User Tools

Site Tools


Behavioral Genetics

There are two major kinds of research that go under this name:

  • Heritability is a statistical measure of the tendency of traits to “run in the family”, and does not require any understanding of the way that inherited genes actually cause this variation. This is an attempt to scientifically resolve the Nature Versus Nurture controversy. See below.
  • Genetic association studies search a subject's DNA for sequences that “go along” (are correlated with) a trait. One recent technique is the Genome-wide association study (GWAS), which tests the DNA of subjects for millions of genetic variants, searching for the variants correlated with a trait.

Heritability is a mature technique, with results going back to the 1920's. Confidence in heritability of intelligence took a big hit in the 1970's when it was suspected that one of the field's founders (Cyril Burt) had fabricated some research results. Since then, the availability of large databases of subjects and computer analytic techniques has allowed wide replication of findings of substantial heritability for traits such as height and IQ, and also a broadening of these heritability findings to an almost embarrassing breadth of behavior traits, such as likelyhood to divorce (40%).

Genetic association is a much newer type of research that has only really gotten going since 2000 and is still rapidly developing. As of 2013, techniques such as GWAS have generated disappointing results. It had been hoped that such genome-level tests would be able to pick out genes that would predict the variation in many traits already known to be heritable, especially disease susceptibility, as well as behavioral traits such as intelligence and personality. This effort has largely failed. For example height is known to be 80% heritable, but GWAS struggles to explain more than 10%. See Still Missing for a more technical summary of this “missing heritability” controversy.

Genetic Variation

How much of personality is caused by genes?

What we humans find interesting about personality is the differences in personality. The Nature Versus Nurture question is how much of the differences between us are “caused by genes”, or (more precisely) caused by genetic differences. While human DNA is 96% similar to chimpanzees and 99.7% similar to Neanderthals, modern humans are on average 99.9% similar to each other. All human genetic variation falls in the remaining 0.1% of genetic diversity. In genetics we sometimes say that siblings “share half of their genes”, but what that really means is that they share half of their diversity. So siblings are expected to be at least 99.95% similar. We say “at least” because parents do not choose each other at random from the entire world population, and tend to favor some degree of similarity (Assortative Mating). See Human genetic variation.


Heritability is a measure of how much of the variation in a trait between individuals is due to genetic differences between individuals. Heritability is a number between 0 (no genetic influence) and 1 (complete genetic determinism), and is also often written as a percentage. Consider the trait of personality extraversion (being outgoing and energetic). If extraversion is 54% heritable, then 54% of the variation in extraversion is due genetic differences, and the rest other causes.

Heritability indirectly measures the effect of genetics on behavioral traits such as intelligence and personality by comparing the trait similarity to the degree of relatedness of family members. A model of the expected genetic differences between siblings, parents and other relatives is compared to the measured trait differences.

The big problem with using relatedness to find heritability is that the people in a family have a lot in common besides their genes. They live in the same place, and are rich or poor together. Also, most behavior is learned, and what young children learn from their parents is clearly important. Identical twins have (nearly) identical genes, and are highly similar for many traits. If the effect of the twins' shared family environment could be removed, then all of the remaining similarity would be heritability. One way to do this is to compare identical twins that have been adopted into different families, but this is rare, so it is difficult to get a large enough sample for reliable statistics.

The classical twin study (see Twin Study) gets around this problem by comparing the similarity of identical twins to the similarity of fraternal twins. The assumption is that in a family with fraternal twins, the shared family environment will cause the same degree of additional similarity between between the fraternal twins as is (on average) caused by the different environment shared by a pair of identical twins in another family. Using the assumption that the genetic effect on the trait is proportional to the genetic similarity, it is possible to solve for the heritability. The remaining variation (not explained by genetic difference) can be further broken down into contributions from the shared family environment and other unknown non-genetic factors.

More recently, heritability and environment effects have been estimated by fitting computer models to the data. These models can make use of pedigree information from several generations, can estimate more parameters than just heritability, shared and non-shared environment, and can relax some of the assumptions made in the classical twin study, such as allowing non-additive effects of genes. See Reconsidering the Heritability of Intelligence in Adulthood for an interesting example of current research in this area.

Heritability of Behavior

For the vast majority of behavioral traits for which there is a reliable test, substantial heritability has been found. From the table in Genetic Influence on Human Psychological Traits:

Personality40% – 60%
IQ (age 5)22%
IQ (age 18)82%
Major depression37%
Alcoholism50% – 60%
Conservatism (over age 20)45% – 65%
Religiousness (adults)30% – 45%

The major findings are that for a wide range of traits examined:

  1. The inherited (presumably genetic) component of individual variation is 50% to 70%.
  2. The component of variation due to family environment (parenting, social status, etc.) is 0% to 20%.
  3. The rest is other unknown causes or measurement error. Though this is more or less by definition the environmental contribution, it may be the prenatal environment rather than the social environment.
  4. Heritability may be reduced in low social classes or in poorer countries. See below.

This is true for a wide range of traits. Intelligence (as defined by IQ) and personality (as defined by personality tests) have gotten the most attention because they most directly relate to the sociopolitical heart of the nature-nurture controversy, but basically the same results have been seen for many other traits such as religiosity, hours of TV watched, and even whether when you cross your arms you put the left or right on top.

This body of research seems quite solid, but was intellectually marginalized and demonized for decades due to its direct connection with the controversies over IQ testing. The researcher who dominated early work in this area was Cyril Burt, who was successfully discredited for research fraud shortly after his death in the 70's. Whatever the truth of this matter is (many of the accusations didn't hold up), it seems clear that his early results were real because they have been reproduced with close agreement.

Heritability and Environmental Deprivation

The bulk of heritability studies have been done in developed countries, where the average levels of nutrition and education are reasonably high, and where homes with abusive or neglectful parents are a minority. If you look specifically at those living their lives in the worst environments, then your findings of the contribution from genetics, family environment and non-shared effects may differ. In fact, this has been found to some degree, and is an ongoing research area (see Heritability of IQ). In particular, the heritable contribution tends to decrease, with increases in contributions from either the family or non-shared environments. The mixed results may reflect variations in the badness of environments between studies, and also the amount of variation of the badness of the bad environment within an individual study.

But why should a deprived environment reduce heritability? If you have a genetic advantage, then wouldn't that appear in any environment? Not necessarily (see The Best Kind Of Person), but a likely explanation is that there is a “good enough” environment which allows you to reach your genetic potential, and that sort of environment is already available to most people in developed countries. Variations in the environments seen by identical twins or between adoptive families mostly don't take the environment out of the “good enough” range, so heritability is high. But if (for example) food is scarce, then the difference between one child getting enough to eat and the other suffering periods of malnutrition can cause significant differences in outcomes such as IQ or even physical height.

There is not yet a consensus on how big the effect of deprived environments is on heritability, but this effect is a big reason for cautiousness in “between group” comparisons. Poor people have in common the fact that they are poor, so you have to consider this as a possible cause for lower IQ scores in this group. If there is any actual difference in the average environments between two groups, then this shared environment (and the shared cultural adaptions that go along with it) may be causing some of the difference in outcomes.

What Does it Mean?

These results were correctly seen by both sides as being incompatible with extreme pro-nurture positions popular in intellectual circles during the 60's and 70's — that all humans have equal potential, and that any difference is due to environment. Though some may still hold these views, the science is not on their side. However, there are subtleties even in this interpretation, see Nature Versus Nurture.

The other interesting conclusion is that the family environment doesn't seem to have much effect on these things, so perhaps it hardly matters what (if anything) parents do beyond providing food, clothing and shelter. Judith Rich Harris attracted considerable attention for this interpretation in The Nurture Assumption. We don't buy this extreme interpretation, but these results do call into question ideas such as “Buying the right baby toys will increase my kids IQ”.

What Do the Numbers Mean?

So it's been shown that some traits that can be measured numerically are more influenced by our biological inheritance than by any other cause, but how should we understand those numbers? A major degree of interpretive freedom comes from the meanings we assign to the things being measured. Perhaps IQ is not the same as “intelligence”, so it could still be that intelligence is strongly influenced by parenting. Clearly personality tests don't measure everything that we would colloquially call personality, so perhaps important aspects of personality are strongly influenced by parenting. We feel this verges on a semantic quibble, and is not very productive. For one thing, many of the heritable traits such as the self-reported importance of religion in one's life seem to have obvious meaning and importance. They don't depend on the semantically troublesome procedure of giving names to the anonymous results that pop out of factor analysis.

Semantic Circularity

One interesting point is that, by their very design, intelligence and personality tests measure something that is relatively stable over a person's life, so will end up measuring the completely stable genetic contribution. The designers set out to measure something that was stable because it is the common understanding of these words that they describe something stable. They chose questions that gave consistent results over a lifetime, and therefore measure something that is not much affected by social experience. So it is not at all surprising that these tests measure tendencies present at birth, and hardly surprising that they measure the genetic contribution so precisely. In other words, if we accept that these tests define intelligence and personality, then it is almost true by definition that intelligence and personality are highly inheritable.

We're willing to more-or-less accept this because it is in line with common-sense popular understandings of these terms. By definition, personality and intelligence are things that are innate (stable over a lifetime) and little influenced by social experience (I'm just a glass-half-empty kind of guy.) It was a plausible idea that personality and intelligence are largely determined by early social experience (in the family) that takes place before verbal tests can be given. If this were true, we would replace the common belief of stability with the more refined one of early plasticity followed by stability, but the evidence has gone the other way. Clever pre-verbal tests in infants (see Descarte's Baby) have found stability of traits such as anxiety from a very early age. The behavioral genetics results are another nail in the coffin. Parenting has little effect on these things.

Genetic or Random

Innate individual differences may random rather than genetic. There is not nearly enough information in the genome to encode the detailed structure of the brain. Rather than being a blueprint, the genome is more like a recipe for an organism, unfolding through a process of development. In heritability studies, the non-heritable component is broadly referred to as “environment”, but this can't be assumed to be caused entirely by visible, meaningful and potentially controllable factors such as nutrition, parenting, peers and education. An unknown (but surely substantial) part of individual behavior variation is basically random. These developmental variations are caused by miniscule meaningless and uncontrollable details of the brain's internal environment.

What Does Parenting Affect?

This does not mean that parenting has no effect, just that it affects other things. One thing that the family environment and parenting style clearly do affect is the kind of family environment the children will create and the parenting style that the children will use when they grow up. This has important implications for children's future life choices and happiness, and evolutionarily significant effects on the number of grandchildren.


All behavior must ultimately have a genetic explanation, just as all behavior must have an electrochemical explanation. This does not mean that we must abandon psychological and sociocultural explanations for behavior.


analysis/nature/behavioral_genetics.txt · Last modified: 2014/03/22 23:21 by ram