Tuesday, June 10, 2014

Genes vs environment: more on the misunderstood subtleties

Yesterday we tried to explain why the idea of a trait being due to genes versus environment was a false distinction.  We proposed two basic theorems of life, that everything is due to genes, and everything is due to environment, just to make the point that arguments about whether one or the other is 'the' cause of a trait of interest are often in never-never land.  Also, even saying 'both' are true, or that they 'interact' is so vague as often to be rather meaningless or unhelpful.  Do we mean pairwise interactions between factor A and factor B?  Is it just multiplicative (A x B yields trait)?  Or, say, A x B-squared?  Or Pi times A divided by B?

And, by the way, how many factors interact?  If it's very many, it's essentially impossible for observational data to resolve the causal process, even if just in pairs.  And are the effects just linear?  We are after all 4-dimensional (space and time) organisms, and action of some factor or factors need not be linear.  Maybe their 'interaction' is like this:

             T=G1 + E1(E5*E5)+ (G2*G3*E2*E11/2)/(π*G4+G5)

This equation is imaginary, but one might ask why on earth would pi or these variables raised to strange-seeming powers ever be relevant?  Because, for example, genes are expressed in quantities like copies of the protein, but they act in 3-dimensional cells with 2-dimensional surfaces interacting with environments, changing over time, that may themselves be linear doses (like amounts of, say, sugar) but may work through other dimensions, like pressure on cell surfaces....)

All this raises another, deeper question that is often if not typically quite misperceived, usually innocently, in the argument.  It is that there are deep semantic issues even in stating the propositions, so naturally there is disagreement about the answers.

More on the semantics of the problem
Part of the problem is the use of the terms themselves.  When we say that 'genes cause X' do we mean that genes (however you may wish to define them) are involved in the mechanism that leads to or maintains trait X?  This is the meaning sometimes used, such as in developmental biology when speaking of how retinal cells end up expressing opsin (light-sensing) genes.  Some regulatory region is bound by a protein and that transcribes the green-opsin in retinal cells, etc. 

But the genes vs environment argument often means something quite different.  If the first meaning of 'genetic' is about mechanism, a second meaning is about variation.  When we say that genes 'cause' something what we really mean is that genetic variation (among individual organisms) is causally responsible, through mechanistic means, for variation in the trait among those organisms.  So, variation in the coding sequence of opsin genes is associated with variation in color perception.  Likewise for variation in height, obesity, intelligence.  Even if what we choose to measure is a cultural choice and our definitions of traits culturally determined, what we choose to consider causation is in this case about inter-individual genetic variation, not the mechanism that's responsible.

The same, of course, goes for environment.  Smoking mechanistically causes heart disease by (say) raising blood pressure.  But epidemiologically, the amount of smoking is associated with the risk of heart attack.  This again is about causal variation.

And it is fair to ask just what one means, and how secure one's assertions are when a 'cause' of some trait--like hypertension or many others, raises your absolute risk by only a few percent (or less) and only by the time you are 50 or 60 years old?  When is that 'causation' in a seriously reliable sense?

The mechanism vs variation distinction is not always clear even when geneticists are speaking about genetics.  Much less for others who don’t really understand the subjects in much depth (e.g., many news reporters, even many outspoken scientists or people pushing some ideology about causation).

If we have to infer causation by relating variation in environmental conditions to genomic variation, we have a different kind of problem from working out mechanisms:  We have to decide how to quantify or measure our putative factors and their variation and to relate them to each other.   In both cases we may need either experimental or observational data.  If we do our science right, which is often far from the case, we should relate these to some theoretical framework that is more than trivial or self-confirming. Naturally, especially given the array of vested interests, perspectives, methodological preferences and approaches, along with the sociopolitical aspects of science and its funding, it is no surprise that the genes-vs-environment imbroglio is so persistent. Cultural inertia and our human fallibilities account for a lot of Einstein's observation about insanity being doing the same thing over and over and expecting different results.  

But even this is not the whole story.

There is a third 'genetic'
The word 'genetic' has an additional meaning, which has to do with the changes that occur in genes among cells within each person during his/her lifetime.  These 'somatic' mutations are inherited by our cells during our lifetime but are not inherited from parent to offspring.  We all harbor genes whose variation among our tissues or within segments of our tissues, affecting their mechanism of action, yield biological effects including disease.  Epigenetic changes in gene usage also occur in local tissues during life, but unless they are in the sperm or egg cells they are (usually) not inherited by the next generation, even if they may be inherited by your body cells when they divide.  There are various clear demonstrations of the importance of somatic genetic changes, and they are not controversial (but usually ignored by scientists, acting like ostriches, because somatic changes are devilish hard to document).

None of these things are secret!  But they show how difficult it is to understand biological causation of complex traits, even for someone not committed to some simplified worldview being defended against all comers.  A lot of confusion, hot air, wasted resources, and unnecessary competition results from failure to understand these things clearly.  Thinking about them seriously can give you a nasty headache, because the realities of Nature are not divided into neat categories, and our means of peeking into Nature's truth are limited, or even at present so deeply flawed that some better ways of thinking or even some better ways of identifying the important questions are called for. 

Thinking about these issues should also raise the difficult but obvious problem that what we choose to call important--be it environment or genes or how we define our trait in the first place--is a subjective decision, often not that clear among peers and often changeable over time (e.g., the definition of blood pressure and how to measure it, the definition of 'hypertension' or 'intelligence', or even a 'thumb').

What is the trait (if it's a trait)?

A philosophical point about genomic and evolutionary causation
One can argue, in principle, that if Nature and natural selection generated life, it would favor less rather than more finely tuned genomic programming.  Given the inevitably probabilistic aspect of survival, finding mates and food and so on, in environmental circumstances that are usually complex and changeable, fine-tuning might seem rather implausible  The default assumption should instead be a lack of rigid programming and a stress on ability to respond to unpredicted circumstances. Adaptability would seem to be obviously more robust than genetic hard-wiring.  Yet too much theory, even by evolutionary biologists and manifestly by those in other fields (including much of biomedicine and anthropology) looks for the hard-wired--it's easier to think about and design projects to find.

Instead, perhaps the burden of proof should be on anyone claiming that their trait, be it disease or adaptation or behavior etc., is hard-wired.  One should have to falsify the more likely assumption that what passes through the sieve of Natural selection is that which can find the holes in the sieve, wherever they may be.  In that sense, causation is 'designed' to be complex and elusive.


Anonymous said...

Adaptive landscapes are depicted as hills with 'up' meaning more adaptive. But maybe we should think of adaptation as an inverted cone, which while snug can trap the organism as if in a well. (Did I steal that metaphor from Larry Moran? Not sure.) There needs to be wiggle room. At the population level that's afforded by drift.

At the individual level it's afforded is the offloading of specification of traits onto developmental processes rather than genes (actually offloading is misleading because some were never 'onloaded' onto tight genetic control in the first place). See Gerhart & Kirschner, Terrence Deacon, J. Scott Turner.


Evelyn Fox Keller

"That form of the debate and the concept of putting nature and nurture in counterpoint is a very Anglo-Saxon idea, and it’s very recent. It really comes with Francis Galton, who introduced the notions of nature and nurture as alternative causes that could be separately weighed. ... To be sure, people talked about nature and nurture before, but they didn’t juxtapose them in that way; rather they tended to regard nature as seed, nurture as the cultivation of that seed."

Ken Weiss said...

Yes, and throw in niche construction, organismal selection (where organisms use that part of their environment that they 'like'--are genetically suited to do well in), developmental selection (where gene-gene-environment interactions don't work and an organism never appears), and a hefty dose of chance (probabilistic intersections between elements of genome and environment), and you've got a more realistic picture.

I'd say that Keller's assertion is about the formality of the distinction, but that the idea of inherent worth vs circumstance can be found all over the place informally, if one wants to look for it. But Galton really formalized it in modern (Darwinian) terms, and that along with other early 20th century science drove the distinction farther.

Of course, quantitative vs qualitative thinking may be even deeper, and genes vs environment another manifestation.