“It’s Genetic”: How scientists determine whether traits are acquired or genetic (and why it can’t really tell you that much)

The nature vs nurture debate is alive as ever. But what does a scientist actually mean when they say that some trait is innate? And what does it mean to say that some trait is acquired through our contact with our environment? I’m gonna take you through the process by which geneticists figure out whether a trait is innate or acquired. I’m also going to show how this research is misinterpreted and then often weaponised by racist pundits. Yes, there are genetic factors at play which contribute to the way we are, but trying to explain broad and vague notions like ‘intelligence’ or ‘strength’ as a simple matter of genetics is ridiculous (and not supported by the literature).

So what do we actually mean when we ask whether some trait is ‘innate’ or ‘acquired’? Well in the first place we are assuming that environmental and genetic factors represent two distinct causal influences on an organism’s development, and one can scientifically ascertain how these causal factors contribute to the development of that organisms traits’. In this respect, the innate/acquired distinction represents an ontological commitment to the idea that ‘genes’ and ‘environment’ constitute two natural and mind-independent causal variables, in addition to the methodological claim that we can ascertain the influence of these variables on the traits (or phenotypes) an organism possesses.

The first issue we have here, in terms of making clean cut decisions about what is genetically inherited and what we get from our environment, is that there is a strong consensus that any given trait requires both environmental and genetic inputs to develop (Heyes 2018; Okasha 2009; Barker 2015, p.54; Lewontin 2006; Block 1995; Lewens 2018; Lynch 2016). So from the get go, when we talk about whether a trait is acquired or inherited, we aren’t talking about whether it’s one or the other, we are talking about the degree of influence each of these factors has on the development of the trait.

Thus, even though we can talk about ‘genes’ and ‘environment’ as two distinct causal influences, this distinction should not be formulated on the assumption that either a trait is caused by genes or it is caused by environmental inputs. So if, for example, a scientist claimed that the preference for chocolate over vanilla is innate, what they are really saying is that genes have a greater effect on liking chocolate than our environment does — yet both still have a role to play.

Heritability Analysis

One of the most common tools scientists use to study the different degrees of influence that genes and environment contribute to the development of a trait is a heritability analysis.

First I will explain how heritability analyses work and what they do, and do not, tell us about genetic and environmental influences on traits. Then I will explain how it lends support to the innate/acquired distinction.

Ned Block (1995, p.101) writes,

Heritability is defined as a fraction: the ratio of genetically caused variation [with respect to some trait] to total variation (environmental and genetic)… If the genetically caused variation is small compared to the environmentally caused variation, then the heritability is low.

Heritability analyses do not directly demonstrate whether some trait is, or isn’t, genetically determined. Rather, they demonstrate how much phenotypic variation within a population is due to genetic variation within that population (Lynch, 2017, p.26). The key point is that heritability analyses show the causes of variation with respect to a trait, not the cause of the trait as such (Lewontin 2006, p.402). That’s all very wordy, but let me explain with an example. Suppose you a took a sample of people and measured their height. What a heritability analysis estimates is how much of the variation in height within this population is caused by genetic variation within that population, and how much is caused by environmental differences. Thus, if I produced the claim, ‘height is 80% heritable’, I am claiming that 80% of the variation we find in some aggregated population is a function of genetic differences within that population.

Heritability analyses do not tell you anything about the causes of a trait in any particular individual, nor do they give a causal explanation of the the differences between any two particular individuals. “Heritability is an irreducibly population‐level parameter” (Okasha, 2009, p.722); it describes a correlation between certain genes and certain traits within a statistically aggregated population. For example, to say that height is 80% heritable does not mean that an individual’s height is 80% caused by their genes, nor that height-as-such is 80% caused by genes; it means that given the total amount of variation we find in a specific population, 80% of this variation is potentially caused by genetic factors. Thus, “heritability analysis, in so far as it licenses any causal inferences, pertains only to population‐level causality” (Ibid). However, this is not to say heritability analyses, in of themselves, demonstrate population-level causality beyond dispute. Like any statistical methodology, they it demonstrates correlations between variables. Of course, one could use heritability analyses, and the correlation statistics they produce, as evidence for some causal claim about genes which is informed by a good scientific theory. Indeed, most scholars agree on the theoretical point that genes have a determinative influence of physical, mental and behavioural traits (Heyes 2018, p.69; Brown and Richerson 2013; Pinker 2004; Hunter 2005; Barker 2015). Hence, as a hypothesis, we would expect to see a correlation between genetic and phenotypic variance within populations when we control for environmental differences — this is precisely what heritability scores indicate. Thus, the innate/acquired distinction is supported by heritability scores since they allow us to separate out, and study the differential magnitude of influence, of environmental and genetic influences on population-level trait development.

Now I am going to present some problems with using this methodology to determine whether something is innate or required. I’ll show you why, and how, you should proceed with caution when dealing with sensationalist claims about something being genetically determined.

The Causal Indeterminacy of Heritability Analyses

Remember that for claims about genetic determination to be valid, we must be able to scientifically establish the the extent to which environmental and genetic factors causally contribute to the development of a trait. One might object that to claim that a trait is ‘heritable’ or ‘not heritable’, says nothing about the extent to which a trait is genetically or environmentally caused. To infer causation from heritability is to falsely equate the causes of variation in a population with the causes of phenotypic development within an individual. To see why this is the case, consider the following example. Consider the trait, ‘having two ears’. Most would agree that having two ears is part of our biological endowment. The majority of people who have less than two ears would have ‘acquired’ through some unfortunate accident. Thus, if we conducted a heritability analysis we would find that almost all of the variation in a population with respect to the variable, ‘having two ears’, would be due to environmental factors. Hence, ‘having two ears’ would have a heritability score of almost zero since almost all variation with respect to this trait would be environmental. However, it would be absurd to conclude from this that having two ears is an acquired trait; that it’s something we simply learn to do. Thus, the causes of variation within a population are not necessarily the causes of a trait. We should not infer from a traits heritability score that this predominantly innate or acquired (Garson 2015, p.85; Block 1995, p.102). The causes of variaton in a trait (losing an ear due to a post-natal accident) are not the causes of a trait (robust genetic codes which give almost all mamals two ears).

Hence we get our first qualification: Any claim about causation derived from heritability analyses must be bolstered by an account of the “proximate mechanism” (Brown and Richerson 2013, p.112) by which a gene gives rise to a trait. That is to say, a high heritability score indicates that genes might be significantly influencing a trait, yet we also need a plausible account of how genes influence said trait; akind of nuts and bolts explanation of how a gene, or cluster of genes, goes on to produce a certin trait. For low heritability scores we need a similar account of the proximate environmental causal mechanism. Hence, in the case where we found that ‘number of ears’ is not very heritable, if we wanted to then assert that the number of ears an individual possess is caused predominantly by environmental factors we would need to provide a proximate account of how environmental factors have a determinative influence on ear development, thereby making it a predominantly acquired trait. But of course, we know that humans’ basic physiological structure is highly robust across environmental differences, and that our genes code for things like ‘having two ears’. The stats give you a general idea that something might be genetetic, but for us to accept something as scientific we need to be able to empirically observe the actually mechanism by which a gene translates into a specific trait.

For example, heritability analyses suggest that height is highly heritable (Jelenkovic et al., 2016). Yet this is further supported by a complex proximate account of how genes can differentially influence the production growth hormones and rates of metabolism, resulting in height differences. Thus, the heritability analysis, corroborated by a strong proximate explanation, allows us to claim that height differences between individuals are, to a certain extent, innate.

There are other traits, like IQ for example, which some suggests are inherited (Herstein and Murray, 1996). However, there doesn’t seem to be any identifiable genetic, or polygenic mechanism, which codes specifically for performing an IQ test, and there is much disagreement over how (and to what extent) genes ffect cognitive capacities in general (Heyes 2018; Pinker 2004; Lynch 2016, p..44; Barker 2015, p.4; Pigliucci 2001). Moreover, “prejudicial treatment based on genetically expressed variation in physiological phenotypes can contribute towards the apparent heritability of other phenotypes” (Lynch 2016, p.30; see also Sober 2001, p.74). This perhaps explains the apparent heritability of traits like IQ, since racial and sex-specific traits (and their underlying genetic basis) are in many cases linked with low educational opportunities — which would explain the apparent genetic basis for variation in IQ. (There is also a broader discussion to be had about how IQ tests only measure those cognitive capacities which are traditionally valued by WASP-Y middle class men).

As such, in lieu of a there being robust account of the proximate mechanism by which IQ could be inherited, and due to the fact that there are other competing and more parsimonious explanations, we should be hesitant to derive any inferences to the effect that IQ is innate, from the mere fact that variance with respect to IQ seems be partially correlated with genetic variance. This is not to say that heritability analyses cannot be used to account for cognitive differences, however, the absence of clear proximate explanations should make us reticent to assert causal arguments with any degree of certainty.

The problem with population samples

Another feature of heritability analyses that potentially makes it a poor methodological support for making claims about genetic determinism is the fact that one cannot talk about heritability in absolute terms; the “heritability of a trait is always relative to a specific population of choice” (Mameli and Bateson 2011, p.439). Since the analysis demonstrates variance within a population, different sample populations can yield very different heritability scores. For example, suppose you want to examine the heritability of ‘height’. You conduct a heritability analysis on persons from a highly in-egalitarian society population in which certain individuals have access to large quantities of high-protein food, while others suffer from malnutrition. And then you conduct an analysis on population in which peoples’ nutritional environment is relatively standardised. In the second population, you would find that height is more heritable, while in the latter you find find that height is significantly less heritable since weight variation would be largely attributable to differences in individual’s access to nutritional food; the principle determinant of variation would be differences in individual’s class. Hence, it makes no sense to make claims like, ‘80% of height difference is caused by genes tout court’, since heritability varies from population-to-population. This perhaps undermines the ontological legitimacy of the innate/acquired distinction. Perhaps heritability scores are more of a reflection of our methodological decisions — like our sampling parameters and the theoretical models — than any natural phenomena occurring in the world.

But this issue can be surmounted by conducting analyses on multiple populations from different parts of the world. This, however, brings forth two qualifications: First, no single heritability analyses can be used to establish that trait is predominantly inherited or not. Before we proceed to any causal inferences about the relative influence of genes and environment on a traits development, we first need to establish that heritability is robust across diverse populations. Second, we cannot give an exact quantitative estimate of how much genes and environment contribute to trait development. Although we find that height is heritable across the board, its heritability varies within certain limits. For example, one study suggests that it is 63.5% heritable in south-east Asian populations, while it is 79% heritable in North America (Jelenkovic et al., 2016). Moreover, height tends to be more heritable in boys than girls. Hence, we can see that differences in height, generally speaking, are due more ‘innate’ differences between individuals than acquired differences — yet we cannot give a univocal quantitative estimate of ‘how’ innate it is. And to reiterate, this innate/acquired distinction is only applicable to the description of population-level phenomena.

Gene-Environment Interactionism

Lynch (2016, p.29) points-out that genes and environment do not work additively to produce traits, but interactively. Heritability analyses erroneously present genes and environment as two discreet causal factors, and imply the idea that a trait’s development can be understood as the additive effect of x amount genetic influence in addition to y amount of environmental influence. However, as many scholars rightly point out (Hunter 2005; Okasha 2009, p.723; Garson 2015, p.82; Hernandez and Blazer 2006, p.10: Lewens 2018, p.8–9) traits arise out of the complex interaction between these factors — not from the additive combination of discreet environmental and genetic inputs. One implication of this is that any single gene, or complex of genes, can produce radically different phenotypes depending on their environmental context (Lerner 1978). Moreover, a gene which in one environmental context makes the development of trait highly likely (making this trait highly heritable within populations in which this environmental context is ubiquitous) can make the development of this trait less likely in another. The development of antisocial personality disorder is an example of this phenomena (Caspi et al. 2002). Therefore, heritability analyses do not allow us to scientifically study the causal mechanisms by which genes and environment contribute to the development of a trait since they completely misrepresent the nature of this causal mechanism by falsely presupposing that genes and environment are additively causal, when they are interactively causal.

The irony of this objection is that many of the studies which have demonstrated the fact that genes and environment are causally interactive have used heritability analyses (Hunter 2005). This has been established, for example, by comparing heritability scores for various diseases from different sample populations with different general environmental conditions (Hunter 2005; Ridge, Mukherjee, Crane and Kauwe, 2013; Liu, Li and Tollefsbol, 2008; Blauwendraat et al., 2019). This allows us to see that in certain environmental contexts these diseases are more heritable because certain genes and that certain genes when paired with certain contexts lead to disease, while in others environmental contexts these diseases are less linked to the presence of this gene. Hence the comparison of heritability analyses across different populations can provide us with information about how the interaction of certain genes with certain environmental contexts effect the development of traits.

One might object that this still undermines the cogency of the innate/acquired distinction since it demonstrates the fact that genes and environment cannot be treated as separate causal influences — but rather, they are inextricably inter-dependent factors which cannot be parsed into two discreet causal variables. We need to abandon the innate/acquired distinction and find a different conceptual vocabulary for describing the interpenetrating influence of genes and environment (Wimsatt 1986, p.185). However, even if heritability analyses do not show us the additive influence of genes and environment nor the discreet causal effects of environmental and genetic factors; but, rather, these factors’ interactive contribution to trait development — we can nonetheless assert that there are genetic (innate) and environmental (acquired) factors influencing the development of a trait. While a gene’s phenotypic effects are sensitive to environmental contexts, it still has phenotypic effects which can be scientifically studied. Even though genes and environment are causally interactive, we can still talk about them as two distinct factors. Therefore, we can still talk about innate and acquired influences on trait development. We simply have to make the qualification that genes do not have phenotypic effects tout court, nor do genes have a univocal causal influences on traits, rather the kind of causal effects a gene has is context sensitive. Whenever we talk about the phenotypic effects of genes, we must specify the “particular group and set of environments” (Barker, 2015, p.84) in which they have these particular effects.

Conclusion: A Qualified Innate/Acquired Distinction

We can now bring each of these qualification together into a nuanced definition of the innate/acquired distinction. The innate/acquired distinction suggests that genes and environment represent two distinct influences on a traits’ development. Heritability scores can be used to determined the extent to which a trait, or variance in populations with respect to some continuous trait (like height), are caused by our innate genetic endowment and acquired environmental context if:

1. Heritability scores are complimented by an account of the proximate mechanisms by which a trait arises.

2. We use heritability scores to make generalisations about statistical populations, not individuals.

3. We take, and compare, heritability scores from a diverse range of populations

4. We express the heritability of a trait, not as an exact quantitative estimate, but as something given within an estimated range.

5. We express the influence of genetic factors in terms of their effects within specific environmental contexts.

Furthermore, we should note that, even if these conditions are met, an innate/acquired distinction which is grounded in heritability analyses can only be applied to aggregated population-level phenomena. Hence provided we are sensitive to these qualifying conditions, the innate/acquired distinction can be used to do productive scientific work.

References

Barker, G., 2015. Beyond Biofatalism: Human Nature For An Evolving World. Oxford: Oxford University Press.

Blauwendraat, C., Heilbron, K., Vallerga, C., Bandres‐Ciga, S., von Coelln, R., Pihlstrøm, L., Simón‐Sánchez, J., Schulte, C., Sharma, M., Krohn, L., Siitonen, A., Iwaki, H., Leonard, H., Noyce, A., Tan, M., Gibbs, J., Hernandez, D., Scholz, S., Jankovic, J., Shulman, L., Lesage, S., Corvol, J., Brice, A., van Hilten, J., Marinus, J., Eerola‐Rautio, J., Tienari, P., Majamaa, K., Toft, M., Grosset, D., Gasser, T., Heutink, P., Shulman, J., Wood, N., Hardy, J., Morris, H., Hinds, D., Gratten, J., Visscher, P., Gan‐Or, Z., Nalls, M. and Singleton, A., 2019. Parkinson’s disease age at onset genome‐wide association study: Defining heritability, genetic loci, and α‐synuclein mechanisms. Movement Disorders, 34(6), pp.866–875.

Block, N., 1995. How heritability misleads about race. Cognition, 56(2), pp.99–128.

Brown, G. and Richerson, P., 2013. Applying evolutionary theory to human behaviour: past differences and current debates. Journal of Bioeconomics, 16(2), pp.105–128.

Caspi, A., 2002. Role of Genotype in the Cycle of Violence in Maltreated Children. Science, 297(5582), pp.851–854.

Garson, J., 2015. The Biological Mind. London: Routledge.

Hernandez, L. and Blazer, D., 2006. Genes, Behavior, And The Social Environment. Washington, DC: National Academies Press.

Herrnstein, R. and Murray, C., 1996. The Bell Curve. New York: Free Press.

Heyes, C., 2018. Cognitive Gadgets: The Cultural Evolution Of Thinking. Cambridge: Harvard University Press.

Hunter, D., 2005. Gene–environment interactions in human diseases. Nature Reviews Genetics, 6(4), pp.287–298.

Jelenkovic, A., Sund, R., Hur, Y., Yokoyama, Y., Hjelmborg, J., Möller, S., Honda, C., Magnusson, P., Pedersen, N., Ooki, S., Aaltonen, S., Stazi, M., Fagnani, C., D’Ippolito, C., Freitas, D., Maia, J., Ji, F., Ning, F., Pang, Z., Rebato, E., Busjahn, A., Kandler, C., Saudino, K., Jang, K., Cozen, W., Hwang, A., Mack, T., Gao, W., Yu, C., Li, L., Corley, R., Huibregtse, B., Derom, C., Vlietinck, R., Loos, R., Heikkilä, K., Wardle, J., Llewellyn, C., Fisher, A., McAdams, T., Eley, T., Gregory, A., He, M., Ding, X., Bjerregaard-Andersen, M., Beck-Nielsen, H., Sodemann, M., Tarnoki, A., Tarnoki, D., Knafo-Noam, A., Mankuta, D., Abramson, L., Burt, S., Klump, K., Silberg, J., Eaves, L., Maes, H., Krueger, R., McGue, M., Pahlen, S., Gatz, M., Butler, D., Bartels, M., van Beijsterveldt, T., Craig, J., Saffery, R., Dubois, L., Boivin, M., Brendgen, M., Dionne, G., Vitaro, F., Martin, N., Medland, S., Montgomery, G., Swan, G., Krasnow, R., Tynelius, P., Lichtenstein, P., Haworth, C., Plomin, R., Bayasgalan, G., Narandalai, D., Harden, K., Tucker-Drob, E., Spector, T., Mangino, M., Lachance, G., Baker, L., Tuvblad, C., Duncan, G., Buchwald, D., Willemsen, G., Skytthe, A., Kyvik, K., Christensen, K., Öncel, S., Aliev, F., Rasmussen, F., Goldberg, J., Sørensen, T., Boomsma, D., Kaprio, J. and Silventoinen, K., 2016. Genetic and environmental influences on height from infancy to early adulthood: An individual-based pooled analysis of 45 twin cohorts. Scientific Reports, 6(1).

Lerner, R., 1978. Nature, Nurture, and Dynamic Interactionism. Human Development, 21(1), pp.1–20.

Lewens, T., 2018. Introduction: The Faces of Human Nature. In: E. Hannon, ed., Why We Disagree about Human Nature. Oxford: Oxford University Press, pp.1–9.

Lewontin, R., 2006. The analysis of variance and the analysis of causes. International Journal of Epidemiology, 35(3), pp.520–525.

Liu, L., Li, Y. and Tollefsbol, T., 2008. Gene-Environment Interactions and Epigenetic Basis of Human Diseases. Current Issues in Molecular Biology, 10(2), pp.25–36.

Lynch, K., 2016. Heritability and causal reasoning. Biology & Philosophy, 32(1), pp.25–49.

Mameli, M. and Bateson, P., 2011. An evaluation of the concept of innateness. Philosophical Transactions of the Royal Society B: Biological Sciences, 366(1563), pp.436–443.

Okasha, S., 2009. Causation in Biology. Oxford Handbooks Online.

Pigliucci, M., 2001. Phenotypic Plasticity. Baltimore: Johns Hopkins University Press.

Pinker, S., 2004. Why nature & nurture won’t go away. Daedalus, 133(4), pp.5–17.

Ridge, P., Mukherjee, S., Crane, P. and Kauwe, J., 2013. Alzheimer’s Disease: Analyzing the Missing Heritability. PLoS ONE, 8(11), p.e79771.

Sober, E., 2001. Separating Nature and Nurture. In: D. Wasserman and R. Wachbroit, ed., Genetics and Criminal Behaviour. Cambridge: Cambridge University Press.

Wimsatt, W., 1986. Developmental Constraints, Generative Entrenchment, and the Innate-Acquired Distinction. Integrating Scientific Disciplines, pp.185–208.

[1]Although they disagree as to what this influence amounts to.

[2]This example is modification of one given by Garson (2015, p.85).

[3]I am not in a position to arbitrate this debate. However, it seems that for there to a be clear proximate relationship between genetics and IQ, we would have to accept the modular model of the mind which the exponents of this relationship endorse. Yet, there is a lot of recent scholarship which seems to demonstrate the implausibility of the massive modularity hypothesis (Heyes 2018, p.6; Barker 2015, p.22).