Evolving Thoughts

Evolution, culture, philosophy and chocolate! John Wilkins' continuing struggle to come to terms with impermanence... "Humanus sum, nihil humanum a me alienum puto" - Terence

Tuesday, April 25, 2006

No further developments in the Mirecki case

The Lawrence World Journal is reporting that the police have closed their files on the Mirecki beating for lack of evidence. At least they didn't try to fit him up with fraudulent reporting, which is what might have been expect when we heard they confiscated his computer.

Predictably, the wingnuts are claiming this is evidence he did fake the beating.

Monday, April 24, 2006

OK, a film meme

Whew. After the last few posts, I'm exhausted, and unable to say anything sensible (or rather, any more sensible than is in those posts). So here's a meme, taken from Janet Stemwedel's Adventures in Science and Ethics, films one ought to have seen. Janet has asterisked the science-relevant ones, I've bolded the ones I have seen:

*"2001: A Space Odyssey" (1968) Stanley Kubrick
"The 400 Blows" (1959) Francois Truffaut
"8 1/2" (1963) Federico Fellini
"Aguirre, the Wrath of God" (1972) Werner Herzog
*"Alien" (1979) Ridley Scott
"All About Eve" (1950) Joseph L. Mankiewicz
"Annie Hall" (1977) Woody Allen
"Bambi" (1942) Disney
"Battleship Potemkin" (1925) Sergei Eisenstein
"The Best Years of Our Lives" (1946) William Wyler
"The Big Red One" (1980) Samuel Fuller (Seen half of it.)
"The Bicycle Thief" (1949) Vittorio De Sica
"The Big Sleep" (1946) Howard Hawks
*"Blade Runner" (1982) Ridley Scott
"Blowup" (1966) Michelangelo Antonioni
"Blue Velvet" (1986) David Lynch
"Bonnie and Clyde" (1967) Arthur Penn
"Breathless" (1959) Jean-Luc Godard
*"Bringing Up Baby" (1938) Howard Hawks
"Carrie" (1975) Brian DePalma
"Casablanca" (1942) Michael Curtiz
"Un Chien Andalou" (1928) Luis Bunuel & Salvador Dali
"Children of Paradise" / "Les Enfants du Paradis" (1945) Marcel Carne
"Chinatown" (1974) Roman Polanski
"Citizen Kane" (1941) Orson Welles
*"A Clockwork Orange" (1971) Stanley Kubrick
"The Crying Game" (1992) Neil Jordan
*"The Day the Earth Stood Still" (1951) Robert Wise
"Days of Heaven" (1978) Terence Malick
"Dirty Harry" (1971) Don Siegel
"The Discreet Charm of the Bourgeoisie" (1972) Luis Bunuel
"Do the Right Thing" (1989) Spike Lee
"La Dolce Vita" (1960) Federico Fellini
"Double Indemnity" (1944) Billy Wilder
*"Dr. Strangelove or: How I Learned to Stop Worrying and Love the Bomb" (1964) Stanley Kubrick
"Duck Soup" (1933) Leo McCarey
*"E.T. The Extra-Terrestrial" (1982) Steven Spielberg
"Easy Rider" (1969) Dennis Hopper
*"The Empire Strikes Back" (1980) Irvin Kershner
"The Exorcist" (1973) William Friedkin
"Fargo" (1995) Joel & Ethan Coen
"Fight Club" (1999) David Fincher
*"Frankenstein" (1931) James Whale
"The General" (1927) Buster Keaton & Clyde Bruckman
"The Godfather," "The Godfather, Part II" (1972, 1974) Francis Ford Coppola
"Gone With the Wind" (1939) Victor Fleming
"GoodFellas" (1990) Martin Scorsese
"The Graduate" (1967) Mike Nichols
"Halloween" (1978) John Carpenter
"A Hard Day's Night" (1964) Richard Lester
"Intolerance" (1916) D.W. Griffith
"It's A Gift" (1934) Norman Z. McLeod
"It's a Wonderful Life" (1946) Frank Capra
*"Jaws" (1975) Steven Spielberg
"The Lady Eve" (1941) Preston Sturges
"Lawrence of Arabia" (1962) David Lean
"M" (1931) Fritz Lang
"Mad Max 2" / "The Road Warrior" (1981) George Miller
"The Maltese Falcon" (1941) John Huston
*"The Manchurian Candidate" (1962) John Frankenheimer
*"Metropolis" (1926) Fritz Lang
"Modern Times" (1936) Charles Chaplin
"Monty Python and the Holy Grail" (1975) Terry Jones & Terry Gilliam
"Nashville" (1975) Robert Altman
"The Night of the Hunter" (1955) Charles Laughton
*"Night of the Living Dead" (1968) George Romero
"North by Northwest" (1959) Alfred Hitchcock
"Nosferatu" (1922) F.W. Murnau
"On the Waterfront" (1954) Elia Kazan
"Once Upon a Time in the West" (1968) Sergio Leone
"Out of the Past" (1947) Jacques Tournier
"Persona" (1966) Ingmar Bergman
"Pink Flamingos" (1972) John Waters
"Psycho" (1960) Alfred Hitchcock
"Pulp Fiction" (1994) Quentin Tarantino
"Rashomon" (1950) Akira Kurosawa
"Rear Window" (1954) Alfred Hitchcock
"Rebel Without a Cause" (1955) Nicholas Ray
"Red River" (1948) Howard Hawks
"Repulsion" (1965) Roman Polanski
"Rules of the Game" (1939) Jean Renoir
"Scarface" (1932) Howard Hawks
"The Scarlet Empress" (1934) Josef von Sternberg
"Schindler's List" (1993) Steven Spielberg
"The Searchers" (1956) John Ford
"The Seven Samurai" (1954) Akira Kurosawa
"Singin' in the Rain" (1952) Stanley Donen & Gene Kelly
"Some Like It Hot" (1959) Billy Wilder
"A Star Is Born" (1954) George Cukor
"A Streetcar Named Desire" (1951) Elia Kazan
"Sunset Boulevard" (1950) Billy Wilder
"Taxi Driver" (1976) Martin Scorsese
"The Third Man" (1949) Carol Reed
"Tokyo Story" (1953) Yasujiro Ozu
"Touch of Evil" (1958) Orson Welles
"The Treasure of the Sierra Madre" (1948) John Huston
"Trouble in Paradise" (1932) Ernst Lubitsch
"Vertigo" (1958) Alfred Hitchcock
"West Side Story" (1961) Jerome Robbins/Robert Wise
"The Wild Bunch" (1969) Sam Peckinpah
"The Wizard of Oz" (1939) Victor Fleming


A few more than Jane, but then I'm ancient, lazy and have little discrimination. If it's moving and on a screen, I'll watch it. I didn't say I understood any of them...

Last one on race

So, as my commenters have noted, I screwed up. Badly. The paper I cited as having 54 races actually had six clusters those populations fit into. I don't mind - I said at the beginning I was ignorant of this, and these posts are my way of learning about the topic and getting a free education ("Yes! He can be taught!").

Before I respond to the very gracious response by Razib (I say "race", and you say "population structure"), which raises interesting philosophical questions, let me just tell you all my learning strategy. I have always found that it is best to take a stance on a topic, and make those who would educate or convince you do the hard work of getting you to change your mind. This way, issues have an edge, and they matter. The idea that one should hold judgement in abeyance in education strikes me as a way to promote rote learning and dogmatism.

But the philosophical issues here aren't undercut by my misreading science. I think that there are a number of deeper issues, which underpin our thinking on matters of detail. Let me see if I can outline them.

1. Classification

Typically we have classified throughout the western era in terms of resemblances. Racial classification is only one of these. There are two main ways to classify things, especially biological things: one is to take a resemblance criterion and identify those who approach or deviate from it. This is known as the typological way of classifying. You take a type specimen and identify all those who are near enough to it for inclusion. [Side note: despite loose usage by Mayr and others, typology is not the same as essentialism; most biologists have been typologists including modern biologists. Almost none have been essentialists.]

The other is by descent. Darwin introduced a real novelty into classification - ancestry. Previous biologists used ancestry as a way of isolating groups. In Cuvier's definition, for example, a species was all descendents of an original pair. Darwin conjectured that the distributions of groups within groups is due to shared ancestry between species. This has led to the phylogenetic manner of classification, sometimes known as "cladistics". But cladistics requires that lineages split. If a lineage recombines with related lineages on a regular basis, then cladistics is no use for classification. Such recombining lineages are called tokogenetic in cladistic terminology. The reason these are uninformative is that the "signal" of past history is lost. But it doesn't follow that the signal is lost for genes within that species - we can treat them as self-standing lineages that have recoverable histories even if the groups which comprise them are not clades.

I once used a figure in a paper to illustrate this:


Reticulate


We can recover the history of the genes (here, the characters) even if we cannot recover the history of populations as such. Genes are at best surrogates for population structure. But they don't, of themselves, make taxa such as races. And using resemblance classes to form racial classification is very sensitive to the identification keys used.

OK, that's the nitpick, although I find this stuff more fascinating than the usual questions raised about race. That's just me.

The second question is

2. The conceptual and social role of "race"

Racial classifications are almost always associated with measures of valuation. that is, we talk often about "advanced" or "civilised" races versus "primitive" or "barbaric" races. I know this is not what was meant by Razib or Matt, but the problem is that simply saying that self-identified races match genetic clustering implies a reality, not to the biological variation, which is real, but to the underlying semantic connotations of the term. And if the "races" used in the clustering really do match biological distributions, it implies, and is regularly taken to imply by those who use race in political and social debates, that the valuations are worth using too.

This is an argument from consequences, and might be a fallacy. If races are real, then the social consequences don't make a damn bit of difference to the factuality. But suppose they are not real, but appear to be because of the way we have investigated it, that is, the way we set up the questionnaire. As I said before, you can find covariance between any two sets of categorials if you care to. The impact of this is dramatic. The nuances of the scientific papers will get lost in the ways the science is employed. This is not trivial, and we need to be very careful here. The human tendency is universal to identify and discriminate against outgroups. Scientists no less than anyone else need to ensure that they don't give aid and comfort to the bigots. So I merely say, be absolutely sure before making claims of the reality of [socially constructed] races.

We humans are pretty good pattern matchers. Like a neural network (because we use them), we can take multivariate inputs and cluster them. But this is not reliable all the time. We can be trained on the wrong data set, for instance. What I am claiming here is that the clusterings are informed by "wrong" data sets to begin with. Rather than trying to match between naive categories and sophisticated and real biological data, perhaps it might be better to see what sorts of clusterings occur just on the basis of the data.

Suppose we do this, and someone will argue that this is what the Rosenberg et al. paper does do, if there are clusters, does this mean all those clusters are equally of the same "depth"? Is, for instance, the Oceanic group of the same rank or significance as the African? How could that be? We have been in Africa for many times the amount of time we were in Asia or Oceania - the amount of unique allelic variation that could evolve in situ must be less than that in Africa. So equating the two clusters is already to make a mis-step.

There's a problem of commensuration here that also occurs in cladistics. There are no ranks in evolution apart from those we impose for convenience. It's very important then to ensure that the convenience suits scientific purposes rather than social ones.

3. Medical issues

One of the main reasons that people today will assert the reality of race is based on the genetic components of things like malarial resistance, lactose intolerance, resistance to various diseases like Yersinia pestis (bubonic plague), and AIDS. There clearly are geographical variations here. It is sometimes convenient to do a Dr House and say that a given disease is identifiable on racial classification. As a first cut, this might work as an inferential guideline, as a heuristic rule. The problem with such abstractions, or as Razib calls them abstracta, is that they are only so good as the distributions of properties they cover. What than gnomic utterance means is that if you specify a type, and not all members of the type have the "typical" traits, then you can be misled as well as informed. Surely it might be better to see what sorts of phenotypic traits covary with the diseases first, before assigning those genes to a phenotypic class.

But Razib has conceded that it doesn't matter from a medical perspective whether we call these races or populations with structure. It does matter than one is likely to suffer from lactose intolerance if one has a substantial east Asian heritage. So let us leave this. I want to play with the philosophical aspects Razib raises about this:

4. Metaphysics

Ah, metaphysics. Meat and drink to a philosopher. Usually not to philosophers of science, though, but I'm atypical (that is, you cannot infer inductively from a knowledge of the type I instantiate to all of my properties <wink>).

Razib says that
we are clashing in the turbulent waters of nominalism
and
the perception by John that I am conceiving of race as an essential and fundamental taxonomical unit. I don't hold to that. I've rejected the Platonic conception of race before.
And so it is. Ultimately this is all about what one thinks classes of things, universals, are. Now while a lot of attack on race concepts focus on "Platonism", I don't, and I don't think Razib is using such a perspective, either. There is a very large difference between saying that something has an essence shared by all its members, to saying that the something is a Platonic form. The former claim is Aristotelian, not Platonic. The difference is this, and bear with me because it has a payoff in this debate:

Plato's forms are not physical things. They are eternal kinds, that no actual object ever properly instantiates, but at best approximates. Aristotle's forms are always physical things, and the descriptions of those things are what has the essences, not the things themselves. This gets a bit messy, because then we have to talk about "properties", but that will do for now. Nominalism, a later development ironically dealing with the confusion between Platonic and Aristotelian forms by the so-called Neoplatonists, holds that universal properties, of essences, exist solely in the words and in the head. This is exactly the issue about abstraction that I dealt with before.

It is my view that we are born Aristotelian essentialists [see
The Essential Child : Origins of Essentialism in Everyday Thought by Susan A. Gelman] but that we learn in our investigation of the world, that it doesn't rely on the nature of the words we use to describe or explain it.

Now, when we classify our world we are often tempted to reify the groups we class. "Races" are such an item. If there are biological clusters, there are biological clusters, but to call one of these a "race" is to invite belief that there is some shared description that is true of all and only members of that "race", which Razib doesn't believe. So we should be nominalists about the abstracta because it hinders our natural tendency to make fallacious inferences from the labels we use.

I better summarise, because otherwise I'll lose more of you than I already have.

  • Race refers to well-marked and stable subspecific varieties in a species. Humans do not have these, I believe
  • They are abstractions based on our tendencies to recognise variation in a typological manner, informed by our social discriminations
  • It is better to generate our abstractions directly from the data to inform our inferential processes, than to claim that prior naive classifications are somehow real. That is, the data explain why we make these discriminations, but they don't license them.
  • The social problems of using the naive folk taxonomy of race are so great that we should be careful about using the notion at all.
I think that is all I have worth saying on the matter. Read the comments - they have been very useful in this series of posts.

Late note: RPM at Evolgen has a nice wrap up, although I wonder did I really start race riots? I hope not.

More on the changes to the ESA

While I prepare a response to the responses... etc., for the race thread, go read this article in the Pinnacle News about changes to the Endangered Species Act, driven by a Republican congressman, Pombo, with respect to critical habitats (in this case, for the red-legged frog in California).

Sunday, April 23, 2006

Some more on race

Razib at Gene Expression has responded to my comments, and the original author Mike McIntosh, has updated his original piece with a few comments in response also. Good for both of them - I am not an expert in this matter, and I need the education. But first, let me deal with a few misunderstandings of what I said brought about by my execrable abuse of the English language (hey, I'm an Australian, and a philosopher. It's a wonder I have any readers at all).

Razib says of my comment:
Within-species groupings are, evolutionarily speaking, ephemeral. Ten thousand years ago, almost none of the non-African races existed. Ten thousand years from now, almost none of the modern races will continue to exist, I warrant.
that
First, I don't agree that "ten thousand years ago" there weren't any non-African races. Europe was settled for about 40,000 years by anatomically modern humans, Australia for 50,000 years, west and south Asia for somewhat longer, and the New World had just been settled. There has been a lot of time for human variation to evolve.
I was unclear. By this I meant that the categories we now call "races" for non-African populations - Caucasian, Asian, Austronesian, Amerind, and so forth, did not apply to populations outside Africa ten thousand years ago. Of course there were humans of variant kinds outside Africa. But take a case I know a little - the Australian aborigines. Work done by Alan Thorne on the Mungo Man skeleton in Australia and other work suggests that there were three or more migrations into the continent, one of which was more robust than the others. These date back at least 45kya. Given the variation in morphology of these migration waves, it looks pretty clear that "Australian" is a temporary class. More recent introgressive mating, from Javanese fishermen over at least a thousand years, and the rest of the ethnic world in the past 200 years, will leave the class even more changed in ten thousand years. I am claiming that the racial categories of modern use are ephemeral, not that there was no variation outside Africa.

Next, he said of my comment
The human species (convention makes me want to type "human race") is massively interbreeding.
that
How do you define massively? In Brazil peoples from various geographical "races" (just replace with "subpopulation" or something if you prefer) admix. In China this does not occur. Of course alleles spread, but they spread via selection as much as neutral processes, and that selection is often contingent on local conditions.

I define "massively" here as being continuous between populations on an evolutionary time scale. There are few (I know only of one. the Tasmanians) populations that are [were] not interconnected with others by gene flow. Perhaps there may be some in Papua or the Amazon. Even the !Mbuti, I am told, occasionally interbreed outside their group. In other species, the record is wildly variable, ranging from constant interbreeding and introgression between species, to almost zero if not zero interbreeding between populations of the same species isolated for millions of years. But humans have to be at the upper end of that scale.

There will be structure, of course But that structure will be limited to a local time scale, as they will be large scale tokogenetic networks similar to individual tokogeny. To be races, I would think that you'd need at the least to have long-term cladogenesis of the type found in haplotype groups of geographically isolated populations of other species. And I do not think that the Han group in China is free of interbreeding. Nor even is the caste structure in India so isolated from a biological perspective. Assortative matings don't make a race.

So, what does? It's a good question, and a hard one to answer. Razib notes that this is semantic; partly it is, partly not. There is a biological case as well. Typically, "race" is a well-defined subspecies in traditional taxonomy. We know there are subgroups in humans, so the question is whether there are sufficiently defined subgroups in humans. It's an empirical question. It's not, though, about alleles as such, but ensembles of alleles. There has to be a large number of alleles that are differentiated into clusters for there to be races. And I think the evidence is that there is not, at least, not that matches anything we call in social use "race". On the other hand I can often identify an African as being from the Afar, Yoruba or Maasai groups. The San are particularly identifiable. Perhaps they should be regarded as "races".

Razib says that there are functional allellic matching problems. Sure there are. This is what the whole research program is about. Is it race? Is it a "hillist-mountainist" problem as he says? Well yes. That's the problem with classification of evolutionary groups. One well-marked division is at the species level; that's why most of the time (exceptions of course, this is evolution) species are consensually accepted by biologists. And I fully concur with him that any taxonomic level is of the same "kind" as any other (see Nelson 1989), for all taxa are just lineages grouped by ancestry and division, and species are the same as any other clade, only with no permanent cladogenesis internally. Races, though, are not permanent (if they were, they'd be species in their own right).

So, given that there are genetic and phenotypic substructures, the only really significant dispute between us is whether the "folk taxonomy" categories match up to any biological reality. I think I'll stick with my claim that there isn't. There are clearly populations with differing clusters of features at the genetic and morphological level, and many of these are clearly the result of selection for local conditions. But would a Martian anthropologist identify our "races" as real races? I think not.

Matt's response is interesting. He makes the point that race is an abstractum, a point I made recently myself about other biological categories. And there is nothing wrong about abstracta - they are in fact necessary to scientific work. An abstraction can be a good representation of underlying causal realities. I merely disagree whether race is such an abstraction. By this I stress that I mean the usual categories used in the self-identification study. If you want to agree that the 54 categories of the Rosenberg study are races, then I have no objection. At least that gives us an empirical basis for cross-comparison. Eppur si muove indeed. But what moves here are not the races of the initial study.

Thanks to both Matt and Razib for their criticism.

Nelson, Gareth J. (1989), "Species and taxa: speciation and evolution", in D Otte and J Endler (eds.), Speciation and its consequences, Sunderland, Mass.: Sinauer.