how to interpret the phenomena created by the apparatus. To put it differently, in laboratories, phenomena are produced, and they need to be explained by suitable models. In the examples we provided of epigenetic studies using animal modeling, the theory that “the social is a biological cause of health and disease” is tested via two types of modeling:

• as an effect of the environment, through simulation procedures (starvation, stress-induction, maternal separation)
• as an effect of the biological, through the production of animals whose traits have been genetically selected or artificially induced

The example of maternal behavior is particularly interesting, being situated as a biological effect and as a cause of the health of the pups. The strain of mother mice selected was known to genetically favor maternal behavior. Thus, although their behavior is viewed as a function of their genes, the health of the pups, on the other hand, is tested as a function of maternal care. A functional connection is a dependence between phenomena in which a change in one phenomenon is accompanied by a change in another. In this specific environmental setting, a change in the provision of maternal care is proved to alter the health of the pups. In both directions, maternal behavior is apprehended as a function only because laboratory practices allow for a thorough control of the mice production and the pups’ living conditions. To consider the pups’ health as a function of maternal behavior is a phenomenon created by this specific style of reasoning and may be interpreted in this context only.

Exporting the epigenetic style of reasoning to make sense of social patterns influencing health does not respect the style of reasoning specific to social epidemiology which, through the use of statistics, relies on a pattern of induction that allows for the calculation of probabilities, not for the testing of functions. This import would even prove to be harmful for social epidemiology and for public policies inspired by such frameworks if the “causal imperialism of the microphysical” (Dupré 1996) imposed biological views and models as to how social patterns work.

• Adler, N. E., W. T. Boyce, M. A. Chesney, S. Folkman, and S. L. Syme. 1993. “Socioeconomic Inequalities in Health: No Easy Solution.” JAMA 269 (24): 3140–45. https://doi.org/10.1001/jama.1993.03500240084031.
• Amsterdamska, O. 2005. “Demarcating Epidemiology.” Science, Technology & Human Values 30 (1): 17–51. https://doi.org/10.1177/0162243904270719.
• Antoni, M. H., S. K. Lutgendorf, S. W. Cole, et al. 2006. “The Influence of Bio-Behavioural Factors on Tumour Biology: Pathways and Mechanisms.” Nature Reviews Cancer 6 (3): 240–48. https://doi.org/10.1038/nrc1820.
• Aronowitz, R. 2008. “Framing Disease: An Underappreciated Mechanism for the Social Patterning of Health.” Social Science & Medicine 67 (1): 1–9.

• Bateson, P. 2001. “Fetal Experience and Good Adult Design.” International Journal of Epidemiology 30 (5): 928–34. https://doi.org/10.1093/ije/30.5.928.
• Bauer, S. 2011. “Appréhender Le Social Dans La Recherche Épidémiologique. Vers Une Politique Des Associations Statistiques?” In De La Vie Biologique à La Vie Sociale. Approches Sociologiques et Anthropologiques, edited by J. Kehr, J. Niewöhner, and J. Vailly, 298–327. La Découverte.
• Beck, J. A., S. Lloyd, M. Hafezparast, et al. 2000. “Genealogies of Mouse Inbred Strains.” Nature Genetics 24 (1): 23–25. https://doi.org/10.1038/71641.
• Bimbenet, E. 2011. L’animal Que Je Ne Suis Plus. Gallimard.
• Birney, E., J. A. Stamatoyannopoulos, A. Dutta, et al. 2007. “Identification and Analysis of Functional Elements in 1% of the Human Genome by the ENCODE Pilot Project.” Nature 447 (7146): 799–816. https://doi.org/10.1038/nature05874.
• Blane, D., M. Kelly-Irving, A. d’Errico, M. Bartley, and S. Montgomery. 2013. “Social-Biological Transitions: How Does the Social Become Biological?” Longitudinal and Life Course Studies 4 (2): 136–46.
• Bourdieu, P. 1984. La Distinction. Les Editions de Minuit.
• Broadbent, A. 2009. “Causation and Models of Disease in Epidemiology.” Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences 40 (4): 302–11.
• Broadbent, A. 2015. “Causation and Prediction in Epidemiology: A Guide to the ‘Methodological Revolution.’” Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences 54:72–80.
• Burrows, H., I. Hieger, and E. L. Kennaway. 1932. “The Experimental Production of Tumours of Connective Tissue.” American Journal of Cancer 16 (1): 56 LP–67. https://doi.org/10.1158/ajc.1932.56.

• Cartwright, N. 1999. The Dappled World: A Study of the Boundaries of Science. Cambridge University Press.
• Cauer, W. 1954. Theorie Der Linearen Wechselstromschaltungen. Vol. 1. Akademie-Verlag.
• Chen, E., G. E. Miller, M. S. Kobor, and S. W. Cole. 2011. “Maternal Warmth Buffers the Effects of Low Early-Life Socioeconomic Status on Pro-Inflammatory Signaling in Adulthood.” Molecular Psychiatry 16 (7): 729–37. https://doi.org/10.1038/mp.2010.53.
• CIFAR. 2014. “From Cell to Society: Creating a World Where All Children Can Succeed.” https://www.cifar.ca/events/from-cell-to-society-creating-a-world-where-all-children-can-succeed.
• Crabbe, J. C., D. Wahlsten, and B. C. Dudek. 1999. “Genetics of Mouse Behavior: Interactions with Laboratory Environment.” Science 284 (5420): 1670 LP–72. https://doi.org/10.1126/science.284.5420.1670.
• Crist, E. 1999. Images of Animals: Anthropomorphism and Animal Mind. Temple University Press.

• Delpierre, C., and M. Kelly-Irving. 2011. “To What Extent Are Biological Pathways Useful When Aiming to Reduce Social Inequalities in Cancer?” European Journal of Public Health 21 (4): 398–99. https://doi.org/10.1093/eurpub/ckr076.
• Desrosières, A. 1993. La Politique Des Grands Nombres. Histoire de La Raison Statistique. La Découverte. La Découverte/Poche.
• Dupré, J. 1996. “Metaphysical Disorder and Scientific Disunity.” In The Disunity of Science, edited by Peter Galison and D. Stump, 101–17. Stanford University.

• Evans, G. W. 2004. “The Environment of Childhood Poverty.” American Psychologist 59 (2): 77.

• Fassin, D. 2000. “Entre Politiques Du Vivant et Politiques de La Vie: Pour Une Anthropologie de La Santé.” Anthropologie et Sociétés 24 (1): 95–116.
• Fodor, J. 1974. “Special Sciences (or: the Disunity of Sciences as a Working Hypothesis).” Synthese 28: 77–115.
• Fodor, J. A. 1997. “Special Sciences: Still Autonomous After All These Years.” In Philosophical Perspectives. Vol. 11, Mind, Causation and World, edited by J. A. Tomberlin, 149-63. Wiley.
• Foucault, M. 1976. Histoire de La Sexualité T1 La Volonté de Savoir. Gallimard.
• Friese, C. 2013. “Realizing Potential in Translational Medicine.” Current Anthropology 54 (7): 129–38. https://doi.org/10.1086/670805.

• Garfinkel, A. 1981. Forms of Explanation. Yale University.
• Glynn, T. J., D. R. Shopland, M. Manley, et al. 1995. “Community Intervention Trial for Smoking Cessation (COMMIT): II. Changes in Adult Cigarette Smoking Prevalence.” American Journal of Public Health 85 (2): 193–200.

• Hacking, I. 1992. “‘Style’ for Historians and Philosophers.” Studies in History and Philosophy of Science Part A 23 (1): 1–20. https://doi.org/https://doi.org/10.1016/0039-3681(92)90024-Z.
• Hacking, I. 1995. “The Looping Effects of Human Kinds.” In Causal Cognition: A Multidisciplinary Debate, edited by Dan Sperber, David Premack, and Ann James Premack. Oxford University Press.
• Haraway, D. J. 1997. Modest_Witness@Second_Millennium. FemaleMan_Meets_OncoMouse: Feminism and Technoscience. Routledge.
• Heijmans, B. T., E. W. Tobi, A. D. Stein, et al. 2008. “Persistent Epigenetic Differences Associated with Prenatal Exposure to Famine in Humans.” Proceedings of the National Academy of Sciences USA 105 (44): 17046–49.
• Hertzman, C. 2012. “Putting the Concept of Biological Embedding in Historical Perspective.” Proceedings of the National Academy of Sciences USA 109 (Supplement 2): 17160 LP–67. https://doi.org/10.1073/pnas.1202203109.
• Hill, A. B. 1965. The Environment and Disease: Association or Causation? Sage Publications.

• Kelly-Irving, M., B. Lepage, D. Dedieu, et al. 2013a. “Childhood Adversity as a Risk for Cancer: Findings from the 1958 British Birth Cohort Study.” BMC Public Health 13 (1): 767. https://doi.org/10.1186/1471-2458-13-767.
• Kelly-Irving, M., L. Mabile, P. Grosclaude, T. Lang, and C. Delpierre. 2013b. “The Embodiment of Adverse Childhood Experiences and Cancer Development: Potential Biological Mechanisms and Pathways Across the Life Course.” International Journal of Public Health 58 (1): 3–11. https://doi.org/10.1007/s00038-012-0370-0.
• King, R. B., and R. M. Bures. 2017. “How the Social Environment Gets Under the Skin.” Population Research and Policy Review 36 (5): 631–37. https://doi.org/10.1007/s11113-017-9447-z.
• Kitcher, P. 1984. “1953 and All That. A Tale of Two Sciences Philip.” The Philosophical Review 93 (3): 335–73.
• Krieger, N. 2003. “Genders, Sexes, and Health: What Are the Connections—And Why Does It Matter?” International Journal of Epidemiology 32 (4): 652–57.
• Kuhn, T. S. 1962. The Structure of Scientific Revolutions. University of Chicago Press.

• Landecker, H. 2011. “Food as Exposure: Nutritional Epigenetics and the New Metabolism.” BioSocieties 6 (2): 167–94. https://doi.org/10.1057/biosoc.2011.1.
• Landecker, H., and A. Panofsky. 2013. “From Social Structure to Gene Regulation, and Back: A Critical Introduction to Environmental Epigenetics for Sociology.” Annual Review of Sociology. 39:333–57. https://doi.org/10.1146/annurev-soc-071312-145707.
• Landrigan, P. J. 1983. “Arsenic.” In Environmental and Occupational Medicine, edited by W. N. Rom, 473–79. Little, Brown and Co.
• Little, J., and J. H. Elwood. 1992. Epidemiology and Control of Neural Tube Defects, Vol. 20. Oxford University Press.
• Lutgendorf, S. K., A. K. Sood, and M. H. Antoni. 2010. “Host Factors and Cancer Progression: Biobehavioral Signaling Pathways and Interventions.” Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology 28 (26): 4094–99. https://doi.org/10.1200/JCO.2009.26.9357.

• MacMahon, B., and T. F. Pugh. 1970. Epidemiology: Principles and Methods. Little, Brown and Co.
• Maleszka, R., H. G. de Couet, and G. L. Gabor Miklos. 1998. “Data Transferability from Model Organisms to Human Beings: Insights from the Functional Genomics of the flightless region of Drosophila.” Proceedings of the National Academy of Sciences USA 95 (7): 3731 LP–36. https://doi.org/10.1073/pnas.95.7.3731.
• Marmot, M. G. 2003. “Understanding Social Inequalities in Health.” Perspectives in Biology and Medicine 46 (3): S9–23.
• Meloni, M. 2013. “Biology without Biologism: Social Theory in a Postgenomic Age.” Sociology 48 (4): 731–46.
• Morange, M. 2009. “How Phenotypic Plasticity Made Its Way into Molecular Biology.” Journal of Biosciences 34 (4): 495–501. https://doi.org/10.1007/s12038-009-0068-5.
• Murphy, M. 2004. “Uncertain Exposures and the Privilege of Imperception: Activist Scientists and Race at the US Environmental Protection Agency.” Osiris 19:266–82.
• Murphy, N. 1998. “Introduction.” In Whatever Happened to the Soul? Scientific and Theological Portraits of Human Nature. Oxford University Press.

• Nagel, T. 1974. “What Is It Like to Be a Bat?” The Philosophical Review 83 (4): 435–50. https://doi.org/10.2307/2183914.
• Nelson, N. C. 2012. “Modeling Mouse, Human, and Discipline: Epistemic Scaffolds in Animal Behavior Genetics.” Social Studies of Science 43 (1): 3–29. https://doi.org/10.1177/0306312712463815.
• Nestler, E. J., and S. E. Hyman. 2010. “Animal Models of Neuropsychiatric Disorders.” Nature Neuroscience 13 (10): 1161–69. https://doi.org/10.1038/nn.2647.
• Niewöhner, J. 2011. “Epigenetics: Embedded Bodies and the Molecularisation of Biography and Milieu.” BioSocieties 6 (3): 279–98.
• Niewöhner, J. 2015. “Epigenetics: Localizing Biology Through Co-Laboration.” New Genetics and Society 34 (2): 219–42. https://doi.org/10.1080/14636778.2015.1036154.

• Oyama, S. 2000. The Ontogeny of Information: Developmental Systems and Evolution. Duke University Press.

• Parascandola, M. 1998. “Epidemiology: Second-Rate Science?” Public Health Reports 113 (4): 312.
• Peto, R. 1984. “The Need for Ignorance in Cancer Research.” In The Encyclopaedia of Medical Ignorance, edited by R. Duncan and M. Weston-Smith, 129–33. Pergamon Press.
• Pickersgill, M., J. Niewöhner, R. Müller, P. Martin, and S. Cunningham-Burley. 2013. “Mapping the New Molecular Landscape: Social Dimensions of Epigenetics.” New Genetics and Society 32 (4): 429–47. https://doi.org/10.1080/14636778.2013.861739.

• Rabinow, P. 2001. “Artifices et Lumières: De La Sociobiologie à La Biosocialité.” In La Maladie Mentale En Mutation: Psychiatrie et Société, edited by Alain Ehrenberg and Anne M. Lovel, 295–305. Odile Jaco.
• Rader, K. 2004. Making Mice: Standardizing Animals for American Biomedical Research 1900–1955. Princeton University Press.
• Reading, R. 2006. “The Enduring Effects of Abuse and Related Adverse Experiences in Childhood: A Convergence of Evidence from Neurobiology and Epidemiology.” Child: Care, Health and Development 32 (2): 253–56. https://doi.org/10.1111/j.1365-2214.2006.00614_2.x.
• Relton, C. L., and G. Davey Smith. 2012. “Is Epidemiology Ready for Epigenetics?” International Journal of Epidemiology 41 (1): 5–9.
• Reubi, D. 2017. “A Genealogy of Epidemiological Reason: Saving Lives, Social Surveys and Global Population.” BioSocieties 13 (1): 81–102. https://doi.org/10.1057/s41292-017-0055-2.
• Rheinberger, H.-J. 2010. An Epistemology of the Concrete: Twentieth-Century Histories of Life. Duke University Press.
• Rose, N. 2007. The Politics of Life Itself: Biomedicine, Power and Subjectivity in the Twenty-First Century Princeton University Press. Princeton University Press.
• Ruphy, S. 2013. Pluralismes Scientifiques: Enjeux Épistémiques et Métaphysiques. Hermann.

• Schaffner, K. F. 2001. “Extrapolation from Animal Models.” In Theory and Method in the Neurosciences, edited by P. McLaughlin, P. Machamer, and R. Grush, 200–230. Pittsburgh University Press.
• Searle, J. R., and S. Willis. 1995. The Construction of Social Reality. Simon and Schuster.
• Sharp, G. C., D. A. Lawlor, and S. S. Richardson. 2018. “It’s the Mother! How Assumptions About the Causal Primacy of Maternal Effects Influence Research on the Developmental Origins of Health and Disease.” Social Science & Medicine 213:20–27. https://doi.org/https://doi.org/10.1016/j.socscimed.2018.07.035.
• Shoji, H., and K. Kato. 2006. “Maternal Behavior of Primiparous Females in Inbred Strains of Mice: A Detailed Descriptive Analysis.” Physiology & Behavior 89 (3): 320–28. https://doi.org/https://doi.org/10.1016/j.physbeh.2006.06.012.
• Skrabanek, P. 1994. “The Emptiness of the Black Box.” Epidemiology 5 (5): 553–55.
• Susser, M. 1973. “Causal Thinking in the Health Sciences Concepts and Strategies of Epidemiology.” In Causal Thinking in the Health Sciences Concepts and Strategies of Epidemiology. Oxford University Press.
• Susser, M., and E. Susser. 1996. “Choosing a Future for Epidemiology: II. From Black Box to Chinese Boxes and Eco-Epidemiology.” American Journal of Public Health 86 (5): 674–77.

• van Speybroeck, L. 2000. “The Organism: A Crucial Genomic Context in Molecular Epigenetics?” Theory in Biosciences 119 (3): 187–208. https://doi.org/https://doi.org/10.1078/1431-7613-00016.
• Vandenbroucke, J. P. 1988. “Is ‘the Causes of Cancer’ a Miasma Theory for the End of the Twentieth Century?” International Journal of Epidemiology 17 (4): 708–9.
• Vaughn, M. G., M. DeLisi, and H. C. Matto. 2013. Human Behavior: A Cell to Society Approach. John Wiley.
• Vineis, P., S. Stringhini, and M. Porta. 2014. “The Environmental Roots of Non-Communicable Diseases (NCDs) and the Epigenetic Impacts of Globalization.” Environmental Research 133:424–30.

• Wahlsten, D., P. Metten, T. J. Phillips, et al. 2003. “Different Data from Different Labs: Lessons from Studies of Gene–Environment Interaction.” Journal of Neurobiology 54 (1): 283–311. https://doi.org/10.1002/neu.10173.
• Waterland, R. A., and R. L. Jirtle. 2003. “Transposable Elements: Targets for Early Nutritional Effects on Epigenetic Gene Regulation.” Molecular and Cellular Biology 23 (15): 5293–300.
• Weed, D. L. 1998. “Beyond Black Box Epidemiology.” American Journal of Public Health 88 (1): 12–14.
• Wesson, D. W. 2013. “Sniffing Behavior Communicates Social Hierarchy.” Current Biology 23 (7): 575–80.
• West-Eberhard, M. J. 2003. Developmental Plasticity and Evolution. Oxford University Press.