Heights And Lengths And Areas Of Rectangles
The year 2013 has just finished and, as is the case at this time of year, media pundits suggest a variety of words and terms that should be banned; some of the most common ones have included, "YOLO," "bromance," "selfie," "mancave," and, of course, please God make it so, "twerking." In these cases, it's not because the terms are wrong, but just because they've become ubiquitous and irritating.
Similarly, some things in the science world beg to be retired. That's rarely the case simply because a term has been ubiquitous and irritating. "Genomic revolution" might be one of those. Another might be, "For 99% of hominid history…," when discussing what humans do in a less artificial setting than our modern world. Personally, I hope this phrase won't be retired, as I use it ubiquitously and irritatingly, with no plans to stop otherwise.
However, various science concepts should be retired because they are just plain wrong. An obvious example, more pseudo-science than science, is that evolution is "just" a theory. But what I am focusing on is a phrase that is right in the narrow sense, but carries very wrong connotations. This is the idea of "a gene-environment interaction."
The notion of the effects of a particular gene and of a particular environment interacting was a critical counter to the millennia-old dichotomy of nature versus nurture. Its utility in that realm most often took the form of, "It may not be all genetic—don't forget that there may be a gene-environment interaction," rather than, "It may not be all environmental—don't forget that there may be a gene-environmental interaction."
The concept was especially useful when expressed quantitatively, in the face of behavior geneticist's attempts to attribute percentages of variability in a trait to environment versus to genes. It also was the basis of a useful rule of thumb phrase for non-scientists – "But only if." "You can often say that Gene A causes Effect X, although sometimes it is more correct to say that Gene A causes Effect X, 'but only if' it is in Environment Z. In that case, you have something called a gene-environment interaction."
What's wrong with any of that? It's an incalculably large improvement over "nature or nurture?", especially when a supposed answer to that question has gotten into the hands of policy makers or ideologues.
My problem with the concept is with the particularist use of "a" gene-environment interaction, the notion that there can be one. This is because, at the most benign, this implies that there can be cases where there aren't gene-environment interactions. Worse, that those cases are in the majority. Worst, the notion that lurking out there is something akin to a Platonic ideal as to every gene's actions—that any given gene has an idealized effect, that it consistently "does" that, and that circumstances where that does not occur are rare and represent either pathological situations or inconsequential specialty acts. Thus, a particular gene may have a Platonically "normal" effect on intelligence unless, of course, the individual was protein malnourished as a fetus, had untreated phenylketonuria, or was raised as a wild child by meerkats.
The problem with "a" gene-environment interaction is that there is no gene that does something. It only has a particular effect in a particular environment, and to say that a gene has a consistent effect in every environment is really only to say that it has a consistent effect in all the environments in which it has been studied to date. This has become ever more clear in studies of the genetics of behavior, as there has been increasing appreciation of environmental regulation of epigenetics, transcription factors, splicing factors, and so on. And this is most dramatically pertinent to humans, given the extraordinary range of environments—both natural and culturally constructed—in which we live.
The problem with "a gene-environment interaction" is the same as asking what height has to do with the area of a rectangle, and being told that in this particular case, there is a height/length interaction.
