Thursday, February 8, 2007

Genetics

Stumbled across two articles today on the impact of genes on behavior, the first on passing bad behavior to kids, the second on beauty. I have never liked these studies, in part because they rest - usually - on flawed premises and leaps of logic. After all, the mere idea of a gene is largely undefined or vague. How does it differ from, say, triplet codons or DNA? How does the genome differ from all of an individual's chromosomes? What does it add to our knowledge. Not much, except in so far as it is a holdover from Mendelian studies on heredity. It is, in a sense, a mask for old science working under the guise of the new.

And this has consequences for the studies that people conduct now. In those two pieces, sweeping claims are made about the inherited nature of certain characteristics. The first one on bad behavior, for instance, compares identical and fraternal twins and the kids they have. An interesting experiment, no doubt. But it's fundamentally flawed. It assumes that variations in behavior are largely the result of "genetic" influence, even though the experiment cannot say what actual pieces of DNA lead to what specific behaviors. There is no line of causality. And this comes out in a critical section buried at the bottom of the piece:

"According to Harden, it's possible that genes involved in risk-taking, sensation-seeking and other aspects of antisocial behavior may make parents more likely to clash, and, when passed on to their kids, make conduct problems more likely."

But once you start allowing for what is possible, once your experimentation is shoddy enough that cannot determine cause and effect, you open the door to all kinds of methodological questions. For example, from the experiment, how is it possible to tease out the effect of learning, teaching, and development as opposed to DNA, genes, and chromosomes? You simply don't know, and it's therefore the researcher taking his or her best guess as to the cause. That's not science. That's belief.

The same can be said for the second article. Our standards of beauty as evolutionarily derived. Again, a key passage:

"Our brains did not evolve with media, and many people see more media images of women than actual women."

The problem here is that humans never stop evolving, and it's odd to think that scientists with a background in evolution would consider that suddenly most or even all change mechanisms stopped once people left the African savanna. And there is the fact that the standards of beauty have altered dramatically and differ widely. Light skin versus dark, proportions of hips to chest, the list could go on and on, but sociobiologists often take samples from particular cultures or societies, which misses the variation that's out there.

We have ample examples from the past where people have assumed the "innate" nature of certain individuals. Women are biologically unsuited for certain public positions, blacks are perenially a subclass of humans who couldn't perform higher orders of thinking, etc. In modern times, all of these ideas were based on pseudo-science that did not maintain the necessary level of methodological rigor that would have culled out the influence of popular perception on verifiable theories. The new wave of biology in human behavior doesn't do much better. The methods are more compelling, but the crippling, flawed assumptions are still there.

8 comments:

HoBs said...

"It assumes that variations in behavior are largely the result of "genetic" influence, even though the experiment cannot say what actual pieces of DNA lead to what specific behaviors. There is no line of causality"

There is no doubt causality that it is genes that affect behavior. This is basic scientific method stuff. You are quite right (I have criticized these studies many times for the same reasons) that it is likely not as simple as a bad behavior gene. The msnbc article itself concedes that. But it is must be genetic. That much can be proven.

Assume Fraternal twins and identical twins are randomly distributed. (this is a fine assumption assuming you take out of your sample fraternal twins born from parents seeking treatment for infertility).

Then, any difference in argumentative between fraternal and identical twins must be due to their identical/fraternal twin-ness.

Any other stories you spin have to be consistent with that. (Ok, one non-genetic possibility is that growign up with someone that looks just like you makes you more argumentative and that may confound the results of the study, but the far more likely explanation is that something genetic is responsible.) It could be something as simple as looks of course.

Maybe since prettier people are treated nicer and tjey are therefore are less argumentative. And since prettieness is inherited, that's what drives their results. But at some level the cause must be genetic.


And yes brains are evolving, but really there is not much selection pressure. If you read Darwin, you need selection pressure for evolution. And right now, the forces are pretty weak. We still haven't evolved to compensate for the bacteria and viruses that have been around for centuries, so that our bodies react inappropraitely.

Chengora said...

Again, you're missing some basic information about evolution. You don't need selection pressure to evolve, for example. This is a simple mistake that many sociobiologists make. Evolution does not simply drill down to natural selection. It encompasses much more: genetic drift, random events, mitotic and meiotic mutation, etc. Evolution, at its base, is a theory tacked onto the observable changes in organisms. But natural selection, while an important factor in these changes, is not the be all and end all of evolution.

The only reason why you need selective pressure in looking at these theories is because they are a (or only) causal factor for explanation. Once you crack that assumption apart, which is ridiculously easy, you no longer have the straight answers you're looking for.

For example, your fraternal/identical twins postulate is interesting, but it relies upon a common misunderstanding of methodology. Take another example: gender roles. One of the main arguments for genetically derived gender roles is that all women "play hard to get", and all men seek to get as much as possible. Since this holds across cultures (which isn't true, but this is the argument), we can assume that genetic factors are the only ones involved.

This of course misses some critical factors. There's no evidence for the assumptions on the one hand. Even within cultures, people have different "sexual strategies" and preferences. This opens at least a door for variation and, more importantly, learning. And homologies that are used to prove this don't hold up. For example, human behavior is often compared to that of colonial birds, since we superficially exhibit the same patterns. But digging deeper reveals that the proximate and deeper causes of those patterns are nowhere near similar. And of course, while the process of evolution occurs on the DNA-level, the pressures or changes being impelled can respond to more than simply genetic ones. For example, environmental issues are up there.

The point is, there are major methodological problems with assuming that people (or creatures more generally) can't learn and pass on different behaviors through non-genetic procedures. It's a failure of evidence and an inappropriate restriction to only one factor in evolution (natural selection) which, while important, is not the only cause. Nor is it a straightforward progression or causality (hence, meiosis).

And this points to a deeper issue in your comments.

"And since prettieness is inherited, that's what drives their results. But at some level the cause must be genetic."

A common criticism of sociobiology is that it creates "just so" stories. Ideas are twisted and reformed to match the evidence available, but which violate methodological rules, don't offer much clarity, and can be contradictory. For example, homosexuality has been simultaneously explained and disproven through genetic explanations, with no experimental ability to discern between them because of the methodological flaws.

And so your statement is inherently problematic. First, there are different standards of beauty. And the lack of set standards in beauty (and they've obviously changed over time) means that at most, you can say people desire what they see as pretty, but no way to say what pretty is. That's a far cry from saying that genes lie at the base of our desires (how does that apply to career ambition, for example?), and it certainly opens the door - as it should - for learning, culture, etc. to either be built on top of that minimal genetic architecture, or to replace it entirely (think of soldiers in war). And of course, there are factors that blunt the impact of genetics. Think of welfare, healthcare, and every person who laments or lauds the fact that physically challenged individuals can still survive and have a meaningful existence.

At a certain point, yeah, you could say "everything is genetic". But the best studies in evolution and development have found that's not the case. DNA requires specific environments to be activated, and there are hints that this environmental conditionality persists even later in life. So you could just as easily say "the environment (nurture) is everything." The problem is teasing out these factors, and because of its adherence to a specific and flawed understanding of natural selection as encompassing most or all of evolution, sociobiology doesn't offer an effective means to do that.

HoBs said...

Sigh, I was on the train this morning listening to This American Life, talking about interviews with politicians, and how things never get resolved because points always get diverted and sidetracked. I agree and always had agreed with everything you said. They are unrelated to my point. Ira Glass amusingly describes a couple politicians quietly high-fiving after a particularly well parried question (that of course did not get answered).

What I was saying about fraternal vs identical twins is not just an interesting postulate, it is the basis of many peer-reviewed papers published in journals like Science and Nature.

If you take a random sample of fraternal and identical twins, all drawn from the same environment. Then if there are differences in outcomes for fraternal vs identical, then those differences must be due to the fact that being a fraternal twin makes a pair different than being a identical twin. Environment cannot explain this because they were drawn from the same environment. Thus any differences must be related to being fraternal or identical. The most obvious difference is that identical sets have the same genes, and fraternal sets have slightly different genes, but yes, admittedly, there are other differences (like having someone else in the world who looks exactly like you). But if you want to question the conclusion that genetics determines these differences, it has to explain why you expect differences between fraternal and identical twins.

Agreed?

Chengora said...

I understand what you're saying, but the issue is that while you agree with what I write, I don't think you fully comprehend all the ins and outs. It has much to do with the specifics of genetics and evolution, but the larger issue is that, while you may agree with my statements, you make statements that go against the logic and processes I describe.

Hence, the problem with your statement about prettiness. You essentially say, at base, things are because of genetics. I say, no, they are because of a combination of the two, which the current approaches in sociobiology can't disaggregate. That's a more appealing position, so you adopt it (consciously or not), but it's not reflected in what you said.

Now, I will say that the identical/fraternal twin case is a very interesting and innovative study. And you're right: many studies have been done off of it, although far more of those studies appear in journals on psychology, not genetics. However, the biggest problem is that when you say the samples are pulled from the "same environment", you're making a jump from there to saying that the cause is genetic. That's not entirely the case. The difficulty with these types of studies is that they take a superficial similarity, try to control all outside factors, then say there is an innate biological cause. But, what if, for example, the cause of aggression is developmental, not genetic? What if differing ph in the womb (as an example, not saying this is the case) between identical births and fraternal births is the cause? The lack of causality - the inability to explore specific processes - makes it impossible to say what is the exact cause. Simply because social environment is taken out doesn't mean that the environment is taken out. I think that's the chief issue with sociobiology: the view of the environment is so small that it truncates meaningful inquiry into other areas. And as a result, there is too often a knee-jerk jump at a genetic explanation, when other factors and the causal processes, have not been explored.

Which is where the example I gave comes in. For gender roles, researchers attempt to hold the environment constant by arguing that all women and men act in certain ways. Environmental issues are therefore eliminated, and the reasons must be genetic. But, as I said, the environmental conditions which activate genes are pervasive, and can impact things like prenatal development and a host of other issues. As a result, it's not possible to say, convincingly, that "it is must be genetic." as you do.

As I said, I think this is an interesting experiment, but not one that clearly demonstrates that genes are the basis upon which behavior is founded. I have no doubt that genes have an impact, but that's not going to be settled by this test.

HoBs said...

Yes, fully acknowledge things like
- ph in the womb
(that's a good one)
could matter.

I also acknowledged factors like growing up with someone who looks just like you could matter.
Or there could be other factors that lead to fraternal twinism like fertility treatments.

But what the studies do do, is they make a huge amount of progress is ruling lots of things out, which it sounds like you acknowledge too. We went from tons and tons of environmental factors to just

- ph in the womb
- fertility drugs
- growing up with a doppleganger
- (probably a few more, but not too many)

That's progress.

About whether it's all genetic, no doubt how genes work depends on the environment. It is just a matter of perspective. I was told of a conversation between a geneticist and a anthropologist on this subject. The anthropologist had the epiphany that to the geneticist, everything was genetic, in the sense that sure, maybe what we call pretty is socially constructed (that is debatable of course but at least to some degree is true). Both would agree that the genes (mostly) determine your appearance, and society determines (mostly) how that appearnance is interpretted, so what is the "cause" of prettyness is a matter of intepretation.

MC Escher was fascinated by stuff like this. There are nice zen koans too, that I can't think of. Like when you clap your hands, it is your left hand or right hand that makes the sound (that is not so clever, there are better ones).

Chengora said...

Good, there's much more area of agreement here. And I fully agree that those studies are good at culling things out. But methodologically, be careful. Culling things away and calling it progress...it is progress of a sort, but we're dealing with unknowns here. It's impossible to tell that we've effectively culled the possible factors down to a relatively small number, and absolutely no reason at all to assume that genes are the answer and therefore that the behavior is hereditary instead of conditional (of being an identical twin, for example). So, we've pulled away a great deal from what I take to be your initial statements, but that's fine.

And of course this impacts the "all about genes, all about environment, both" thing. It's not simply a matter of perspective, at least not for sociobiologists and evolutionary psychologists. The inability to examine and test causal processes is a huge stumbling block, mostly because, in speaking in specific terms, those individuals don't assign "factor status" to the environment. And usually, when you get to that point where the experimenter is just throwing up his hands and saying "whatever", you've hit a bad experiment that has missed some critical factors.

Also, don't forget, genes don't really determine much of your appearance. Sure, coloration and to an extent features. But what we do to ourselves is the most important thing (weight, plastic surgery, dental work, exercise, color dying, clothing, personality, confidence, etc.). Interestingly, most sociobiologists seem to focus on the face as a marker of "beauty", even though your skin is technically your largest and most visible organ. Hmm...I wonder if there's something more there.

hcduvall said...

...I feel like I'm bounding in again, but I think fundamentally Chengora doesn't think enough of the complexity of human behavior has been addressed yet, or can be compensated for, to even really begin these sorts of studies/experiments. The fact that people are trying is not faith enough. Mainly, they're not ready to rule out as much as they do for full control, and make prouncements from it. To borrow from a visual metaphor for complexity (it was used by some guy for illustrating what he was trying to do with his ai research, and I find it exceeding useful)...say we sum up a thought, concept, cultural moral as a light of a certain frequency, but that you get that from a laser passing through unknown number of prisms and mirrors to hit just that frequency. I think Chengora thinks they're more mirrors and prisms out there that makes fiddling with five useful. I'll stake the ground inbettween to say that the data is probably useful, even with possibly flawed methodology, but it still sounds gun jumpy. That they haven't gotten all the major influences accounted for, let alone all of them (which is perceptiably impossible until we sublimate into the ether and become energy beings).

Is there a term for the study of how people study, specifically? Not jsut how people think, I mean. It sounds like the study of science could benefit from borrowing a bit of literary theory: maybe starting with bloom and the anxiety of influence or that school resentment business.

HoBs said...

lots of literary theory directed toward the study of science. the germany critical theory school, which lead to Foucault, and Habermas...

the study of how people study is one of the study of knowledge words that i vaguely am getting a grasp on, but get mixed up.

epistemology is the study of what we know. maybe that's close? but i think there's something better.
not ontology or teleology... anyway