New Study Helps Show That Genes Do Not Determine Personality

In my post here I document various false ideas taught by the  Smithsonian Institution and its site www.smithsonianmag.com.  Over the years the  Smithsonian Institution has played a large role in the propagation of false or unbelievable ideas taught by materialists. The latest false headline at that site is an article with this false clickbait headline:

Part of Your Personality Is Encoded in Your Genes. Huge Analysis Links Characteristics to More Than 1,200 Common Genetic Variants

It is not true at all true that part of your personality is encoded in your genes.  Genes only specify low-level chemical information such as which amino acids make up a particular protein molecule. 

You may start to realize the utter impossibility of a gene specifying a personality trait if you study the genetic code, and how simple it is. Below is a visual representing the genetic code. 

The letters such as A, C, T and G are types of nucleotide base pairs found in DNA and its genes. Using the scheme of representation depicted above, particular triple combinations of these base pairs can stand for or represent particular amino acids, the twenty types of chemicals shown in the outer right ring of the diagram. So, for example, the combination of CTC stands for the amino acid leucine. 

A gene is long sequence of such A,C, T and G "letters," typically many hundred or thousands, with each triplet of such "letters" standing for an amino acid in a sequence of amino acids used by a particular protein. There is no way a gene can represent anything mental or any aspect of a personality. There is no encoding scheme in DNA or its genes allowing such high-level representation, just as there is nothing in DNA or its genes allowing the specification of any complex structure bigger than a protein molecule. 

What is the study the Smithsonian article is referring to? It is the study "Robust inference and correlates from genetic associations with personality," which you can read here. Did that study find any decent evidence that personality is determined by genes? No, it found results entirely consistent with the idea that genes have no major influence on personality. 

The paper is a meta-analysis that crunches data from studies that gathered genomic data on lots of people who had taken personality tests. Such tests might give someone a scoring of how he rates on what is called the Big Five personality traits. Those who use this dubious term describe those traits as "openness to experience, conscientiousness, extraversion, aggreableness and neuroticism."  The paper attempted to find some examples of what are called SNPs that are correlated with particular personality traits. An SNP or single-nucleotide polymorphism is a tiny variation in a gene. The Google Gemini infographic below explains what an SNP is. The depicted portion of a gene is only a tiny fraction of a gene, which is many times longer. 

The paper says, "On average, the estimated effective number of independently associated common SNPs for each trait was 16,180 (Supplementary Table 17)." That number may sound impressive, but it actually is not any good reason for suspecting that genes determine  personality. The reason why is that the total number of SNPs in the human genome has been estimated as 600 million, and purely because of chance we would expect that an analysis like this would have produced as many correlations as the authors found, even if genes and their SNPs have no effect at all on personality. 

There are five traits in the Big Five list of traits. If on average each trait correlated with 16,180 SNP's, that would mean that roughly 100,000 SNPs out of 600 million would have some correlation with one of the Big Five traits (about 1 in 6000). Is that a number higher than we would expect to occur if genes and SNPs have no causal relation to personality traits? No, it is not. 

The correlations reported are weak correlations. When one thing exactly varies whenever some other thing varies, with the strength of the variation in the first thing always matching the strength of the variation in the second thing, then scientists say that is a correlation coefficient of 1.0. When there is a much weaker numerical association, scientists assign a much smaller number.  So, for example, the correlation between between height and weight is only about .3 and .4. Greater height tends to suggest greater weight, but the fact that you are 6 feet tall does not prove you weigh more than someone who is 5% shorter. 

The correlations reported in the new study are very weak correlations  We read this:

"Individual SNP effects were extremely small, reflecting the highly distributed genetic architecture of personality. For example, the median association estimate among extraversion lead SNPs, corrected for the winner’s curse, is 0.009 s.d. [standard deviation] per effect allele (corresponding to scoring at the 50.35th versus the 50th percentile in extraversion)."

A statistical effect of  0.009 s.d. [standard deviation] is a negligible effect, basically nothing. 

But should we be impressed that some non-zero correlations were found? No, we should not. That's because when analyzing causally unrelated things, you will often find small correlation coefficients greater than .1, even when the first thing has no real influence or effect on the second thing. 

I created an Open Office spreadsheet to demonstrate this reality. I typed in the Rand() function into one of the cells, and then copied that cell to fill up the two columns shown below. The spreadsheet then had two columns of completely random numbers, as shown below.

At the bottom of the spreadsheet, I typed in a formula that calculates the correlation between these two columns of random numbers. The formula is below:

=CORREL(B28:B57;C28:C57)

Now, the way this spreadsheet works, each time I press the F9 key on my keyboard, all of the random numbers are randomly regenerated, which causes the correlation number at the bottom to change. Each times that happens we are left with two columns of freshly generated entirely random numbers, and there is absolutely zero causal relation between those two columns.  Here are different correlation coefficient numbers I got at the bottom, after pressing the F9 key 10 different times. Each time I did that, there was recalculated the correlation between two new columns of freshly regenerated random numbers. 

Try 1: .007

Try 2: .118

Try 3: .127

Try 4: -.26

Try 5: .127

Try 6: .171

Try 7:  -.160

Try 8: .450

Try 9: .364

Try 10:  .0007

What does this spreadsheet and my ten recalculations of it show? It shows that correlations can always be found between two sets of causally unrelated data. It is not at all true that when analyzing two sets of data that have no causal connection, that you will always find a correlation coefficient very close to 0.  It is, in fact, extremely common to find weak correlation coefficients (below .35) when comparing two data sets that have no cause and effect relation at all.

The new meta-analysis paper analyzing personality traits and genome data fails to clearly report any correlations between personality traits and genomes/SNPs higher than we would expect to get from chance if there was no causal connection between personality traits and genes. What's going on seems to be what I call correlation-fishing or noise mining. That's when you analyze two things that have no causal relation, and find tiny little correlations here and there, without finding any decent evidence of a causal relation.

The trick of reading the paper's figures may elude some readers, as key graphs are found in a kind of sidebar on the far right of the paper. If you look at the numbers reported in those graphs, you will fail to find any strong positive correlation being reported. We see lots of weak-looking numbers such as .15 or .25, with apparently no sign of good correlations. There are also some indications that the correlations reported are mostly negative correlations.

I can imagine some ways in which that might work. For example, some SNP resulting in a defective gene leading to a genetic disease or genetic abnormality might result in a sick or ugly subject who is weak in "extraversion," because extraversion (an outgoing personality) is more common in those who are healthy and good-looking.  Similarly, someone with a broken gene resulting in sickness or ugliness might tend to be less self-confident because of such a shortfall; and that might show up in a slightly lower "openness to experience"  tendency. But you are torturing language in a misleading way if you try to make something like that sound like genes determining personality. Much better to just say, "Body problems can have a negative effect on personality."