Tuesday, November 29, 2005

Unnatural groups   posted by Razib @ 11/29/2005 05:52:00 PM

We have talked in detail about the problems with group selection before, so before someone assumes that I am promoting that thesis in the generality I want to caution that I am not. But, it occurred to me today that I am convinced that a high ratio of intergroup:intragroup variance on cultural shibboleths is attainable. If your tribal members are characterized by a splotch of red paint on their forehead, while your traditional rivals brandish green forehead splotches, I think one can safely assume that variation of forehead color between the groups exceeds the variation within the group. This is in contrast to genetic variation, as neighboring groups (e.g., Serbs and Croats) will likely exhibit far more within than between group variation, in other words, the two population subsets will intersect over the preponderance of a given sample space.

In contrast to the genetic case cultural shibboleths can be directly and consciously manipulated. A trivial case would be the semi-hostility that was common in my middle school in western Pennsylvania when the Steelers and Browns would play each other. Sports ties and affinities are partly heritable, but, I also know of many individuals who consciously chose to diverge from the conventional familial allegiances to "make a personal statement." Jerry Seinfeld once remarked that what you are really rooting for are the jerseys, if a given player is traded to a rival team you are likely to detest him, simply because he has switched sides. I would contend that this phenomenon, often puzzling to those who have little interest in organized professional sports, is not learned, but cued to innate cognitive mechanisms. Recall that Constantinople was almost destroyed by rioting as the rival chariot team factions of the Blues and Greens joined forces against the imperial regime of Justinian and Theodora, not only are these passions relating to sports not modern, they can be extremely powerful in mobilizing group action.

Research reported in The Nurture Assumption and Not by Genes Alone suggest that a reflexive "groupishness" does exist in humans. Some of this groupishness seems almost perverse, for instance, studies which show that people tend to give more to individuals who they don't know, and will never meet again (controlled psychology experiments), when those individuals are arbitrarily labeled as members of the same meaningless group (e.g., "you are people whose favorite brand of cereal is Cheerios, while the other group likes Frosted Flakes"). But taking into account the behavior in relation to sports teams, I don't know if we should be that surprised. In Grooming, Gossip and Evolution Robin Dunbar argued that the maximum size of "natural" social units probably does not exceed more than a few hundred individuals. By natural, I mean social groups which don't need to be mediated by legal coda or formalized rules, but rather are managed through direct interpersonal interactions. It may be that there is a structural constraint upon how far the human mind can scale (the combinitorics of modeling social networks implies exponential increase in complexity as the number of individuals increases). It might also be that groups were never much larger than a few hundred individuals. I suspect it is likely a combination of both, though the former is probably a cap (to some extent) on the latter.

In any case, one could posit that for the vast majority of modern human existence small groups have been the operative units of human organization. One could take this to be the environment under which our innate intuitions were shaped. There are debates about the extent of massive modularity and the extent of hard-wired innateness. I will ignore the details and assert that though the extent of innateness may very, it seems highly plausible that some "mental organs" do exist. Language for instance is a good candidate. A minimal degree of innate social competence is also another one (a Theory of Mind). There are some that are fixed in human populations, while others may vary as a function of group, and likely there are many which vary throughout the population (perhaps because Evolutionarily Stable Strategies coexisting). But in any case, the omnipresence of groupishness across human societies indicates that it is a trait which has had widespread value, just as language has. This doesn't mean there has to be a "group module" in the brain, lodged within the "social organ," rather, there might be a few simple integrated heuristics, starting with "do with the majority does," which generate the groupish behavorial phenotype.

I began thinking of the idea that humans are adapted toward cuing onto somewhat absurd shibboleths as group identifiers after re-reading this old post of mine over after Matt McIntosh quoted me in regards to how people have problems categorizing groups in the modern world. The gist of what I was saying is that it makes sense that categorization schemes that are optimized for small groups would be maladapted to the modern world where groups have become much larger and amorphous. This post is an attempt to be a bit more clear in regards to that point: in the pre-modern world small groups could enforce on group identifying markers or behaviors upon members to a very high degree, ergo, intergroup variance could be maximized so that generalizations would carry a great deal of weight and predictive power. But now consider modern American Jews. Many Jews by declaration eat pork, enjoy shrimp and work and play on Saturdays. But they are still "Jews." Some still practice traditional Rabbinical Judaism, which was dominant between 500 and 1800 in some form throughout the Jewish world, but this group is not numerically the majority of identified American Jews. Infact, one could argue that in regards to many values non-Rabbnical Jews and gentiles now have more in common than non-Rabbnical Jews have with those who adhere to halakah in a strict fashion. In a similar vein, in The Future of Religion Rodney Stark and William Bainbridge found from survey data of habits and preferences that conservative Christians are a cultural outgroup in relation to non-conservative Christians and seculars. Even though Christians notionally belong to one group, in practice the norms and values of believers in this religion in the United States exhibits such a high level of intragroup variance that the "Left" half of the religious distribution intersects almost perfectly with the values of secular Americans who disavow any religious values or beliefs.

The big picture point is that we live in an age of unnatural groups and coalitions, where shibboleths that once might have had life and death implications (i.e., your tribe abandons you if you violate their taboos) have become cross-cultural affairs, and those groups now encompass billions (Christians and Muslims). Where in groups of 50-200 innate cognitive mechanisms could work to enforce strict adherence to a host of common practices and values to a group demarcated by narrow set identifying practies and beliefs, the same notions of group identity are now abstracted to "believers in Christ" or "the Ummah" where only the core practices and beliefs themselves are points of commonality. In pre-modern times groups were smaller, and lived in close proximity, today there is far less possibility of enforcement values across continents. Muslims might not eat pork, but I know from personal experience that South Asian Muslims prefer Indian food to Arab food, and that Arab Muslims prefer Eastern Mediterranean food, regardless of religion, over the food of Pakistan. Nevertheless, the same cognitive mechanisms which react to other groups can get triggered and make us react as if a group of 1 billion was functionally equivalent to a group of 100, even though the reality on the ground might be characterized by a great deal of variance in practice. A tribe of 100 is a far easier entity to generalize about than a tribe of 1 billion becuase a tribe of 100 can synchronize their practices and preferences far more easily. One might even ask if there can ever be truly a tribe of 1 billion in any real sense. One can see that these pancultural and continental groups, often under the aegis of universal religions, attempt to recapture the reality of small groups via their rhetoric of brotherhood and fellow feeling. And to some extent, it works, though not perfectly. The imperfection explains the temporary coexistence of liberal and segregationist Democrats within the same party for decades. In explains the modern day coexistence of libertarians and social conservatives within the Republican party. Residual notional affinity, often derived from past short-term utility driven alliances, can have a strong cohesive effect. But nevertheless, the reality is that the dynamics of the two structures, small groups and large quasi-groups are different on many points, and that is where our mental models start turning out unsatisfactory suboptimal results. The reality is that though liberal and conservative Democrats were of the same official kind, they divided their loyalties and had a host of other affiliations which worked at cross-purposes with their political identity (ACLU vs. White Citizens' Councils). Multiplicity of group allegiance has always existed, St. Paul was a Jew, a Christian and a Roman citizen simultaneously, but, I would argue that the differentiation of identities has exploded in the modern era as small scale groups have dissolved in the face of modern nuclear families, consumerism and geographical mobility, and the power of modern communication which has resulted in the dominance of the macrogroup in the mindshare of most individuals (i.e., American vs. over resident of East Village, Muslim as opposed to a citizen of Britain!). The synergy between our cognitive templates and the power of grand group rhetoric result in the disjunctive tendencies in discourse and practice in everyday life.

Addendum: I could go on for many pages on the many examples of disjunction between a) macrogroup identity b) reality on the ground and c) the tendency to extrapolate lower order relevant cognitive biases onto a), and the tensions it causes with b). But, I want to suggest on example that I have wondered about, in part because of my own ignorance in this area: language. One of the major issues I would have with Cavalli-Sforza's charts where he shows the concordance between languages and genes is that it ignores the reality that while gene frequencies are usually clinally varying as a function of space, languages are more sharply differentiated. Wait though, back up a minute...a few years ago I read in The Power of Babel by John McWhorter that the sharp boundaries between Slavic languages in the Balkans are artificial, that in reality the dialects graded into each other clinally. McWhorter said to the effect that "Bulgarian" and "Macedonian" on either side of the respective borders of these two nations would likely exhibit more similarity than two dialects of "Macedonian" at opposite ends of the nation. I know that in relation to Turkic languages this is clearly true, that the different national languages (i.e., "Uigher") are to some extent a hodgepodge of dialects thrown together due to the vicissitudes of history.

OK, so those sharp boundaries aren't always real. But it gets more complicated, because I'm sure you would object that "national Bulgarian," which is the official language, is standardized and sharply differentiated from "national Macedonian." Fair enough, but that gets into the issue of diglossia, the tendency for their to be a "high" elite literate language and a "low" dialect. While the high written languages might be distinct the low dialects grade into each other because of their scales of variation. A few years ago something really strange happened to me though...my "second language" is Bengali. Normally, the few Hindi movies I have watched I could only really make out 25% of the words clearly (depending on context and cues), so that they were basically unintelligible to me. When I heard spoken Hindi, or its sister dialect, Urdu, usually uttered by elite individuals (they almost always had Ph.D.s or M.D.s), it was also basically Greek to me (not quite, but it wasn't intelligible in more than the vaguest sense). Nevertheless, an acquaintance passed recently to me a print story in a English newspaper where there was an exact transliteration of the local argot of eastern Uttar Pradesh state, where Hindi is the official language. And guess what, I understood 80% of it! And the honest truth is that "high Bengali" is almost as Greekish to me as Hindi. So here we might have a situation where the colloquial dialects across a wide geographic range share more commonalities than the elite languages which serve as their variant national vehicles.

The whole point of that digression is that this shit is complicated. If you think I have a hard time being pithy, well, you try to wrap up those nested realities and relations into a few sentences (that's a rhetorical dare).

Addendum II: Another tendency, which I'd like to emphasize the back-projection of modern perceptions of differences between factions. To give examples, the Blues and Greens in the Nika riots are often said to be proxies for the aristocracy and aspiring mercentile classes, respectively. My understanding is that this is grossly over-exaggerated in the attempt to find a 'rational' explanation from the factions. Similarly during late republican Rome the populares are often portrayed as 'progressive' and the optimates as the 'traditionalists.' The reality is that the optimate dictator, Sulla, was as 'traditionalist' as the modern Chinese state is communist. The typologies gloss over the fact of capricious and stochastic nature of factional alliances contingent upon personal relations and random acts of history.