Saturday, 17 March 2012

Personality Without Genes?

According to a paper just published (but available online since 2010), we haven't found any genes for personality.

The study was a big meta-analysis of a total of 20,000 people of European descent. In a nutshell, they found no single nucleotide polymorphisms (SNPs) associated with any of the "Big 5" personality traits of Neuroticism, Extraversion, Openness to Experience, Agreeableness and Conscientiousness. There were a couple of very tenuous hits, but they didn't replicate.

Obviously, this is bad news for people interested in the genetics of personality. But I wonder if the implications are even wider -

We know that there are SNPs associated with physical traits like height, weight, hair colour, eye colour, and the risk of various diseases. If none of those SNPs are associated with personality, then none of those traits are causally associated with personality.

"Short man syndrome"? A myth. Rod Stewart was wrong about blondes. There's no such thing as a "fat personality". And so on. Maybe that's not surprising, but more generally, the implication would be that the genes we inherit have no direct or even indirect influence on our personality, which is a pretty radical conclusion when you think through it.

I'm making some assumptions here. Maybe some genes are correlated with personality, but the currently popular "Big 5" approach is just a poor way of measuring of personality. It could also be that there are so many interacting genetic and environmental effects on personality that any given effect is tiny by itself, and even bigger sample sizes, or multivariate data analysis, would be needed to detect such effects.

ResearchBlogging.orgde Moor, M., et al. (2010). Meta-analysis of genome-wide association studies for personality Molecular Psychiatry, 17 (3), 337-349 DOI: 10.1038/mp.2010.128


Ivana Fulli MD said...

Thanks NS for a very nice and useful post.Once again but very much so.

My ( middle-age bitter pro bono clinician psychiatrist) take on it is:

You cannot find what doesn't exists and psychologists and psychiatrists have been using such a poor and poorly validated way of deciding what is a personality - not to mention the DSms'committees electing personality disorders.

Sorry about my bitterness and poor spelling and grammar but I lack time and find you posts so fascinating that I cannot resist commenting.

Anonymous said...

Perhaps a better corelation could be found for choleric/sanguine/melancholic/phlegmatic ?


Disgruntled PhD said...

Yeah, I would tend to come down on the side of the BIg5 being a pretty poor way of measuring personality.

The model comes from loads and loads of exploratory factor analyses, but has not been confirmed by either item response theory or confirmatory factor analysis (in fact, one of the authors of the Big% theory, McCrae, suggested that this is because CFA doesn't work for personality).

In my mind, this is far more likely than there being no genetic variants associated with particular types of personality.

estnihil said...

Agree with the above commenters: personality is a fuzzy vague kind of thing to be talking about, looking for specific traits and mannerisms would be better. Also agree that the Big 5 is a terrible way of measuring personality - five traits? Five traits!? Nothing is that simply in the human body or brain, or at least, that's my guess at things.

Is there an evolutionary argument suggesting that personality is at least partly genetic? Would it be an evolutionary advantage to inherit some forms of personality, or would it instead simply be better for each person to adapt to enviornmental circumstances - EVEN IF THESE ENVIRONMENTAL CIRCUMSTANCES RESULT IN LESS ATTRACTIVE PERSONALITIES?

mihai martoiu ticu said...

Maybe personality does not exist and one should try to find the behavioral components underlying the observed personality. What do I mean? There are genes that result in different behaviors. One gene can lead one to steal or give some talents helping to become a CEO (what is a kind of criminal). But it would be stupid for scientists to look for criminality gene or a CEO-gene.

Bernard Carroll said...

The solution is clear, as the authors themselves stated: let's do an even bigger study... there's a pony in there somewhere! And when they 'discover' two or three SNPs, how much of the 5-factor variance will these account for?

It seems the name of the game is to keep the game going.

Ivana Fulli MD said...

Thanks mihai martoiu ticu;

It is refreshing to get to read an intelligent and articulate lawyer on NS blog 's comments;

In France only failure at university or low academic expectations put young people to read law-with some exceptions for sure like my cruel eldest son who could have done a PhD in physics -even at cambridge UK- and works on one in international law!

FunPsych said...

Or, maybe temperament and personality are partly due to epigenetic factors (i.e. not from different SNPs but differential expression of genes due to methylation).

After all, we know that personality disorders can result from environmental stressors, environmental stressors can lead to different DNA methylation patterns, and DNA methylation patterns can be inherited.

Thus, I think it's a bit simplistic to think that personality is not influenced by genes just because no SNP associations were found.

Nina Bella said...

Maybe the reason we can't find a gene is that personality is controlled by our nervous system, and there are surely genes that control how that works. Whether you are sensitive like your mother or logical like your father or nervous like your grandma or ditsy like your aunt may be a function of how your nervous system is wired, and that is definitely something that is genetic. Maybe we don't need a gene for personality when we already have one (or several) for neurology.

mihai martoiu ticu said...


I’m taking only a master in International Law, but I’m not a lawyer. I graduated in philosophy, being interested in the problem of free will. I also took a course in cognitive neuroscience for this purpose. That’s why I get excited when I read stuff about genes influencing behavior.

But if those guys published in Molecular Psychiatry, they should know that genes only do some physical stuff, like regulating the chemicals in the brain. A behavior is the result of many chemicals and therefore of many genes. It might be true that some genes might play a bigger role in some components of some behavior. But the higher we go in abstractions, the more difficult it is to find a gene that has the most important causal role.

In short, there might be some genes that make me seek adrenaline shots, and to satisfy my need I might watch action movies, but I would not search for an action-movie-gene.

It’s like a car. Good old Newtonian laws make every component work as it works. But it would be stupid to search for a natural law especially ‘designed’ for turning cars left.

Jesse Marczyk said...

The radical conclusion of "genetics don't matter" is clearly wrong: personality is heritable, whether or not they have any specific genes in mind that have large, measurable, and consistent effects.

If genetics didn't matter, we wouldn't expect MZ twins to be any different from DZ twins, siblings, or strangers in personality from each other.

That said, I'm sure there's always room for improvement in personality measures.

Ivana Fulli MD said...


NS is worth reading several times for you since you wrote:"The radical conclusion of "genetics don't matter" is clearly wrong: personality is heritable."

No offense intended to you.

I cite NS:

"Maybe some genes are correlated with personality, but the currently popular "Big 5" approach is just a poor way of measuring of personality. It could also be that there are so many interacting genetic and environmental effects on personality that any given effect is tiny by itself, and even bigger sample sizes, or multivariate data analysis, would be needed to detect such effects."

Ivana Fulli MD said...

Nina bella,

Genetic component in the sense "mihai martoiu tico" elaborates, reasonable hypothesis.

Still I learned in Quebec several decades ago that in a family where a father, a mother and children live together for some years the first born child tends to reproduce the social ways of being of the father and the second born child social behavior imitates more the mother's.

Just to give an example of the "nurturing side" of the problem- which, by the way, makes one want to see the birth rank of the samples be taken into account in that sort of study.if the use of the big silly 5 did not makes one little hopeful about their results.

Anonymous said...

Correct me if I'm wrong, but hasn't it been shown that inheritance of a defective form of mono amine oxidase in a family was correlated with aggressive behaviours? Can't find the link now, but think it came out 6 years ago

gimpy said...

Aside from the concerns about the personality classifications, if personality is determined by the small influence of many genes then studies such as this would miss them.

The strategy to find them would be to look for genes whose mutation has a huge deleterious effect on personality (somebody mentioned MOA) and infer from this that in a heathy individual the activity of this gene contributes to personality traits similar to those affected when it goes wrong. Classic reverse genetics in other words.

Or use studies where pharmacological compounds are used to alter behaviour, then determine from that which genes are up or downregulated to manifest these changes.

Epigenome said...

Here are some papers about the epigenetics of personality:

mihai martoiu ticu said...

If I may use a new analogy: the fact that we describe people in a certain way, does not have to correspond to something in the physical reality. We describe some people as being stupid, but I bet there is no gene for stupidity.

Tal said...

The issue here is clearly not one of personality measurement. You don't have to reify the five-factor model to accept that virtually every well-validated self-report measure of personality shows high temporal stability, internal consistency, external validity, etc. (see Brent Roberts' beautiful Power of Personality paper). It's true that there's nothing special about the Big Five (psychometric descriptions of personality are arbitrary to a large extent), but it's clearly not right to suggest that we're not capturing personality properly.

The key point to keep in mind is that these are the very same personality measures that show high heritability in behavioral genetic studies. So one can't argue that personality is heritable, but something's wrong with the Big Five. There can be absolutely no question at this point that additive genetic variance contributes directly to self-reported scores on the Big Five (as well as pretty much every other personality measure).

The more likely explanation is that, as you suggest, the bulk of the variance is explained by high-dimensional interactions between variants and/or environmental factors, and hence no variant contributes more than a tiny (and effectively undetectable) portion of the variance when you look at main effects. Alternatively, it's possible that what's driving personality variance is not common variants at all, but rare ones that have very low base rates in the population and hence are not detected by standard GWAS studies (if this is true, the coming era of genome-wide scans will tell us).

As to the idea that genes don't contribute even indirectly to personality, I don't think we can conclude anything like that. Variables like height, attractiveness, etc., even if (causally) correlated with personality, are likely to have very small effects. If you think the correlation between, say, height and extraversion is on the order of .1 - .2, and we can't identify any SNPs that explain more than a fraction of a percent of the variance in height, it would be exceedingly unlikely that we'd successfully pick on on an indirect contribution to personality even with 20,000 subjects. The unfortunate reality is that the world is complex and nothing has simple causes--certainly not personality.

CM said...

If you want to know about the relationship between height and personality, you should look at that relationship (many have). The genetics doesn't add much here.

So many people getting carried away with brains and genetics and so forth when a good old behavioural measure will do the trick nicely. Look for answers in the most direct way possible.

Callum J Hackett said...

I think it seems fairly obvious that some personality traits are genetically heritable. Although I can't think exactly how I'd do it, I think research like this would be more fruitful if the Big 5 attributes were viewed as complex cultural manifestations of a much larger number of simpler traits which could feasibly be coded for at the gene level.

The Neurocritic said...

There was a rather extensive discussion of this issue when the paper by Verweij et al. was published in 2010 - see Bad News for the Genetics of Personality and Heritability, personality, and genomics.

LG said...

Darn ! I can't blame the genes then?

Anonymous said...

Is this more blank slate nonsense? I can't wait for the cultural anthropologists to declare, "with the right early puppy-hood interventions we can make pit bull pups into poodles"

omg said...

Did they go on safari ?

The phrase Big Five game was coined by white hunters and refers to the five most difficult animals in Africa to hunt on foot.[1] The term is still used in most tourist and wildlife guides that discuss African wildlife safaris. The collection consists of the lion, African elephant, Cape buffalo, leopard, and rhinoceros.[2] The members of the Big Five were chosen for the difficulty in hunting them and the degree of danger involved, rather than their size.

Ivana Fulli MD said...

Thanks omg,

For a bit of cultural distance and german humor and your political experience. I really like your comments very much- although I sometimes disagree.


It would be great fun to listen to an oral debate between you and mihai martoiu ticu.

I will just give you a personal experience of the relativity and cultural construct of "personalities"

Homeopathic remedies are prescribed considering the homeopathic personalities of the clients for the chronic cases-

It has nothing to do with the big five and those "homeopathic remedies personalities" do work-

in the sense that when you know those homeopathic personality - or use a computer software- you find them in people and those people feel understood

-and the more so when they read books by homeopaths dealing on the different personalities in need of those remedies.

NB: I am not telling hmo remedies works and agree that they can just be a very sophisticated placebo. my point is that with training you learn to recognize the arsenicum album personality and to differenciate it from the argentum nitricum personality etc...

Neuroskeptic said...

Great comments, people.

Anonymous: "Is this more blank slate nonsense?" - I hope not - it was published in Molecular Psychiatry. I think it's a fair reflection of the state of GWAS studies though. Personally I don't bother reading psych GWAS studies any more. Rare variant studies seem to be where the action is.

Neuroskeptic said...

omg: Heh. Maybe that's more than a coincidence...

Lion - Extraversion (they're the King).

Leopard - Neuroticism (always running around at full pelt, tiring themselves out.)

Elephant - Conscientiousness (they never forget).

Buffalo - Agreeableness (to the tastebuds)

Rhino - Openness to Experience... maybe.

Anonymous said...

"We know that there are SNPs associated with physical traits like height, weight, hair colour, eye colour, and the risk of various diseases. If none of those SNPs are associated with personality, then none of those traits are causally associated with personality."

This is an invalid inference, is it not, since correlation is not transitive? If A is (causally) correlated with B, and B is (causally) correlated with C, that does not entail that A is causally correlated with C. Consequently you cannot infer from "C is not correlated with A" that there is no causal relationship between B and C. Isn't it possible that, while SNPs are not correlated with personality factors, phenotypes associated with SNPs may be correlated with personality factors by virtue of social learning, acculturation, etc? Few of the associations under discussion here are all or nothing -- to take one simplified example, someone may have the "gene for" being 2 meters tall, but because of growing up in poverty and being systematically malnurished, their adult phenotype is that of being 1.7 meters tall.

Am I just missing something here?

Anonymous said...

Correct me if I'm wrong here, but the inability to locate SNPs that match the Big Five shouldn't necessarily rule out either the validity of the Big Five model OR potential connections to genetics, should it? I mean, just because a SINGLE nucleotide difference (i.e. single gene) is not responsible for the expression of these traits, that doesn't mean that gene complexes cannot play a role (as they almost certainly do in many other psychological traits, like intelligence), does it? In fact, one wouldn't expect single genes to explain personality differences, given that these traits are generally considered to be expressed on a spectrum or continuium (i.e. you could be a strong introvert, strong extrovert, or many variants in between). Personality traits are not like attached/unattached earlobes, where you have them or you don't.

NimrodV said...

Im not so sure. Surely there are genes for intelligence. A lack of normal function in these genes would lead to stupidty.

David Duffy said...

anonymous enquired...

"This is an invalid inference, is it not, since correlation is not transitive?"

Given certain assumptions about mechanism, eg (a) the relationship between shorter stature (S) and a particular personality type (P) is that S causes P (b) genotypes always cause height rather than height causing genotypes (c) no confounders, eg short stature is associated with various developental disorders (d) some kind of linear model is appropriate - then, yes, path analysis or structural equation modelling or causal analysis makes these kinds of testable predictions. As someone mentioned earlier, the expected height SNP-personality correlations will be tiny ((biggest height SNP) 1-1.5% of variance * (height-personality r) say 0.1. A traditional bivariate twin or family analysis might be able to say something, given it "uses" whole genome resemblances.

"...inability to locate SNPs that match the Big Five shouldn't necessarily rule out either the validity of the Big Five model OR potential connections to genetics, should it?"

No, but it's a real bummer ;). Actually
I'm just reading some of the papers on oxytocin receptor variants, where quite large effect sizes are claimed. I would expect some of this stuff on sociality and trust to be tapped by the big 5.

Ivana Fulli MD said...


You wrote "Personally I don't bother reading psych GWAS studies any more. Rare variant studies seem to be where the action is."

Please elaborate since for diabetis type2, obesity and myocadial infarction GWAS studies have shown (since spring 2007 in the review "Science"and the results have been replicated) genetic variations strongly corelated with the presence of those ilnesses in area that were not previously suspected to be linked to theses deseases at all.

As humbling and embarrassing for the cardio-vascular genetic research teams as for the psyche genetic research teams when you find no serious contendant for " The Schizophrenia gene " and too many variations common to schizophrenia sufferers and affective disorders sufferers- not to mention the autisms and ADHD

See the Eur project cadioGenics or whatever for references.

ramesam said...

Maybe I am wrong; but does this suggest an honorable place for Free Will - "the genetic blue print cannot decide your personality as if etched on stone, you can mold it the way you want" sort of a stand!
And again people are talking about "man" being a Superorganism - the sum total effect of the biome rather than human genes only?

Ivana Fulli MD said...


The subject of this post suggests only that either the "Big Fives" are not a valid model of personalities of the human specie or that NS was right in his clear and concise statment:

I re- cite NS:

"Maybe some genes are correlated with personality, but the currently popular "Big 5" approach is just a poor way of measuring of personality. It could also be that there are so many interacting genetic and environmental effects on personality that any given effect is tiny by itself, and even bigger sample sizes, or multivariate data analysis, would be needed to detect such effects."

NB: Tal and friends are clever enough to validate a 5 color model for karmas visual analogic scale- if they were to put their minds and brains into it that is.

and submit it to the right "Psycho whatever Scientific Journals".

A failure to find a genetic basis for their model wouldn't make you change religion would it?

Scud said...

Because the findings don't match your expectations, the whole concept of the Big 5 has to be wrong?

That's a ... new line of argumentation for you neuro.
And a bad one on top.

Ivana Fulli MD said...


I re-re cite NS:

I re- cite NS:

"Maybe some genes are correlated with personality, but the currently popular "Big 5" approach is just a poor way of measuring of personality. It could also be that there are so many interacting genetic and environmental effects on personality that any given effect is tiny by itself, and even bigger sample sizes, or multivariate data analysis, would be needed to detect such effects."

Neuroskeptic said...

Scud: I never said it had to be wrong. I'm just saying that if you accept that there are no Big 5 SNPs, then the explanation must be either:

1) Personality is independent of genetics
2) The Big 5 is a poor measure of personality
or 3) We just haven't found the personality SNPs yet because our sample sizes are too small or our stats aren't clever enough.

Or a mixture of those.

Neuroskeptic said...

Oops, meant to include under 3) there, "or personality is controlled by genetic variants but SNP GWAS haven't found them because they're not SNPs".

Andrew Oh-Willeke said...

The evidence that at least some personality traits are significantly hereditary and congenital is there. Genetic links to traits that are called mental disorders but look a lot like personality traits in character have been identified.

There are a lot of traits that are almost surely genetic for which we can't find genes. This says more to me about our deficient understanding of how the genome codes this kind of trait than it does about the absence of something somewhere in the genome that does.

It could very well be that simple SNPs don't code personality, or that we don't have good definitions of the kinds of traits that are actually coded. But, somehow or other, something is there to be found, even if we haven't found it in the crudest, simplest model of additive genetic variance available.

Andrew Oh-Willeke said...

Don't forget parasites.

the study by J. Xiao, L. Jones-Brando, C. C. Talbot, R. H. Yolken. "Differential Effects of Three Canonical Toxoplasma Strains on Gene Expression in Human Neuroepithelial Cells." Infection and Immunity, 2010; 79 (3): 1363 DOI: 10.1128/IAI.00947-10 suggests from a mouse model that there may be an important subclass of mental health conditions and/or personality traits with a parasitic infectious agent as a cause or trigger.

Ivana Fulli MD said...


And do not forget syphilis- as a curable by penicillin - cause of psychosis.

But what has this to do with the big 5 and genetic research?

Ivana Fulli MD said...


What should I do to get my first attempt to answer dearest Andrew deleted?

On 19 March 2012 17:59 some letters did not showed and I corrected it 19 March 2012 18:01.

Thanks again for a great post on a very important subject indeed: do we have to spend time and money for researches based upon "premature dogmas" as C Nemeroff put it (discussing hyppocampus and anti-depressant)or on "recognized truth " like the big 5 to the DSMs?

Anonymous said...

Any dog breeder can tell you of the many behavioral characteristics of the breeds. A working dog like an Australian cattle dog has a very different "drive" than a Labrador. I would bet that they have been well established in the literature as well.

Ivana Fulli MD said...

Anonymous 20 March 2012 23:30

There are so many naughty people outhere that you should be careful dealing about dog's since ignorant people can easily conclude that human races are equivalent to dogs' breed...

Plus, I saw a documentary film with a buddist "chief monks" having an adult lion as a pet of sort living peacefully with other animals inb a monastery and a pice of TV news an Italian man obliged by judges to send to Africa or a Zoo his pet lion who had to go to the African place where his beloved lion had ended to teach the keepers of his lion how to cook proper "pasta al dente" since his beloved lion was starving himself to death refusing to eat the overcooked pasta its African keeper had tried to give the "Italian pet adult lion" to eat. I am not making that up.

(In France a 24 years old French man of North African family origine is under "military elite forces siege" after the killing of three Jewish French children and 4 adulkts French adult men- one jewish, one black and two arabs.

A great scientist had to resign after having claim that the blacks were a race of inferior intelligence in a country with a half-black president of a much more than average intelligence.

Ivana Fulli MD said...

The 3 French Jewish children to be buried in Israel where killed at the entrance of a jewish school with a teacher and the three other adults French men killed with the same weapons were French army paratroopers in uniform.

The alledged killer is a 24 years old claiming that he wanted to be the avenger of Palestinian children victims.

The scientists have some morals responsabilities in society.

As much as some of Neuroskeptic learned commentator might hate it I think the XVI century François Rabelais (physician and writer) "Science sans conscience n'est que ruine de l'âme" ("Science without conscience is only the ruin of the human soul") still holds truth in it.

omg said...

Science has no soul. It tortures aka dissects for truth.

No comment about the Sarkozy debacle. It's an election year everywhere. Ivana, you should consider becoming the president. I reckon they'll have scientists instead of lawyers as presidents oneday.

oMg I had a bit of a belly laugh about NS' Big 5. Should've been the domesticated fowl, cow etc.

Ivana Fulli MD said...


Scientists must have conscience

be it only because a Northern Europe sociologists study I cannot put even an eye on is supposed to show -with good enough methodology for that kind of studies that is- that mental illnesses stigmatisation is much worst in people believing in the biological and heriditary origins of it.

*I agree NS has a nice sense of humor -anf gift with English making his post and comments interesting.

Thanks for the moral support.

Anonymous said...

Twin study?

Davidson Loomington said...

Hi N.S.-

I enjoy your commentaries!

I'm surprised at how willingly you find these results to be... interesting. It's one thing to assume genes have a role in determining our personality, and it's completely another thing to assume there is such a "thing" as "personality". The way we talk, the way we interact, the decisions we make and the behaviors we engage in... these are not evidence of some other-mechanism called "personality". What's our evidence that "personality" exists? The actions and behaviors we use to report on them.

The next question to ask, and I'm surprised you didn't ask it this time (maybe you have in the past), is how do we know these behaviors are evidence of this "thing" of "personality"? The answer personality psychologists fall back on is: our statistics. The answer, if you realize that statistics don't create evidence, they are tools to diagnose it, is: We know these behaviors are evidence of personality because people "of this personality" engage in these behaviors.

So it comes down to a logical error of reification. Personality exists because people who behave such a way have the personality that makes them behave such a way and we know they have that personality because they behave that way and on and on and ad nauseum.