SCIENCE AND ETHNOSCIENCE
E. N. Anderson
Dept. of Anthropology, University of California, Riverside
Part 3. Case Study: Classification
One fact that is devastating to the view that science is purely a cultural or social construction is the broad consonance between folk and scientific systems of classification. People everywhere classify plants and animals about the same way, recognizing categories like “bird,” “snake,” and so on (Atran 1990; Berlin 1992; Brown 1984). Moreover, they focus on inferred biological relationships. They classify dogs with dogs, cats with cats, and oak trees with beech trees, rather than—say—shepherd dogs with sheep and human shepherds, hounds with ducks, cats with grass, and oak trees with potatoes.
People classify things. The fundamental, original purpose of this is to make the world manageable. If we had to react to every stimulus as a new and unprecedented thing, we would never get out of bed in the morning. Thus, as Kant (1978) pointed out, we assimilate and differentiate as we need to. Humans seem to be natural classifiers. Modern psychology confirms Kant’s points: we essentialize categories, treating things we class together as if they were “the same” and exaggerating the differences of things we put in different categories (Atran 1990; Atran and Medin 2008).
Classifications are fundamentally about being useful. We classify so that we can identify edible and useful plants, dangerous or poisonous animals, types of tools we need for projects, breeds of dogs used for different tasks, types of paintings (Impressionist, abstract expressionist, op-art…). Most classifications are developed from actual interaction with the things we are classifying. We classify them in ways that make for maximal efficiency in using them.
However, our love of classifying runs far beyond utility. We classify all manner of things, and learn about them. Folk biology everywhere includes an incredible number of names and facts, many of them essentially useless to the people who know them. The Maya, for instance, have names for all manner of tiny insignificant birds, and know their life histories.
We classify everything: dogs, personality types, ideas, gods, kinfolk, potatoes (some Peruvian farmers know hundreds of varieties by sight), and kinds of love. People even develop classifications for fun, like the classifications of imaginary creatures (unicorns, dragons, and so on) in fantasy books.
Conversely, people may develop classifications to keep people in line—from those endless classifications of sins and impurities in the Bible to those endless classifications of traffic violations in the modern civil codes. So one main, and universal, use of classification systems is to maintain control not only over natural complexity but also over people’s lives and social actions (Bowker and Star 1999; Foucault 1970).
Classifications range from legally defined, like types of property, to biologically based, like types of fir trees. Classifications may be very clear, simple, and sharp, like classifications of living elephants: there are only two, not much like each other, and not at all like any other animal. Conversely, classifications of philosophic theories are so endlessly argued by philosophers that one may wonder whether the reason for having the classifications at all is to stir up debate. Obviously, we will never have an accepted classification of philosophies.
Classifications may be universal (modern scientific nomenclature—among scientists, at least), cultural (English bird names), or at lower levels. My classification of foods I like is unique to me. My children’s classification of foods they would not eat was an all too significant family reality in their early years, but is of no significance today.
Classifications may be broadly true in some sense. The modern scientific classifications of chemicals, stars, and living things are grounded in real and demonstrable facts. We classify animals and plants on the basis of biological relationships. Linnaeus had to infer these from appearance, and brilliantly saw that flowers are basic rather than leaves, stems, and roots. We can now use cladistic analysis backed up by comprehensive genetics, and prove directly the genetic relationships we once had to infer.
On the other hand, classifications can be ad hoc, or plain wrong, or utterly ridiculous, like José Luis Borges’ “Chinese encyclopedia” parody cited in Foucault (1970:xv).
The purpose of classification is not to be right but to be useful. Even Borges’ is useful: it is intended to shock the reader into thinking about the whole philosophical issue of classification. Anyone reading it realizes it is a joke, because the units are totally non-comparable; a real system has to have units that are comparable, in ways that matter within that particular system. In technical terms, there has to be an “emics” to the system.
A great deal of research has been devoted to the history and cross-cultural variation of classification.
Kinship terminology has received by far the most effort. Kinship is unique in that it is equally important and elaborate in all cultures. It is the only realm in which every culture has an elaborate, precise, formalized, and almost universally known system. Australian aboriginals may not have elaborate physics or chemistry, but their kinship systems are so formal and elaborate that many brilliant English-speaking scholars have spent years unsuccessfully trying to analyze them. Thus they provide ideal material for comparative analyses of human thought. Therefore, theorizing about family and kin has always been basic to anthropology. Lewis Henry Morgan’s vast classic work Systems of Consanguinity and Affinity (1871) put the seal on kin classification as a major field for anthropological endeavor (Trautman 1987).
Since the 1870s, much work has been devoted to taxonomies of animals and plants, and sometimes other living things (like fungi). Comparative work has shown that people everywhere see real biological relationships, and use them as one basis for classifying. This has even led authorities on the subject to postulate that people have a natural tendency to classify on the basis of perceived basic similarities (Atran 1990; Berlin 1992; Brown 1984).
In classifying living things, every culture has a general classification systems versus special purpose classifications. Brent Berlin found that this distinction is basic and apparently worldwide (Berlin 1992). The general system is the one based on apparent biological relationships or real-world appearance. It is the one that provides everyday names. Everywhere, it is based on inferred similarities out there in nature. Everywhere, if you simply ask “what is that?” you get the name in the general system. The name of an animal or plant is always understood to be its name in the general system unless you specify otherwise. Local utilitarian factors influence all general purpose systems (Hunn 1982), but do not determine them, so they end by looking very much like the modern international scientific system.
On the other hand, Roy Ellen has long emphasized real differences between cultural classification systems (e.g. Ellen 1993). Similarly, Geoffrey Lloyd (2007) has found many areas that are not accurately perceived in folk biology. He makes much of the microorganisms, which are irrelevant to his case, but he makes the more serious point that traditional classifications generally fail at the higher levels. Nobody seems to have words for “mammal,” and many cultures lack a word for “animal.” Plants are variously assembled. Carol Kaesuk Yoon (2009) has recently held that there is a “clash” between “instinct”—the natural categorization that humans do—and “science.” She maintains this because genetics has now shown that fish fall into several classes, with the bony fish closer to humans than to cartilaginous fish. One might add that birds are closer to some “reptiles” (dinosaurs) than those are to other reptiles.
So science is indeed a cultural construction. Even modern classifications are not 1:1 maps of biology, and folk systems certainly are not.
Yet, in fact, folk and traditional people see natural categories astonishingly well—not as well as the best modern geneticists, but well enough to show that nature is hard to ignore in these matters. Sometimes the traditional small-scale societies had views closer to modern genetics than the Linnaean biologists did. Fungi were still “plants” when I was an undergraduate, but the indigenous peoples of Mexico correctly place them closer to animals (Hunn 2008; Lampman 2008). I found that the Yucatec Maya categorize orioles according to the best modern analysis. And certainly my friends on the Hong Kong waterfront were aware that “fish” (yu) was a functional class (swimming aquatic life), not a natural biological one. They knew from inspection, for instance, that cuttlefish were closer to octopi than to bony fish, though cuttlefish were “fish” and octopi were not. It is significant that this example translates perfectly; folk English does the same thing.
The point is that it is constructed on the basis of ongoing interaction with reality. (Even those unicorns are based on reality, at a couple of removes. The original “unicorn” was the rhinoceros, and tales of it—a huge horselike creature with one horn in the middle of its forehead—were duly interpreted as reasonably as possible: a horse with a narwhal tusk for a horn, the Europeans having no other one-“horned” animal to compare.) Obviously, no society could exist if it did not base its knowledge on truth learned by experience.
One proof is the development of dictionaries in Arab (Carter 1990) and Chinese civilizations. Technical vocabularies specialized on particular subjects, such as horses (Carter 1990) or drugs and medicines, show the classification systems appropriate to those matters. Early Arabic dictionaries sometimes arranged words by linguistic domains, and these were much like ours or anyone else’s. The early Greek and Latin writers also classified plants and animals in ways not irrational or incomprehensible. They are not the same as our ways, but they are close enough that we still use many of Theophrastus’ and Pliny’s names as scientific names, either for the same plants or for similar or related ones. (Still, one sometimes wonders about the more modern sages! Kaktos, Greek for a kind of thistle, wound up applied to some plants that have nothing in common with thistles except prickliness. Dozens of other names were similarly applied any old way, just to recycle a Greek name, no matter how inappropriately. This started early; kardamon, another thistle, had already—and mysteriously—become the name of a spice in late antiquity.)
Special purpose classifications classify plants and animals in relation to human wants and needs. In Hong Kong, when I asked “what is that fish?” I got the name in the general system, relating fish to fish—classifying them as soles, sharks, groupers, and so on. I slowly learned there were many other ways to classify fish: by price, by technique used to catch them, by habitat, by sacred and ceremonial significance, and by eating qualities. These were five separate, salient, well-known systems. They were not merely ad hoc. The fishermen never confused them with the basic system (Anderson 1972). When I asked “what is that fish?” I always got a name from the basic system, never “a netted fish” or the like.
Proof that even the arcana of fish classification can suddenly become important is found in the striking book Trying Leviathan by D. Graham Burnett (2007). This book is the history of a trial that took place in New York City in 1818 to decide whether a whale was a fish or a mammal. The state had passed a law requiring inspection of fish oil, with a fee to be paid by the seller. This being New York, a whale-oil dealer immediately challenged the law on the basis of science: whales had recently been classified as mammals by Linnaeus and Cuvier. This early example of New York chutzpah got him haled into court. The trial involved the formidably brilliant icthyologist Samuel Mitchill as witness for the defense, but the verdict went against the dealer, since the plaintiff could establish that the state legislature had passed the law based (at some remove) on the supposition that whales were fish and whale oil would be inspected.
This was long before Darwin. There was no obvious reason to prefer lactation, air-breathing and live birth over fins, aquatic habitat, and streamlined shape as classification markers. The lawyers were astute enough to realize that classification could be ambiguous; they made reference to the “duck-billed beaver” (platypus) and other anomalies. Not only New York lawyers find whales confusing. My fishermen friends in Hong Kong told me that whales and porpoises were anomalous because they looked like fish but acted intelligent (unlike fish) and were “like pigs” internally. My friends thought these creatures were uncanny, and avoided catching them.
In anthropology, studies of classifying everything from religious ceremonies to art objects have continued to proliferate. An unwise attack on such studies was launched in the 1960s by Marvin Harris (1966, 1968) and others. Harris chose to criticize a study of Maya firewood knowledge by Duane Metzger and Gerald Williams (1966), branding it—and by extension all such research—as “trivial.” He could not have picked a worse target. The Maya depend on firewood for cooking and warmth. They live in a wet climate where good dry wood is hard to find and must be carefully chosen. Like hundreds of millions of other people around the world, they spend up to several hours a day searching for firewood. Knowing how to get the best wood in the shortest time is a life-and-death matter for them. Firewood use is a matter of enormous councern worldwide, since about 1/3 of all the wood used in the world goes for this purpose, greatly contributing to global warming and deforestation. Nothing could be less trivial, either to the Maya or to the planet.
An area in which folk classification is infamously important, inaccurate, and pernicious is “race.” Americans are addicted to the notion that everything important is genetic and that genetics is a simple science. (Many have wondered how a nation of overachieving immigrants from all manner of other cultures can believe this.) Thus, as noted above, Richard Herrnstein and Charles Murray in The Bell Curve (1994) give us a “Latino race” with an IQ of 89! Quite apart from the absurdity of such aggregated measures of intelligence, Herrnstein and Murray simply ignored the fact that Latinos can be white, black, Native American, East Asian, or any and all mixtures of these. Similar confusion surrounds “Black” Americans, Native Americans, and other categories. We have lately been inflicted with something called “race medicine,” which prescribes different drugs for Black and White Americans. Yet there is a total continuum. Millions of Whites are part Black, and almost all Blacks are part White—frequently 15/16, since anyone with any African appearance is called “Black.” These 15/16 Caucasian patients are given “Black” drugs! Even such appalling bureaucratic monstrosities as “Asian-Pacific Islander”—the creation of arbitrary Census Bureau labeling—have become “real” to Americans. This shows how “race” classifications can not only change arbitrarily but can be invented out of whole cloth. The strange, if not downright surrealistic, history of “race” labels has been well covered in anthropology by Lee Baker (1998), Audrey Smedley (2007), and Jonathan Marks (Marks 2001), among others.
Even Linnaean classification is related to economic and aesthetic theories of the Enlightenment elite (Foucault 1970). Foucault also saw many other interesting aspects of classification that go far beyond its immediate utility. He wrote: “Take…animal and plant classifications. How often have they not been rewritten since the Middle Ages according to completely different rules: by symbolism [the medieval use of animal and plant symbols], by natural history, by comparative anatomy, by the theory of evolution. Each time this rewriting makes the knowledge completely different in its functions, in its economy, in its internal relations” (Foucault, in Chomsky and Foucault 2006:26; cf. Foucault 1970). There is some truth in this, but Foucault misses the key point that actual everyday classification of creatures did not change significantly during this period. Dogs were dogs, cats were cats, whales were whales. Nor, of course, was it “completely different in its functions”; it still functioned largely to let people name what they saw, and give similar names to similar creatures.
Over centuries, many new plants and animals were added to European knowledge, necessitating major changes in everyday words and usages, but the basic system did not change. However, Foucault is correct in that elite scholars’ interests and perspectives really did change. The Medieval churchmen were more interested in animals as symbols than in animals as animals (see e.g. Herbert Friedman’s superb account of birds in art, 1980; also Rowland 1978). The function vs. anatomy tension lies behind the whale trial described above, and did indeed affect how we folk speakers classify whales, but we still talk about the “whale fishery,” as well as “shellfish” and “cuttlefish” and other non-anatomical “fish.” Darwin profoundly changed human thought, but not folk taxonomical usage.
Consider the European goldfinch (Carduelis carduelis). In the Middle Ages it was a symbol of Christ, and thus the child Christ is shown holding one in many Renaissance paintings. In Linnaeus’ taxonomy it got its present name—just its old Latin name, doubled—and was classed with finches. Anatomists then separated the finches into several groups—they turned out to be more different inside than outside—and the goldfinch got its own family, Carduelidae (which includes a lot of its relatives). Darwinians have gone on to debate the actual relationships and membership of the Carduelidae. So Foucault is right.
But not right at the deepest level. Throughout all of this, the goldfinch remains a goldfinch, and every English speaker who notices birds knows it. Germans similarly call it a distelfink, “thistle finch,” as they have for centuries, in honor of its regular food (thistle seeds). Folk classification still makes it a “finch” along with zebra finches and Mexican ground finches, although we now know these birds are not closely related.
This emphasizes a difference between folk and elite understandings. This is not so much a matter of better or worse education, or of snobbism, but of needs. We ordinary people, and this includes scholars and scientists on their off days, need to have a quick, convenient, pragmatic label to refer to things we regularly interact with. Scientists need to have labels based on understanding of deeper, less obvious, but more biologically important processes.
Hence scientists refer to C. carduelis in the lab and in the technical literature, but call it a “goldfinch” when they see it in their thistle patch. Anywhere in the world, if you ask “what’s that?” as a small yellow finch with a red face flies by, you’ll be told “a goldfinch” (or local equivalent). You will never be told “that’s the Christ child,” and you would not have been told that in the Italian Renaissance, either. The medieval symbolic system is very much a special purpose classification, and the medieval artists knew that.
Medieval and Renaissance artists and writers often spoke of four levels of symbolism—traditionally defined as “image, symbol, metaphor, and allegory” or something similar (see e.g. Schneider 1992:17). This survived in religious music until the present. A good, and thoroughly modern, example is Mississippi John Hurt’s “Slidin’ Delta Blues,” in which the image–a train nicknamed “Slidin’ Delta”—is a symbol of parting from one’s love, which is a metaphor for death, which is an allegory for transcendence. Hurt sings: “Lord, I’m goin’ somewhere / I never been before,” where the “somewhere” is a faraway real place, death, mystical experience, and Heaven, depending on the level at which one is listening.
“Totemism,” in the broad sense, is similar. Classifying people into Eaglehawk and Crow moieties or into Wildcat and Coyote moieties does not mean that people are animals. It is not the basic classification of the traditional peoples that use this system, either, contra Emile Durkheim and Marcel Mauss (1963). It is simply the use of well-known animals as symbols for social groups (Lévi-Strauss 1962, 1963). This is a special purpose system, and it depends on the prior existence and widespread knowledge of the basic or general system. It often depends also on knowing the animals’ habits. Wildcats were associated in California Native cultures with valleys, coyotes with mountains, and thus valley and lowland animals are in the Wildcat moiety, hill and mountain ones in the Coyote moiety. People are distributed according to birth rather than residence, however. One is automatically in one’s father’s moiety, no matter where one lives. What has happened is that human social divisions are projected onto nature. “Nature” and human society are not radically separated in Native American cultures, so this is a “natural” thing to do (Durkheim and Mauss 1963).
Such social classifications are universal; consider our school mascots. Symbol, metaphor, and allegory are amazingly important to humans (Lakoff and Johnson 1980).
An odd kind of “classification” is found in linguistic gender and other grammatical systems. German, Spanish, and other gender systems are notoriously decoupled from sexual reality; “maiden” is neuter in German. Several Australian languages have four genders: masculine, feminine, neuter, and useful plants (Lakoff 1990 discusses one such language). Many languages, including Chinese and Maya, have classifying particles, added to numbers and demonstratives, that identify broadly the type of noun to follow. Thus Chinese says yi ben shu “one volume book” and yi tiao yu “one length fish.” This allows one to see that Yucatec Maya has a category for “plants” in general: there is no actual word for “plants,” but there is a classifier (k’ul) that includes all and only plants.
Maps are, in a sense, another form of classification. Not all cultures make maps, but all have extremely detailed knowledge of places and paths in their environments (Hunn 1991, 2008). The idea that small-scale societies are somehow intuitively aware of the environment without making mental maps or representations of it is wrong (Istomin and Dwyer 2009).
Yet, humans seem compelled to think causally. This is another inborn habit of thought. We have to find a motive. Typically, we first look for an active, thinking agent. If that fails, we look for a covering law—not usually a formal one, just a rule of thumb that will serve. Only if that too totally fails do we accept blind chance, or probabilistic factors, as a reason (see e..g Nisbett and Ross 1980). As Geoffrey Lloyd says, “humans everywhere will will use their imaginations to try to get to grips with what happens and why, exploiting some real or supposed analogy with the schemata that work in otherwise more mundane situations” (Lloyd 2007:130).
Aristotle described four types of cause, or rather of aition (pl. aitia), which has also been translated “factor” (Aristotle 1952:9, 88f). The first is material cause—what the object we are contemplating is made of. This would not occur to modern people as a “cause”—the hickory wood does not cause the baseball bat—but Aristotle was thinking partly of the elements of Greek thought. Earth, air, fire, and water were generally thought to have dynamic qualities that made them evolve into things. Chlorine purifies water by virtue of its violently oxidizing nature, which destroys bacteria and toxins; this is an example of material cause in action.
Second is formal cause: the definition of the object, its pattern, its essential character. A baseball bat is a rounded stick made of hickory wood, and is patterned so as to hit balls in a game. Third is efficient cause—the direct, proximal cause, specifically the causing agent, of an action. The bat is made by a factory to be sold to a player, who then uses it to hit a ball; the chlorine is bubbled through water, where it reacts chemically with toxins and bacterial membranes. Fourth is the final or ultimate cause, the reason for the action or object: the water is purified so people can drink it safely; the bat is used in a game for the purpose of entertaining people. This last can go into infinite regress: the bat is to hit a ball, so that the game will go on, so that people will be entertained, so that they will enjoy life and buy the sponsors’ products, so that…. And this only scratches the surface of Aristotle’s theory of cause, and he was only one Greek philosopher (see Lloyd 2007:108-130).
The endless debate on cause in philosophy since Aristotle need not concern us, since we are here considering folk and traditional knowledge. In that realm, our heuristics and biases play out at their most florid. Aristotle’s efficient cause is stated in agent terms. This default attribution to intentional action by an agent gives us the universal belief in gods, spirits, and other supernatural beings.