Thursday, April 13, 2006

No Two Alike: the role of noise   posted by agnostic @ 4/13/2006 08:15:00 PM

I. Introduction

Almost ten years ago, Judith Rich Harris (The Nurture Assumption) reviewed the literature on the tiny role played by the "shared environment" -- i.e., anything two people growing up in the same household share -- in shaping adult personality and intelligence, a central finding of behavior genetics. Depending on the particular trait measured, roughly 40 - 50% of the variance in personality and intelligence among individuals in a population is accounted for by genetic variation; typically 0 - 5% is accounted for by the shared environment; while the rest is lumped into the category "nonshared environment." It is this latter category to which Harris now turns her attention in No Two Alike. That is, what causes identical twins reared together to turn out any other way than exactly identical?

II. Micro noise

Rather than lay all her cards on the table at the start, she crafts the book to read like a detective novel in which suspects are eliminated one by one, only to reveal her original thesis near the end. However, I'll take a more conventional approach in reviewing her work by getting straight to the punchline, stipulating that none of the usual suspects (birth order, etc.) is the real culprit -- except for "developmental noise" (DN). Her phrasing of DN (p. 46):

There are biological differences between identical twins, due to unpredictable things that happen before and after they are born. A neuron zigs instead of zags. One fetus occupies a better position in the uterus. One twin falls down the stairs or contracts a virus. Exposure to pathogens has been proposed as a possible etiological trigger both for schizophrenia and for type 1 diabetes. The schizophrenic twin may have developed the disorder not because her life was more stressful but because she happened to encounter a virus at the wrong time.

I'm going to belabor this point, looking only at infection, since Harris did not elaborate on it, as she was concerned with social causes of personality differences. In fact, the boldfaced phrase above could be amended to say that "both twins contract a virus, but in one, and not the other, it travels to the nervous system." Sound crazy? This is just what happens with polio (pdf), which typically resides in the gastrointestinal tract, but occasionally gets into the nervous system:

Up to 95% of all polio infections are inapparent or asymptomatic. Estimates of the ratio of inapparent to paralytic illness vary from 50:1 to 1,000:1 (usually 200:1). . . Fewer than 1% of all polio infections result in flaccid paralysis.

And consider the MZ and DZ concordance rates (in %) for the following known infectious diseases, where presumably both twins were infected (Vogel & Motulsky, Human Genetics, p. 237; search the Amazon page for "tuberculosis" and flip to that page):

Measles - MZ = 97, DZ = 94
Scarlet Fever - MZ = 55, DZ = 47
Pneumonia - MZ = 32, DZ = 18
Tuberculosis - MZ = 53, DZ = 21
Leprosy - MZ = 59, DZ = 10

If MZ twins are barely more concordant than not for tuberculosis, and more discordant than concordant for pneumonia, the chain from infection to symptoms must incorporate a fair amount of chance. In principle, the same thing could play a role in individual differences in personality. I earlier wrote (here, here, and here) about why we should expect something like this. Briefly, the very existence of the brain-blood barrier (BBB; kiddy link with more sophisticated links therein) suggests a timeless struggle to keep microorganisms out of our control center. Because microbes evolve (and quickly), we shouldn't expect the defense to hold up forever, nor against all invaders. Aside from known infectious diseases of the brain such as bacterial meningitis (which uses a Trojan Horse ruse to cross the BBB), the mere fact that we can chemically affect mood and personality using drugs such as Prozac demonstrates how easily microbes could invade the human central nervous system and alter mood / personality.

This is all the more likely in regions of the brain without a BBB, such as the hypothalamus, which in turn is part of the limbic system, thought to be the emotional center. Moreover, the BBB is not fully formed even in toddlerhood. In any case, this is where you would want to head if you were a virus that had long-term (rather than acute) plans for your host -- commandeering the visual or auditory systems wouldn't do much for you (if anything, it'd make the host blind or deaf); ditto for the seats of higher reasoning. If the human brain is like an airplane that you, the long-term-minded virus, are hijacking, you want to keep pretty much everything in working order -- except the part that deals with setting the destination, which is like the motivational and personality components of the brain that set goals and steer toward them. All a virus would have to do is supply a different set of coordinates to the pilot; everything else would take care of itself.

As an aside, some viruses might not even be keen on hijacking the brain -- some might merely be interested in a preemptive strike against the brain's most sensitive pathogen-detectors: our "chemosensory" senses of taste and smell (we can't hear, see, or feel microbes). It so happens that parts of the limbic system (such as the amygdala and hypothalamus) also deal with these less conscious aspects of taste and smell (e.g., how palatable something is, as opposed to discriminating two brands of soda if you were blind-folded). So, a virus which attempted to knock out our pathogen-detectors might cause minor collateral damage that subtly altered personality.

The notion that a tweak so subtle could turn one from an introvert to an extrovert may sound exotic, but in a 2002 review (pdf) of animal models of infectious obesity, the authors emphasize:

Noteworthy in our experiments was the absence of gross histological changes in the brain of many obese animals; had the fact of an initial virus infection not been known, there would be no virological or histological evidence that the observed obesity syndrome had its origin in a preceding viral infection.

In other words, an infection both fleeting and nearly impossible to detect (unless you already knew what to look for) had a large effect in turning infected animals obese. By the way, the part of the brain infected in their studies was the hypothalamus. In sum, I see no reason why something similar couldn't make one more extroverted, less conscientious, and so on.

III. Social noise

Moving on, her original thesis is that some of the nonshared environment consists of differences caused by your competing for status by examining other people's mental "profile" of you. She proposes three psychological mechanisms to account for this: the relationship system, a mind-reading device, and a status system.

The first is like a mental rolodex whose purpose is to keep track of socially relevant information of individuals: their face and name by which to uniquely identify them, their relationship to you (brother, close friend, boss, spouse, etc.), and other information relevant to how well they can contribute to your survival and reproduction -- are they mean or welcoming, bright or dull, do they return favors or not, do they still owe you any favors or not, are they sexually interested in you or not, and so on. The mind-reading device is the standard theory of mind module used in many cognitive & evolutionary psychology models of how our mind is built. In essence, its purpose is to allow you to peer into the mind of another individual to facilitate social behavior -- does that narrow-eyed look she's giving you mean she's trying to threaten you or seduce you? (Or both, perhaps, if it's role-play.) What did she mean when she put a funny intonation on "I'll" as in "All right then, I'll do the dishes"? And so forth. Last, the status system pushes you to attain high status within your peer group, which is mostly based on physical dominance in young males and good looks in young females. Since in pre-modern times higher status (so defined) likely increased reproductive success monotonically, the purpose of this system is clear enough: it compels you to reach a status level where you'll have the most kids.

Now, here's how she accounts for the non-shared environment: part of personality formation comes from you adjusting your personality settings in light of what others think about you, a process that plays out really in adolescence, not infancy. That is, you use your mind-reading device to see what information is listed in your profile in someone else's relationship system. However, because you want a large sample size to base judgment on, you look into the relationship systems of everyone who has an entry for you and average the results. This composite profile is what the "Generalized Other" thinks of you, and you adjust your personality accordingly. If they regard you as a diminuitive wimp, it won't pay in the status competition to be aggressive and cocky. If they think you're a dependable leader, it pays to be more extroverted. This ties in to the non-shared environment because identical twins won't necessarily have identical composite profiles once they've read the minds of the Generalized Other -- e.g., one may owe more favors than the other, and thus may be perceived as less cooperative.

I'll quote her at length since this is the core of her thesis. Here are the three ways she believes this interaction between the three devices can produce personality differences between identical twins reared together (p. 230-1):

[W]hen [MZ twin 1] tries to read what you've recorded about him on his page in your lexicon [i.e. relationship system], he will see something different from what [MZ twin 2] sees. . . They don't get the same social feedback because your relationship system has discriminated between them. The relationship system is far more interested in the differences between people -- the things that make each one unique -- than in their similarities. (original emphasis)

In other words, to the extent that MZ twins have non-identical profiles, they will respond in non-identical ways. However, recall that a person examines the average profile of themselves which they've read from the minds constituting the Genaralized Other -- not just of one other person. Therefore, this scenario could produce differences only if the average profiles were different -- and that just pushes the question back a level: what led to MZ 1 and MZ 2 having different profiles in adolescence? Since under Harris' view their genes and almost all relevant social experiences have been near identical up to this point, it must be that chance factors large or small have resulted in this discrepancy. For example, Donald's amygdala was subtly lesioned by infection but not George's, resulting in divergent composite profiles gotten from the Generalized Other. So, this is due to developmental noise (DN).

Within a group of boys, only one can be the toughest. Within a group of girls, only one can be the prettiest. If a group contains a pair of identical twins -- even if they're both very tough or very pretty -- one will inevitably be second best and the other will rank higher in the attention structure. The twin who receives more gazes [i.e., the one the others pay more attention to] will tend to speak out more in the group and will consequently receive yet more gazes, so that what might have started out as a tiny difference will widen.

This differential attention-getting could be due to pre-existing differences -- which in this scenario cannot be due to their shared genes or shared environment, and thus would be due to DN -- or when they're together, by chance MZ 1 gets 49 of 100 gazes, while MZ 2 gets 51 of 100, which then snowballs via a postive feedback loop in favor of MZ 2. So again, this is just DN.

Last, social cues you use to read the minds of the Genaralized Other may be ambiguous:

[S]ocial cues that you repeatedly receive from others in your group may be based on things that shouldn't have any evolutionary significance at all, because they're random. You tripped and fell at a dinner party and landed in the salmon mousse. . . You made an offhand remark that was interpreted as a prediction and the prediction happened to come true. Incidents like these can give an individual a reputation that can persist for years, spread and perpetuated through gossip.

Harris admits that some random events don't have lasting consequences, but others do: e.g., "a random event can cause everyone in your group to think you're wise or clumsy or funny or brave, and to go on thinking that for years." The Generalized Other, in her view, may be responding in a predictable, non-random fashion -- thinking you're clumsy when you tripped into the mousse -- but the initial event was due to chance, like choosing the seed of a random number generator, which then obeys a rule-governed algorithm for arriving at the result. Once more, we come back to DN.

To sum up: all three reasons Harris gives for why the interaction between the three devices can account for personality differences in MZ twins are actually variations on DN. Her earlier treatment of DN was based on chance phenomena at the "micro" level of axons and hormones, but her proposal is just a "macro" or social version of the same. This is not a criticism, for although I don't think Harris' proposal is what she says it is (i.e., an alternative to DN), it specifies the mental mechanisms for how DN can occur systematically at the social level in personality formation, a novel contribution to personality psychology.