Nature, Nuture and Luck

An ingrained assumption of the scientific literature on heredity is that non-hereditary variation between people is a function of their differential environmental influences. Yet, the more I read, the less plausible this seemingly safe assumption appears.

For example, I've recently been reading Entwined Lives: Twins and What They Tell Us About Human Behavior, by Nancy L. Segal (1999), in which the author repeatly employs this conceit.

For example, regarding IQ the author notes (in accord with many other credible sources on the subject) that:

IQ heritability (the extent to which genetic differences among people explain IQ differences) is estimated to be 20% in infancy, 40% in childhood, 50% in adolescence, and 60% in adulthood. . . . Typical IQ correlations are .86 for identical twins reared together, .75 for identical twins reared apart, .60 for fraternal twins, .42 for parents and children, and .15 for cousins. IQ correlations for adopted relatives are generally lower then those for biological relatives. They reflect the effects of shared environmental influences on intellectual development.

I have seen other accounts that put peak heritability of IQ at closer to 50% than 60%.

Yet, she also notes that:

One of the most compelling contributions from twin studies has been the knowledge that living together does not make people alike in intelligence. . . . Adoptive siblings show some IQ resemblence in childhood (.29-.34 correlation), suggesting a modest but real contribution from the shared environment. When adoptive siblings are studies in adolescence and adulthood, however, their IQ correlation reduces to zero (below the .15 cousin correlation), showing that shared environmental effects on intelligence diminish as they age.

One sees a similar pattern in the somewhat more nebulous area of a wide host of personality traits. "Approximately 20%-50% of individual differences in personality are genetically based." Of the "Big Five" personality traits (extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience), some have a higher heritability figure (49% for extraversion), some are lower (39% for agreeableness) and others come in between. As in the case of intelligence, environment seems to play a bigger role in personality in infancy and childhood, but "[p]ersonality studies concentrating on adolescence and adulthood report escalating genetic influence which eventually stablizes until late adulthood before diminishing somewhat, probably due to accumulated individual experiences."

Among other personality traits exhibiting genetic impacts are characteristic levels of happiness, shyness, "well being", stress reaction, aggression, chance of never marrying, religiousity, propensity to watch television, and "Traditionalism, the endorsement of traditional family and moral values." But, again, the studies show that "Shared environments make a small contribution to personality similarity among relatives." For example, variations in Traditionalism "did not show common family effects" which were not attributable completely to genetics . . . . "living together does not make people living in a family alike." Instead, these "similarities are explained primarily by shared genes."

Some studies have even found greater personality similarities in twins reared apart than in twins reared together, a "highly counterintuitive" finding perhaps explained by the theory that "twins living apart express themselves more freely because they are not influenced by their twin, while twins living together 'create' differences in the interest of individuation." She also notes that "Physical features, while similar in identical twins, are not linked to personality traits so even if similar responses are triggered by similar faces and figures they will have little effect on the developing personality."

She concludes this analysis with the common place conclusion that "If genes explain 20%-50% of the variation in personalitity and shared environment has a modest impact, then non-shared environments must account for the rest." She then goes on to discuss at length the notion that families have different ecological niches in them that influence personality and development.

The other possibility, however, is that there is an impact other than birth right and environment (shared or non-shared) on intellectual development. Perhaps, like iris patterns or finger prints, a significant share of how our genes express themselves is simply random and has neither a genetic cause, nor a meaningful environmental source.

Related to the notion of the impact of luck is the fact that almost no mental traits, with the exception of adult intelligence (and certain well known single gene mutations most of which are associated with various forms of mental retardation), break the 50% barrier for heritability. Given the very modest contributions of a shared environment to variation in these traits, perhaps the balance of the variation is genuinely random, perhaps even by means of some as yet not well understood process that causes intellectual development traits to manifest themselves only 50% of the time in people with identical genes.

A particularly striking example is schitzophrenia, which is known to have a strong birth right component, much of which is genetic in nature.

Approximately 1% of the general population is at lifetime risk for schitzophrenia, but if a person is affected, the risk rises to about 6% for his or her parents, 9% for his or her siblings, 13% for his or her children, 17% for his or her fraternal twin and 48% for his or her identical twin.

(A link between advanced paternal age and schitzophrenia suggest that age related mutation may explain much of the balance of the variation in the statistics shown above.)

Thus, only half of genetically, paternal age and pre-natal environment identical individuals express schitzophrenia, right at the upper limit observed for most mental traits with a genetic component.

The fact that the difference between identical twins who express schitzophrenia and those who do is not genetic in nature is further born out by another convincing study of "schizophrenia in children born to normal identical and fraternal twins and their schitzophenic twin siblings."

Remarkably, children of both well and ill identical twins showed the disorder with near equal frequency (17.4% and 16.8%), while children of well and ill fraternal twins differed greatly (2.1% and 17.4%). These findings confirm that identical twins inherit the same genetically influenced disease predisposition that can be passed to children even by a twin who does not express the disease, and that living with a schitzophrenia parent is not necessary for a disorder to appear in a child.

Some linkage in propensity to express schitzophrenia has been linked to the number of fingerprint ridges a child has, and in cases in which there are the biggest differences between twins, to birth weight and delivery traumas. Also even twins who are not described as actually expressing schitzophrenia are described as having "personality quirks and idiosyncrasies" to a greater degree than unrelated individuals, suggesting that expression of schitzophrenia genes is not an all or nothing affair, with the difference being the intensity of the expression rather than the existence of the trait at, at least, a subclinical level. Similarly, a famous study of the "Genain Quadruplets" who were raised together, all four of whom developed schitzophrenia, all expressed this disorder to different extents in different developmental patterns.

Affected twins had subtle developmental delays, disruptive tendencies and shyness relative to their unaffected twins in childhood, but overall there were "few differences in childhood predictors of schitzophrenia." And, of course, all of these childhood symptoms seem to point more towards the notion that expression of the disorder or lack thereof is a course set early in childhood, not to an environmental cause. (Notably, Alzheimer's disease appears to be another mental condition which is present at a subclinical level long before it can be diagnosed, often even decades earlier.)

It is fair to infer from the data, however, that a very large share of the likelihood that someone with a set of genes for schitzophrenia will expess that in a full blown case of schitzophrenia are just as random as the finger print patterns that have been linked to some extent to its expression in ways that have nothing to do with either genetics or any logically significant aspect of a shared or non-shared environment. Schitzophrenia expression may be just as random as the question of whether a particular sperm cell from a man with a given set of genes will fertilize an egg to become a boy or a girl.

Psychiatrists, geneticists, and psychologists would like to think that we live in a basically deterministic world where every effect has a determinable cause. But, the evidence seems to be mounting that expression of genetic predispositions towards mental traits may be as inherently stochastic in practice as the behavior of individual particles in the field of quantum mechanics.

It is an unsatisfactory conclusion that is hard to definitively prove because one never knows when one might find a missing link that clearly predicts who will and will not express a trait, or, at least, when the determination is made about who will and will not express it in the developmental process.

Most mental illnesses express themselves remarkably late in life for conditions with such strong genetic elements, and we also know that many of them routinely arise in the absence of any truly extreme conditions such as an intense psychological trauma (post-traumatic stress syndrome and possibly multiple personality disorder appear to be the most well known exceptions, with both having strong environmental causes).

Schitzophenia, for example, has among the strongest genetic elements of mental disorders (bipolar disorder may have an even stronger genetic element), but is typically a condition that manifests in late adolescence or young adulthood. We don't know, however, if one's fate to develop fully expressed schitzophrenia is determined in the womb, around the time of birth based upon birth traumas, in infancy, in childhood, or close to the time at which the disorder actually expresses itself.

Given the lack of any clear post-natal environmental factor in the epidemiology of one of the most carefully studied mental health conditions known (urban v. rural environment is perhaps the only environmental factor with well established statistical significance, and its impact is modest), it seems plausible that post-natal environmental factors aren't very important in determining if someone with a genetic predisposition does develop the condition. Likewise, much of the non-genetic variation observed in IQ and personality may simply be a matter of luck.

Not surprisingly, the National Institute of Mental Health is vigorously looking at the issue of why genetic predispositions to mental illnesses express themselves sometimes and not at other times.

UPDATE: The use of the word "luck" in the headline is, of course, something that no self-respecting behavioral scientist or evolutionary biologist would be in the title of the journal paper. The notion of a third factor distinct from genetics or environment needs a far more dignified name. I hereby dub this concept "random developmental variation" or "RDV" for short, until such time as I discover that there is another name for this concept in the literature that isn't totally lame.