Wednesday, August 1, 2018

It's Already Eugenics.

Over the past several years, I have spent  time taking on studies related to the genetics of mental illness, IQ, personality traits, etc.  I have done so for a few reasons, but largely to counter what I think is, effectively, a reductionist mindset that pervades the scientific community and, by extension, society at large, leading to a kind of eugenic attitude towards society's problems.  I believe that this is quite harmful.  My general approach has been to attack the studies directly, which have historically been relatively easy to pick apart, even for someone who does not work directly in such research.   It was my view, and still is, that the "findings" in these studies are not what is generally being suggested (genes for the traits noted above). 
Up until recently, these studies repeatedly failed to replicate at all, and the mantra from scientists was that they just needed bigger databases.  I had hoped this would not be the case, but I can see now that as the databases are getting into the millions, it appears that they are starting to get some associations that are relatively consistent (at least within specific groups). (Update: Please see Addendum 2 below.  it is still unclear to me to what extent there is replication).
What these associations mean,  of course, is open for debate and speculation.  I am of the view that they don't really have anything to do with the actual traits and are probably some form of population stratification, but even if that is the case, it does allow for a bit of statistical predictive capability.  In my opinion, this  minor predictive success will soon become asymptotic, but that alone is enough to fuel years of these studies, striving for better and better predictability.   Nevertheless, this puts me in a rather difficult position.
There is simply no way that I am going to be able to pore over the barrage of complicated studies coming out and gain enough knowledge of the study to sow any doubt that could in any way compete with splashy editorials in the New York Times and stories in other mainstream media outlets.  Without my old standby of no replicated associations, I would need to delve deeper to realistically contest the findings, but when I look at This recent study, for example, with N = 1.1 million and mathematical calculations like this:





I realize that I am in over my head.  Do I think that this kind of math likely obscures, consciously or unconsciously, flaws in the studies in question?  Sure, in fact I assume as much.  But does this over-the-hill, retired psychiatrist have the ability to parse these equations and raise any reasonable questions as to the implied conclusions and present that in an understandable manner?  Not by a long-shot.  These studies have a mathematical quality to them that is more akin to science being conducted by statisticians with genetic loci evaluated in a manner similar to economic indicators.  This makes me very nervous.
The reason I find it so unsettling is that there is not any kind of theory of the human mind underlying any of this.  It seems like anything learned on the way to procuring a PhD, could be tossed in the trash bin as irrelevant.  Forget Freud or Jung, forget Plato, forget art or creativity (unless you can find a genetic association), forget just about everything and focus exclusively on the prevalence of a few molecules.  It's a bizarre kind of nihilism.  Even if you lend any validity to these studies, it strikes me as akin to trying to figure out how a car works, armed only with a partial schematic of the electrical wiring.  My impression is that few of the scientists working in this field have any interest in delving into psychodynamic issues, exploring their own psychological proclivities, or taking any kind of broader view about how the mind works.  In fact, I think many of them would laugh at this notion.  Just find your gene, find some way to link it to some part of the brain and suggest some vague way in which this might make sense.  The fantasy is that some working model will come about at some point in the future.
It's one thing to study genes that might be related to height or eye color, or even breast cancer, without a working understanding of how genes might affect those things, but it is unconscionable to do so when researching psychological traits, then providing conclusions in the studies that read like Rorschach tests from attendees of a "Man as Computer" seminar.   If you are going to do studies in this field, take some time to relate what you are doing to the field of psychology, have a working idea of the mind, read some philosophy, and open yourself to some self-examination.  Don't excuse yourself from this just because you can conduct a study and crank out mathematical equations to examine the data, without taking an interest in the field.  If that's your attitude, then you should be focusing your research elsewhere.  I realize that psychology, by it's nature, is not very conducive to being easily studied, but reducing it to this bean counting of genetic links, just because this is the cleanest way to get quantifiable results, is ruining the field.
Unfortunately, even if I'm correct in assuming that little of practical value will come of this, the way the results of these studies are presented in mainstream media outlets, creates a perception that we just keep finding genes for just about everything.  Just in the past few months, there have been studies purporting to find genetic associations for church attendance, risky behavior, loneliness, merit, being a gym rat, bar hopping, being "hangry," along with various intelligence and educational attainment studies.  Even the most diehard proponent of genetic causality has to have at least a moment of doubt when studies like this come out.  But the studies all use the same approach.  Can we veto the ones that sound absurd, without raising questions about the validity of the others?
There is also the issue of cross cultural differences in the results.  For example,  the polygenic score results in the above noted study don't seem to apply to African Americans, or presumably, other cultural groups.  Although the authors try to explain this away, some damage has already been done.  Anytime you have a study that appears to show a genetic difference between white people and black people, there is a ready-made audience receptive to this idea.  That is both in the general population and amongst a particular group of science writers and poseurs, who need no introduction and pounce on these kinds of findings regardless of their validity.  As I said above, I don't believe that there are any real causal genes for educational attainment, so what do we make of it if different polygenic scores are used for different cultural groups?  Well, I think we know what will be made of it.  So, the eugenic damage has, to an extent already been done, and will likely continue with new studies coming out in the future.
I've witnessed a similar trend in my field of psychiatry, as the strong emphasis on psychodynamic issues related to mental illness when I started my Residency, faded away, replaced by a pharmacology model, fueled by drug companies and managed health care systems.  Part of the cover for these changes was also to point at these genetic studies, implying that there were genetic causes for all these mental illnesses that could only be addressed chemically.  The fact that none of those studies ever stood up to any rigor has become immaterial, because psychiatry has been changed permanently.  Again, the damage was done.
I have reached the conclusion that there is not much that can be done about any of this until it runs it course. Presumably, Godel's Incompleteness theorem will take hold and let this construct of reductionist thinking collapse like a house of cards. That's my version of optimism.
Even if that were ever to happen, though, we have the problem of the consequences in the interim, which is that an acceptance of these studies, right or wrong, has consequences for society.  Our society has been inundated with genetic hyperbole over the past few decades, and it appears to be intensifying.  This creates a mindset which pervades decision-making for many policies and treatments.  How we treat ADHD, how we view the prison population, how we view race and socioeconomic status, to give a few examples, are all now viewed, at least in part, through this genetic paradigm, even though no real causal genes have been demonstrated for any of these things.  And to what end?
I hear some scientists try to assuage fears about the implications of this kind of research, and offer optimistic assessments of its potential value.  I'm not particularly impressed by this wishful thinking.  It takes little imagination to see where these attitudes will take us, because we have been seeing it for some time.  It will only serve to divide people further, at its most extreme resembling some sort of a caste division.
Let me close by addressing individuals involved in this area of genetic research.  Coincidentally, I was invited to a movie halfway through writing this, called, "Three Identical Strangers,"  about a set of triplets adopted separately at birth to different families, all unaware of the existence of the others.  Although an important theme of the movie is a nature vs. nurture debate, I doubt it would change anyone's opinion on either side of that argument.  More importantly in the context of what I am saying here (without spoiling the movie), is the sense of how easy it is to accept ethically questionable rationales, when you narrow your focus to scientific pursuits over the ethical questions they raise. Ethical issues aren't spelled out in advance of new science.  That comes after the fact.  You may think that you have sewn up any ethical dilemmas related to this research and that you are thus immune to any unintended consequences, but in the end, you will be judged exactly for those unintended consequences.  That may seem unfair,  but a lack of foresight won't be excused by saying you were just following the numbers.  As a scientist, it is also your job to anticipate the consequences of the studies attached to your name.

Addendum Related to a Twitter discussion on this topic: By my count (feel free to confirm), of the 74 SNP's that reached significance in the Okbay study for educational attainment, 10 met significance in this most recent study (which found 1,271 total SNP's meeting significance).  When this was expanded to the 162 SNP's with the inclusion of the "replication" dataset in Okbay, 30 of the SNP's were listed as significant.  Not sure why this wasn't quantified in the original study or why it fell on me to work that out (with the help of a friend), but there it is.

Addendum 2:  After what I think could be considered some stonewalling on the part of the authors, etc., related to confirming the crossed out material above, I will now say that it appears likely that the "replication"  that was referred to in the study, had included the data from the previous study (Okbay, et al.).  I base this on my communication with a corresponding author (also Okbay).  Dr. Okbay concurred with the others who stated that the reason for the discrepancy between the listed SNP's and the claimed p-value significance for the previous 162 SNP's, was that there were other SNP's within the locus that had higher p-values.  She also provided me with a list of those 162 SNP's and their p-values.  I have no further information and remain skeptical of those results, but, of note, the 30 of the 162 SNP's that were listed amongst the 1,271 in the current study, had p-values that matched what is shown on the supplementary table.  The p-values from this table, as I understand it, were calculated using the full 1.1 million dataset.  Therefore, approximately 400,000 would have been part of the previous study in which these SNP's were found to be significant.  That is obviously going to bolster any claimed "replication" significantly.  How significantly, you might ask?  Well, who knows?  There is really a simple way to find out, which, based on the lack of cooperation I've received, might have already been done and left out of the study.  I did try to follow up with Dr. Okbay about the issue of independence of the dataset, but did not receive a follow-up response.  I don't know whether Nature Genetics has much interest in my concern, but I have been in contact with them and will update this if I hear from them further or receive further correspondence from any of the authors of this study.

No comments:

Post a Comment