[Paleopsych] BBS: Folk biology and the anthropology of science

[Premise Checker]

Folk biology and the anthropology of science:
Cognitive Universals and Cultural Particulars
http://www.bbsonline.org/documents/a/00/00/04/23/bbs00000423-00/bbs.atran.html

  Scott Atran
  Centre National de la Recherche Scientifique (CREA - Ecole
  Polytechnique)
  1 rue Descartes
  75005 Paris
  FRANCE
  and
  Institute for Social Research
  The University of Michigan
  Ann Arbor MI48106-1248
  USA
  satran at umich.edu

Keywords

Folk biology, taxonomy, cognitive universals, modularity, evolution,
culture, Maya, anthropology

Abstract

This essay in the "anthropology of science" is about how cognition
constrains culture in producing science. The example is folk biology,
whose cultural recurrence issues from the very same domain-specific
cognitive universals that provide the historical backbone of
systematic biology. Humans everywhere think about plants and animals
in highly structured ways. People have similar folk-biological
taxonomies composed of essence-based species-like groups and the
ranking of species into lower- and higher-order groups. Such
taxonomies are not as arbitrary in structure and content, nor as
variable across cultures, as the assembly of entities into
cosmologies, materials or social groups. These structures are routine
products of our "habits of mind," which may be in part naturally
selected to grasp relevant and recurrent "habits of the world." An
experiment illustrates that the same taxonomic rank is preferred for
making biological inferences in two diverse populations: Lowland Maya
and Midwest Americans. These findings cannot be explained by
domain-general models of similarity because such models cannot account
for why both cultures prefer species-like groups, despite the fact
that Americans have relatively little actual knowledge or experience
at this level. This supports a modular view of folk biology as a core
domain of human knowledge and as a special player, or "core meme," in
the selection processes by which cultures evolve. Structural aspects
of folk taxonomy provide people in different cultures with the
built-in constraints and flexibility that allow them to understand and
respond appropriately to different cultural and ecological settings.
Another set of reasoning experiments shows that the Maya, American
folk and scientists use similarly structured taxonomies in somewhat
different ways to extend their understanding of the world in the face
of uncertainty. Although folk and scientific taxonomies diverge
historically, they continue to interact. The theory of evolution may
ultimately dispense with the core concepts of folk biology, including
species, taxonomy and teleology; in practice, however, these may
remain indispensable for scientific work. Moreover, theory-driven
scientific knowledge cannot simply replace folk knowledge in everyday
life. Folk-biological knowledge is not driven by implicit or inchoate
theories of the sort science aims to make more accurate and perfect.

INTRODUCTION [1]

In every human society, people think about plants and animals in the
same special ways. These special ways of thinking, which can be
described as "folk biology," are fundamentally different from the ways
humans ordinarily think about other things in the world, such as
stones, stars, tools or even people. The science of biology also
treats plants and animals as special kinds of objects, but applies
this treatment to humans as well. Folk biology, which is present in
all cultures, and the science of biology, whose origins are particular
to Western cultural tradition, have corresponding notions of living
kinds.

Consider four corresponding ways in which ordinary folk and biologists
think of plants and animals as special. First, people in all cultures
classify plants and animals into species-like groups that biologists
generally recognize as populations of interbreeding individuals
adapted to an ecological niche. We will call such groups - such as
redwood, rye, raccoon or robin - "generic species" for reasons that
will become evident. Generic species are usually as obvious to a
modern scientist as to local folk. Historically, the generic-species
concept provided a pretheoretical basis for scientific explanation of
the organic world in that different theories - including evolutionary
theory - have sought to account for the apparent constancy of "common
species" and for the organic processes that center on them (Wallace
1889/1901:1)

Second, there is a commonsense assumption that each generic species
has an underlying causal nature, or essence, which is uniquely
responsible for the typical appearance, behavior and ecological
preferences of the kind. People in diverse cultures consider this
essence responsible for the organism's identity as a complex,
self-preserving entity governed by dynamic internal processes that are
lawful even when hidden. This hidden essence maintains the organism's
integrity even as it causes the organism to grow, change form and
reproduce. For example, a tadpole and frog are in a crucial sense the
same animal although they look and behave very differently, and live
in different places. Western philosophers, such as Aristotle and
Locke, attempted to translate this commonsense notion of essence into
some sort of metaphysical reality, but evolutionary biologists reject
the notion of essence as such. Nevertheless, biologists have
traditionally interpreted this conservation of identity under change
as due to the fact that organisms have separate genotypes and
phenotypes.

Third, in addition to the spontaneous division of local flora and
fauna into essence-based species, such groups have "from the remotest
period in... history... been classed in groups under groups. This
classification [of generic species into higher- and lower-order
groups] is not arbitrary like the grouping of stars in constellations"
(Darwin 1872/1883:363).[2] The structure of these hierarchically
included groups, such as white oak/oak/tree or mountain
robin/robin/bird, is referred to as "folk-biological taxonomy."
Especially in the case of animals, these nonoverlapping taxonomic
structures can often be scientifically interpreted in terms of
speciation (that is, related species descended from a common ancestor
by splitting off from a lineage).

Fourth, such taxonomies not only organize and summarize biological
information; they also provide a powerful inductive framework for
making systematic inferences about the likely distribution of organic
and ecological properties among organisms. For example, given the
presence of a disease in robins one is "automatically" justified in
thinking that the disease is more likely to present among other bird
species than among nonbird species. In scientific taxonomy, which
belongs to the branch of biology known as systematics, this strategy
receives its strongest expression in "the fundamental principle of
systematic induction" (Warburton 1967, Bock 1973). On this principle,
given a property found among members of any two species, the best
initial hypothesis is that the property is also present among all
species that are included in the smallest higher-order taxon
containing the original pair of species. For example, finding that the
bacteria E-scheriehia coli share a hitherto unknown property with
robins, a biologist would be justified in testing the hypothesis that
all organisms share the property. This is because E. coli link up with
robins only at the highest level of taxonomy, which includes all
organisms.

As we shall see, these four corresponding notions issue from a
specific cognitive structure, which may be a faculty of the human mind
that is innately and uniquely attuned to perceiving and conceptually
organizing living kinds. The evolutionary origins of such a faculty
arguably involved selection pressures bearing on immediate utility,
such as obtaining food and surviving predators and toxins. In no
society, however, do people exclusively classify plants and animals
because they are useful or harmful. This claim goes against the
generally received view that folk biologies are primarily utilitarian,
and that scientific biology emerged in part to expel this utilitarian
bias from systematic thinking about the living world. Rather, the
special ways people classify organic nature enable them to
systematically relate fairly well-delimited groups of plants and
animals to one another in indefinitely many ways, and to make
reasonable predictions about how biological properties are distributed
among these groups, regardless of whether or not those properties are
noxious or beneficial.

Although folk biology and the science of biology share a psychological
structure, they apply somewhat different criteria of relevance in
constructing and interpreting notions of species, underlying causal
structure, taxonomy and taxonomy-based inference. Given the universal
character of folk biology, a plausible speculation is that it evolved
to provide a generalized framework for understanding and appropriately
responding to important and recurrent features in hominid ancestral
environments. By contrast, the science of biology has developed to
understand an organization of life in which humans play only an
incidental role no different from other species.Thus, although there
are striking similarities between folk taxonomies and scientific
taxonomies, we will also find that there are radical differences. To
explore how these different criteria of relevance function, the
folk-biological taxonomies of American students and Maya Indians are
compared and contrasted below with scientific taxonomies.

In this target article, we first describe universal aspects of folk
biology. We then show where and why folk biology and scientific
biology converge and diverge. In the final part, we explain how folk
biology and scientific biology continue to interact in the face of the
historical differences that have emerged between them. The focus is on
taxonomy and taxonomy-based inference. The general approach belongs to
"the anthropology of science," which this paper illustrates. The
examples of biology do not apply straightaway to all of science, any
more than those of systematics apply to all of biology, but they are
central enough in the history of science to be a good place to begin.

1.Folk-Biological Taxonomy.

Over a century of ethnobiological research has shown that even within
a single culture there may be several different sorts of
"special-purpose" folk-biological classifications that are organized
by particular interests for particular uses (e.g., beneficial versus
noxious, domestic versus wild, edible versus inedible, etc.). Only in
the last decades has intensive empirical and theoretical work revealed
a cross-culturally universal "general-purpose" taxonomy (Berlin,
Breedlove & Raven 1973) that supports systematic reasoning about
living kinds, and properties of living kinds, in the face of
uncertainty (Atran 1990). For example, learning that one cow is
susceptible to "mad cow" disease one might reasonably infer that all
cows are susceptible to the disease but not that all mammals or
animals are.

This "default" folk-biological taxonomy, which serves as an inductive
compendium of biological information, is composed of a fairly rigid
hierarchy of inclusive groups of organisms, or taxa. At each level of
the hierarchy, the taxa, which are mutually exclusive, partition the
locally perceived biota in a virtually exhaustive manner. Lay
taxonomy, it appears, is everywhere composed of a small number of
absolutely distinct hierarchical levels, or ranks. Anthropologist
Brent Berlin (1992) has established the standard terminology for
folk-biological ranks as follows: the "folk-kingdom" rank (e.g.,
animal, plant), the "life-form" rank (e.g., bug, fish, bird, mammal,
tree, herb/grass, bush), the "generic" or "generic-species" rank
(e.g., gnat, shark, robin, dog, oak, clover, holly), the
"folk-specific" rank (poodle, white oak) and the "folk-varietal" rank
(toy poodle; spotted white oak). Taxa of the same rank tend to display
similar linguistic, biological and psychological characteristics.

   1.1. The Significance of Rank.

Rank allows generalizations to be made across classes of taxa at any
given level. For example, the living members of a taxon at the
generic-species level generally share a set of biologically important
features that are functionally stable and interdependent
(homeostasis); members can generally interbreed with one another but
not with the living members of any other taxon at that level
(reproductive isolation). Taxa at the life-form level generally
exhibit the broadest fit (adaptive radiation) of morphology (e.g.,
skin covering) and behavior (e.g., locomotion) to habitat (e.g., air,
land, water). Taxa at the subordinate folk-specific and folk-varietal
levels often reflect systematic attempts to demarcate biological
boundaries through cultural preferences. .

The generalizations that hold across taxa of the same rank (i.e., a
class of taxa) thus differ in logical type from generalizations that
apply only to this or that taxon (i.e, a group of organisms). Termite,
pig and lemon tree are not related to one another by virtue of any
simple relation of class inclusion or connection to some common
hierarchical node, but by dint of their common rank - in this case the
level of generic species. Notice that a system of rank is not simply a
hierarchy, as some suggest (Rosch 1975, Premack 1995, Carey 1996).
Hierarchy, that is, a structure of inclusive classes, is common to
many cognitive domains, including the domain of artifacts. For
example, chair often falls under furniture but not vehicle, and car
falls under vehicle but not furniture. But there is no ranked system
of artifacts:[3] no inferential link, or inductive framework, spans
both chair and car, or furniture and vehicle, by dint of a common
rank, such as the artifact species or the artifact family. In other
words, in many domains there is hierarchy without rank, but only in
the domain of living kinds is there always rank.

Ranks and taxa are of a different logical order, and confounding them
is a category mistake. Biological ranks are second-order classes of
groups ( e.g., species, family, kingdom) whose elements are
first-order groups (e.g., lion, feline, animal). Ranks seem to vary
little, if at all, across cultures as a function of theories or belief
systems. In other words, ranks are universal but not the taxa they
contain. Ranks represent fundamentally different levels of reality,
not convenience. Consider:

The most general rank is the folk kingdom,[4] that is, plant or
animal. Such taxa are not always explicitly named but they represent
the most fundamental divisions of the biological world. These
divisions correspond to the notion of "ontological category" in
philosophy (Donnellan 1971) and psychology (Keil 1979). From an early
age humans cannot help but conceive of any object they see in the
world as either being or not being an animal, and there is evidence
for an early distinction between plants and nonliving things (Gelman &
Wellman 1991, Keil 1994, Hickling & Gelman 1995, Hatano & Inagaki
1996). Conceiving of an object as a plant or animal seems to carry
certain assumptions that are not applied to objects thought of as
belonging to other ontological categories, like person, substance or
artifact.

The next rank down is that of life form.[5] The majority of taxa of
lesser rank fall under one or another life form. Most life-form taxa
are named by lexically unanalyzable names (primary lexemes), and have
further named subdivisions, such as tree and bird. Biologically,
members of a single life-form taxon are diverse. Psychologically,
members of a life-form taxon share a small number of perceptual
diagnostics, such as stem aspect, skin covering and so forth (Brown
1984). Life-form taxa may represent general adaptations to broad sets
of ecological conditions, such as competition among single-stem plants
for sunlight and tetrapod adaptation to life in the air (Hunn 1982,
Atran 1985a). Classification by life form may occur relatively early
in childhood. For example, familiar kinds of quadrupeds (e.g., dogs
and horses) are classified separately from sea versus air animals
(Mandler, Bauer & McDonough 1991; Dougherty 1979 for American plants;
Stross 1973 for Maya).

The core of any folk taxonomy is rank of generic species, which
contains by far the most numerous taxa in any folk-biological system.
Taxa of this rank generally fall under some life form, but there may
be outliers that are unaffiliated with any major life-form taxon.[6]
This is often so for a plant or an animal of particular cultural
interest, such as maize for Maya (Berlin, Breedlove & Raven 1974) and
the cassowary for the Karam of New Guinea (Bulmer 1970). Like
life-form taxa, generic-species taxa are usually named by primary
lexemes, such as oak and robin. Occasionally, generic-species names
exhibit variant forms of what systematists refer to as binomial
nomenclature: for example, binomial compounds, such as hummingbird, or
binomial composites, such as oak tree. In both these cases the
binomial makes the hierarchical relation apparent between the generic
species and the life form.

Generic species often correspond to scientific genera or species, at
least for those organisms that humans most readily perceive, such as
large vertebrates and flowering plants. On occasion, generic species
correspond to local fragments of biological families (e.g., vulture),
orders (e.g., bat) and, especially with invertebrates, even
higher-order taxa (Atran 1987a, Berlin 1992). Generic species also
tend to be the categories most easily recognized, most commonly named
and most readily learned in small-scale societies (Stross 1973).

Generic species may be further divided at the folk-specific level.
Folk-specific taxa are usually labeled binomially, with secondary
lexemes. Such compound names make transparent the hierarchical
relation between a generic species and its subordinate taxa, like
white oak and mountain robin. However, folk-specific taxa that belong
to a generic species with a long tradition of high cultural salience
may be labeled with primary lexemes, like winesap (a kind of apple
tree) and tabby (a kind of cat). Partitioning into subordinate taxa
usually occurs as a set of two or more taxa that contrast lexically
along some readily perceptible dimension (color, size, etc.); however,
such contrast sets often involve cultural distinctions that language
and perception alone do not suffice to explain (Hunn 1982). An example
is the Itzaj Maya contrast between red mahogany (ch%k ch%k-al~te') and
white mahogany (s%k ch%k-al~te'). Red mahogany actually appears to be
no redder than white mahogany. Rather, red mahogany is preferred for
its beauty because it has a deeper grain than white mahogany. It is
"red" as opposed to "white" probably because Lowland Maya
traditionally associate red with the true wind of the East, which
brings rain and bounty, and white with the false wind of the North,
which brings deception (Atran in press).

In general, whether or not a generic species is further differentiated
depends on cultural importance. Occasionally, an important
folk-specific taxon will be further subdivided into contrasting
folk-varietal taxa, such as short-haired tabby and long-haired tabby.
Varietals are usually labeled trinomially, with tertiary lexemes that
make transparent their taxonomic relationship with superordinate
folk-specifics and generic species. An example is spotted white oak.

Foreign organisms introduced into a local environment are often
initially assimilated to generic species through folk-specific taxa.
For example, European colonists originally referred to New World maize
as "Indian corn," that is, a kind of wheat. Similarly, Maya initially
dubbed Old World wheat "Castillian maize." Over time, as the
introduced species acquired its own distinctive role in the local
environment, it would assume generic-species status and would, as with
most other generic species, be labeled by a single lexeme (e.g.,
"corn" in American English now refers exclusively to maize).

Finally, intermediate levels also exist between the generic-species
and life-form levels. Taxa at these levels usually have no explicit
name (e.g., rats + mice but no other rodents), although they sometimes
do (e.g., felines, palms). Such taxa - especially unnamed "covert"
ones - tend not to be as clearly delimited as generic species or life
forms; nor does any one intermediate level always constitute a fixed
taxonomic rank that partitions the local fauna and flora into a
mutually exclusive and virtually exhaustive set of broadly equivalent
taxa. Still, there is a psychologically evident preference for forming
intermediate taxa at a level roughly between the scientific family
(e.g., canine, weaver bird) and order (e.g., carnivore, passerine)
(Atran 1983, Berlin 1992).

   1.2. The Generic Species: Principal Focus of Biological Knowledge.

People in all cultures spontaneously partition the ontological
categories animal and plant into generic species in a virtually
exhaustive manner. "Virtually exhaustive" means that when an organism
is encountered that is not readily identifiable as belonging to a
named generic species, it is still expected to belong to one. The
organism is usually assimilated to one of the named taxa it resembles,
although at times it is assigned an "empty" generic-species slot
pending further scrutiny (e.g., "such-and-such a plant is some
[generic-species] kind of tree," see Berlin in press). This
partitioning of ontological categories seems to be part and parcel of
the categories themselves: no plant or animal can fail to belong
uniquely to a generic species.

The term "generic species" is used here, rather than "folk genera/folk
generic" (Berlin 1972) or "folk species/folk specieme" (Bulmer 1970),
for three reasons:[7] (1) a principled distinction between biological
genus and species is not pertinent to most people around the world.
For humans, the most phenomenally salient species (including most
species of large vertebrates, trees, and phylogenetically isolated
groups such as palms and cacti) belong to monospecific genera in any
given locale.[8] Closely related species of a polytypic genus are
often hard to distinguish locally, and no readily perceptible
morphological or ecological "gap" can be discerned between them (Diver
1940).

(2) The term "generic species" reflects a more accurate sense of the
correspondence between the most psychologically salient
folk-biological groups and the most historically salient scientific
groups (Stevens 1994). The distinction between genus and species did
not appear until the influx of newly discovered species from around
the world compelled European naturalists to sort and remember them
within a worldwide system of genera built around (mainly European)
species types (Atran 1987a).

(3) The term "generic species" reflects a dual character. As salient
mnemonic groups, they are akin to genera in being those groups most
readily apparent to the naked eye (Cain 1956). As salient causal
groups, they are akin to species in being the principal centers of
evolutionary processes responsible for biological diversity (Mayr
1969).

   1.2.1. The Evolutionary Sense of an Essence Concept.

From the standpoint of hominid evolution, the concept of such an
essential kind may represent a balancing act between what our
ancestors could and could not afford to ignore about their
environment. The concept of generic species allows people to perceive
and predict many important properties that link together the members
of a biological species actually living together at any one time, and
to distinguish such species from one another. By contrast, the ability
to appreciate the graded phylogenetic relationships between scientific
species, which involve vast expanses of geological time and
geographical space, would be largely irrelevant to the natural
selection pressures on hominid cognition.

Ernst Mayr (1969) calls such "local" species, which are readily
observed over one or a few generations to coexist in a given local
environment, "non-dimensional species" for two reasons: they are
manifest to the untrained eye, with no need for theoretical
reflection; and the perceptible morphological, ecological and
reproductive gaps separating such species summarize the evolutionary
barriers between them. Mayr argues that the awareness of
non-dimensional species provides the necessary condition for further
insight and exploration into phylogenetic species; any sufficient
condition for scientific understanding, however, must go beyond
essentialism.

People ordinarily assume that the various members of each generic
species share a unique underlying nature, or essence. This assumption
carries the inference of a strong causal connection between
superficially dissimilar or noncontiguous states or events - an
inference that other animals or primates do not seem capable of making
(cf. Kummer 1994). People reason that even three-legged, purring,
albino tiger cubs are by nature large, striped, roaring, carnivorous
quadrupeds. This is because there is presumably something "in" tigers
that is the common cause of them growing large, having stripes, eating
meat and roaring under "normal" conditions of existence. People expect
the disparate properties of a species to be integrally linked without
having to know precise causal relationships.

A biological essence is an intrinsic (i.e., nonartifactual)
teleological agent, which physically (i.e., nonintentionally) causes
the biologically relevant parts and properties of a generic species to
function and cohere "for the sake of" the generic species itself. For
example, even preschoolers in our culture consistently judge that the
thorns on a rose bush exist for the sake of there being more roses,
whereas physically similar depictions of barbs on barbed wire or the
protuberances of a jagged rock are not considered to exist for the
sake of there being more barbed wire or jagged rocks (Keil 1994).

This concept of underlying essence goes against the claim that
"biological essentialism is the theoretical elaboration of the
logical-linguistic concept, substance sortal" that applies to every
count noun (Carey 1996:194). Chair may be defined in terms of the
human function it serves, and mud in terms of its physical properties,
but neither have deep essences because neither is necessarily assumed
to be the unique outcome of an imperceptible causal complex. For
example, a three-legged or legless beanbag chair does not lack "its"
legs, because although most chairs "normally" have four legs they are
not quadrupedal by nature (cf. Schwartz 1978). Neither is the notion
of essence merely that of a common physical property. Red things
comprise a superficial natural class, but such things have little in
common except that they are red; and they presumably have few, if any,
features that follow from this fact.

People the world over assume that the initially imperceptible
essential properties of a generic species are responsible for the
surface similarities they perceive. People strive to know these deeper
properties but also assume that the nature of a species may never be
known in its entirety. This cognitive compulsion to explore the
underlying nature of generic species produces a continuing and perhaps
endless quest to better understand the surrounding natural world, even
though such understanding seldom becomes globally coherent or
consistent.

   1.2.2. A Taxonomic Experiment on Rank and Preference.

Given these observations, cognitive studies of the "basic level" are
at first sight striking and puzzling. In a justly celebrated set of
experiments, Rosch and her colleagues set out to test the validity of
the notion of a psychologically preferred taxonomic level (Rosch,
Mervis, Grey, Johnson & Boyes-Braem 1976). Using a broad array of
converging measures, they found that there is indeed a "basic level"
in category hierarchies of "naturally occurring objects," such as
"taxonomies" of artifacts as well as living kinds. For artifact and
living kind hierarchies, the basic level is where: (1) many common
features are listed for categories, (2) consistent motor programs are
used for the interaction with or manipulation of category exemplars,
(3) category members have similar enough shapes so that it is possible
to recognize an average shape for objects of the category, (4) the
category name is the first one to come to mind in the presence of an
object (e.g., "table" versus "furniture" or "kitchen table").

There is a problem, however: The basic level that Rosch et al. (1976)
had hypothesized for artifacts was confirmed (e.g., hammer, guitar);
however, the hypothesized basic level for living kinds (e.g., maple,
trout), which Rosch initially presumed would accord with the
generic-species level, was not. For example, instead of maple and
trout, Rosch et al. found that tree and fish operated as basic-level
categories for American college students. Thus, Rosch's basic level
for living kinds generally corresponds to the life-form level, which
is superordinate to the generic-species level (cf. Zubin & Köpcke 1986
for findings with German).

To explore this apparent discrepancy between preferred taxonomic
levels in small-scale and industrialized societies, and the cognitive
nature of ethnobiological ranks in general, we use inductive
inference. Although a number of converging measures have been used to
explore the notion of basic levels, there has been little direct
examination of the relationship between inductive inference and basic
levels. This is all the more surprising in view of the fact that a
number of psychologists and philosophers assume that basic-level
categories maximize inductive potential as intuitive "natural kinds"
which "scientific displines evolve to study" (Carey 1985:171; cf.
Gelman 1988, Millikan in press). Inference studies allow us to
directly test whether or not there is a psychologically preferred rank
that maximizes the strength of any potential induction about
biologically relevant information, and whether or not this preferred
rank is the same across cultures. If a preferred level carries the
most information about the world, then categories at that level should
favor a wide range of inferences about what is common among members
(cf. Anderson 1990).

The prediction is that inferences to a preferred category (e.g., white
oak to oak, tabby to cat) should be much stronger than inferences to a
superordinate category (oak to tree, cat to mammal). Moreover,
inferences to a subordinate category (spotted white oak to white oak,
short-haired tabby to tabby) should not be much stronger than or
different from inferences to a preferred category. What follows is a
summary of results from one representative set of experiments in two
very diverse populations: Midwestern Americans and Lowland Maya (for
complete results see Atran, Estin, Coley & Medin in press; Coley,
Medin & Atran in press).

   1.2.2.1. Subjects and Methods.

The Itzaj are Maya Amerindians living in the Petèn rainforest region
of Guatemala. Until recently, men devoted their time to shifting
agriculture, hunting and silviculture, whereas women concentrated on
the myriad tasks of household maintenance. The Itzaj comprised the
last independent native polity to be conquered by Spaniards (in 1697)
and they have preserved virtually all ethnobiological knowledge
recorded for Lowland Maya since the time of the initial Spanish
conquest (Atran 1993). Despite the current awesome rate of
deforestation and the decline of Itzaj culture, the language and ethic
of traditional Maya silviculture is still very much in evidence among
the generation of our informants who range in age from 50 to 80 years
old . The Americans were self-identified as people raised in Michigan
and recruited through an advertisement in a local newspaper.

Based on extensive fieldwork with the Itzaj, we chose a set of Itzaj
folk-biological categories of the kingdom (K), life-form (L),
generic-species (G), folk-specific (S), and folk-varietal (V) ranks.
We selected three plant life forms: che' = tree, ak' = vine, pok~che'
= herb/bush. We also selected three animal life forms: b'a'al~che'
kuxi'mal = "walking animal," i.e., mammal, ch'iich' = birds including
bats, k%y = fish. Three generic-species taxa were chosen from each
life form such that each generic species had a subordinate
folk-specific, and each folk-specific had a salient varietal.

Pretesting showed that participants were willing to make inferences
about hypothetical diseases. The properties chosen for animals were
diseases related to the "heart" (puksik'al), "blood" (k'ik'el), and
"liver" (tamen). For plants, diseases related to the "roots" (motz),
"sap" (itz) and "leaf" (le'). Properties were chosen according to
Itzaj beliefs about the essential, underlying aspects of life's
functioning. Thus, the Itzaj word puksik'al, in addition to
identifying the biological organ "heart" in animals, also denotes
"essence" or "heart" in both animals and plants. The term motz denotes
"roots," which is considered the initial locus of the plant puksik'al.
The term k'ik'el denotes "blood" and is conceived as the principal
vehicle for conveying life from the puksik'al throughout the body. The
term itz denotes "sap," which functions as the plant's k'ik'el. The
tamen, or "liver," helps to "center" and regulate the animal's
puksik'al. The le', or "leaf," is the final locus of the plant
puksik'al. Properties used for inferences had the form, "is
susceptible to a disease of the <root> called <X>." For each question,
"X" was replaced with a phonologically appropriate nonsense name (e.g.
"eta") in order to minimize the task's repetitiveness.

All participants responded to a list of over 50 questions in which
they were told that all members of a category had a property (the
premise) and were asked whether "all," "few," or "no" members of a
higher-level category (the conclusion category) also possessed that
property. The premise category was at one of four levels, either
life-form (e.g. L = bird), generic-species (e.g. G = vulture),
folk-specific (e.g. S= black vulture), or varietal (e.g. V =
red-headed black vulture). The conclusion category was drawn from a
higher-level category, either kingdom (e.g. K = animal), life-form
(L), generic-species (G), or folk-specific (S). Thus, there were ten
possible combinations of premise and conclusion category levels: L->K,
G->K, G->L, S->K, S->L, S->G, V->K, V->L, V->G, and V->S. For example,
a folk-specific-to-life form (S->L) question might be, "If all black
vultures are susceptible to the blood disease called eta, are all
other birds susceptible?" If a participant answers "no," then the
follow-up question would be "Are some or a few other birds susceptible
to disease eta, or no other birds at all?"

The corresponding life forms for the Americans were: mammal, bird,
fish, tree, bush and flower (on flower as an American life form see
Dougherty 1979). The properties used in questions for the Michigan
participants were "have protein X," "have enzyme Y," and "are
susceptible to disease Z." These were chosen to be internal,
biologically based properties intrinsic to the kind in question, but
abstract enough so that rather than answering what amounted to factual
questions participants would be likely to make inductive inferences
based on taxonomic category membership.

   1.2.2.2. Results.

Representative findings are given in Figure 1. Responses were scored
in two ways. First we totaled the proportion of "all or virtually all"
responses for each kind of question (e.g., the proportion of times
respondents agreed that if red oaks had a property, all or virtually
all oaks would have the same property). Second, we calculated
"response scores" for each item, counting a response of "all or
virtually all" as 3, "some or few" as 2, and "none or virtually none"
as 1. A higher score reflected more confidence in the strength of an
inference.

Figure 1a summarizes the results from all Itzaj informants for all
life forms and diseases, and shows the proportion of "all" responses
(black), "few" responses (checkered), and "none" responses (white).
For example, given a premise of folk-specific (S) rank (e.g., red
squirrel) and a conclusion category of generic-species (G) rank (e.g.,
squirrel), 49% of responses indicated that "all" squirrels, and not
just "some" or "none," would possess a property that red squirrels
have. Results were obtained by totaling the proportion of "all or
virtually all" responses for each kind of question (e.g., the
proportion of times respondents agreed that if red oaks had a
property, all or virtually all oaks would have the same property). A
higher score represented more confidence in the strength of the
inductive inference. Figure 1b summarizes the results of Michigan
response scores for all life forms and biological properties.

Response scores were analyzed using t-tests with significance levels
adjusted to account for multiple comparisons. Figure 2 summarizes the
significant comparisons (p-values) for "all" responses, "none"
responses and combined responses. For all comparisons, n = 12 Itzaj
participants and n= 21 American participants (for technical details
see Atran et al. in press).

Following the main diagonals of Figures 1 and 2 refers to changing the
levels of both the premise and conclusion categories while keeping
their relative level the same (with the conclusion one level higher
than the premise). Induction patterns along the main diagonal indicate
a single inductively preferred level. Examining inferences from a
given rank to the adjacent higher-order rank (i.e., V->S, S->G, G->L,
L->K), we find a sharp decline in strength of inferences to taxa
ranked higher than generic species, whereas V->S and S->G inferences
are nearly equal and similarly strong. Notice that for "all"
responses, the overall Itzaj and Michigan patterns are nearly
identical.

Moving horizontally within each graph in Figures 1 and 2 corresponds
to holding the premise category constant and varying the level of the
conclusion.[9] Here we find the same pattern for "all" responses for
both Itzaj and Americans as we did along the main diagonal. However,
in the combined response scores ("all" + "few") there is now evidence
of increased inductive strength for higher-order taxa among Americans
versus Itzaj. On this analysis, both Americans and Itzaj show the
largest break between inferences to generic species versus life forms.
But only American subjects also show a consistent pattern of rating
inferences to life-form taxa higher than to taxa at the level of the
folk kingdom: G->K vs. G->L, S->K vs. S->L, and V->K vs. V->L.

Finally, moving both horizontally and along the diagonal, for Itzaj
there is some hint of a difference between inductions using
conclusions at the generic-species versus folk-specific levels: V->G
and S->G are modestly weaker than V->S. Regression analysis reveals
that for Itzaj, the folk-specific level accounts for a small
proportion of the variance beyond the generic species (1.4%), but a
significant one (F > 4). For Michigan participants, the folk-specific
level is not differentiated from the generic-species level (0.2, not
significant). In fact, most of the difference between V->G and V->S
inductions results from inference patterns for the Itzaj tree life
form . There is evidence that Itzaj confer some preferential status
upon trees at the folk-specific level (e.g. savanna nance tree). Itzaj
are forest-dwelling Maya with a long tradition of agroforestry that
antedates the Spanish conquest (Atran 1993).

   1.2.2.3. Discussion.

These results indicate that both the ecologically inexperienced
Americans and the ecologically experienced Itzaj prefer taxa of the
generic-species rank in making biological inferences; the findings go
against a simple relativist account of cultural differences in
folk-biological knowledge. However, the overall effects of cultural
experience on folk-biological reasoning are reflected in more subtle
ways that do not undermine an absolute preference for the generic
species across cultures. In particular, the data point to a relative
downgrading of inductive strength to higher ranks among industrialized
Americans through knowledge attrition owing to lack of experience and
a relative upgrading of inductive strength to lower ranks among
silvicultural Maya through expertise.

A secondary reliance on life forms arguably owes to Americans' general
lack of actual experience with generic species (Dougherty 1978). In
one study, American students used only the name "tree" to refer to 75%
of the species they saw in a nature walk (Coley, Medin & Atran in
press). Although Americans usually can't tell the difference between
beeches and elms, they expect that biological action in the world is
at the level of beeches and elms and not tree. Yet without being able
at least to recognize a tree, they would not even know where to begin
to look for the important biological information. The Itzaj pattern
reflects both overall preference for generic species and the secondary
importance of lower-level distinctions, at least for kinds of trees. A
strong ethic of reciprocity in silviculture still pervades the Itzaj;
the Maya tend trees so that the forest will tend to the Maya (Atran &
Medin 1997). This seems to translate into an upgrading of biological
interest in tree folk-specifics.

These findings cannot be explained by appeals either to cross-domain
notions of perceptual "similarity" or to the structure of the world
"out there." On the one hand, if inferential potential were a simple
function of perceptual similarity then Americans should prefer life
forms for induction (in line with Rosch et al.). Yet Americans prefer
generic species as do Maya. On the other hand, objective reality -
that is, the actual distribution of biological species within groups
of evolutionarily related species - does not substantially differ in
the natural environments of Midwesterners and Itzaj. Unlike Itzaj,
however, Americans perceptually discriminate life forms more readily
than generic species. True, there are more locally recognized species
of tree in the Maya area of Peten, Guatemala than in the Midwest
United States. Still, the readily perceptible evolutionary "gaps"
between species are roughly the same in the two environments (most
tree genera in both environments are monospecific). If anything, one
might expect that having fewer trees in the American environment
allows each species to stand out more from the rest (Hunn 1976). For
birds the relative distribution of evolutionarily related species also
seems to be broadly comparable across temperate and rainforest
environments (Boster 1988).

An inadequacy in current accounts of preferred taxonomic levels may be
a failure to distinguish domain-general mechanisms for best clustering
stimuli from domain-specific mechanisms for best determining loci of
biological information. To explain Rosch's data it may be enough to
rely on domain-general, similarity-based mechanisms. Such mechanisms
may generate a basic level in any number of cognitive domains, but not
the preferred level of induction in folk biology.

Perhaps humans are disposed to take tight clusters of covariant
perceptual information as strong indicators of a rich underlying
structure of biological information. This may be the "default" case
for humans under "normal" conditions of learning and exposure to the
natural world. By and large, people in small-scale societies would
live under such "normal" conditions, involving the same general sorts
of ambient circumstances that led to the natural selection of
cognitive principles for the domain of folk biology. People in urban
societies, however, may no longer live under such "default" conditions
(except for hunters, bird watchers etc., Tanaka & Taylor 1991.)

How, then, can people conceive of a given folk-biological category as
a generic species without always (or mostly) relying on perception?
Ancillary encyclopedic knowledge may be crucial. Thus, one may have
detailed knowledge of dogs but not oaks. Yet a story that indicates
where an oak lives, or how it looks or grows, or that its life is
menaced may be sufficient to trigger the assumption that oaks comprise
a generic species just as dogs do. But such cultural learning produces
the same results under widely divergent conditions of experience in
different social and ecological environments. This indicates that the
learning itself is strongly motivated by cross-culturally shared
cognitive mechanisms that do not depend primarily on experience.

In conjunction with encyclopedic knowledge of what is already known
for the natural world, language is important in targeting preferred
kinds. In experiments with children as young as two years old, Gelman
and her colleagues showed that sensitivity to nomenclatural patterns
and other linguistic cues helps guide folk-biological inferences about
information that is not perceptually obvious, especially for
categories believed to embody an essence (Gelman, Coley & Gottfried
1994; Hall & Waxman 1993). Language alone, however, is not enough to
induce the expectation that little known generic species convey more
biological information than better known life forms for Americans.
Some other process must invest the generic-species level with
inductive potential. Language alone can only signal that such an
expectation is appropriate for a given lexical item; it cannot
determine the nature of that expectation.

Why assume that an appropriately tagged item is the locus of a "deep"
causal nexus of biological properties and relationships? It is
logically impossible that such assumptions and expectations come from
(repeated exposure to) the stimuli themselves. Input to the mind alone
cannot cause an instance of experience (e.g., a sighting in nature or
in a picture book), or any finite number of fragmentary instances, to
be generalized into a category that subsumes a rich and complex set of
indefinitely many instances. This projective capacity for category
formation can only come from the mind, not from the world alone.

The empirical question, then, is whether or not this projective
capacity of the mind is simply domain-general, or also
domain-specific. For any given category domain - say, living kinds as
opposed to artifacts or substances - the process would be
domain-general if and only if one could generate the categories of any
number of domains from the stimuli alone together with the very same
cognitive mechanisms for associating and generalizing those stimuli.
But current domain-general similarity models of category formation and
category-based reasoning fail to account for the generic species as a
preferred level for folk-biological taxonomy across cultures.

Our findings suggest that fundamental categorization processes in folk
biology are rooted in domain-specific conceptual assumptions rather
than in domain-general perceptual heuristics. Subsistence cultures and
industrialized cultures may differ in the level at which organisms are
most easily identified, but they both still believe that the same
absolute level of reality is preferable for biological reasoning,
namely, the generic-species rank. This is because they expect the
biological world to partition at that rank into nonoverlapping kinds,
each with its own unique causal essence, whose visible products may or
may not be readily perceived.

People anticipate that the biological information value of these
preferred kinds is maximal whether or not there is also a visible
indication of maximal covariation of perceptual attributes. This does
not mean that more general perceptual cues have no inferential value
when applied to the folk-biological domain. On the contrary, the
evidence points to a significant role for such cues in targeting
basic-level life forms as secondary foci for inferential understanding
in a cultural environment where biological awareness is relatively
poor, as among many Americans. Possibly there is an evolutionary
design to having both domain-general perceptual heuristics and
domain-specific learning mechanisms: the one enabling flexible
adaptation to the variable conditions of experience; the other more
invariable in steering us to those abiding aspects of biological
reality that are causally recurrent and especially relevant for the
emergence of human life and cognition.

   1.3. Evolutionary Ramifications: Folk Biology as a Core Domain of Mind and
   Culture.

A speculative but plausible claim in light of our observations and
findings is that folk biology is a core domain for humans. A core
domain is a semantic notion, philosophically akin to Kant's "synthetic
a priori." The object domain, which consists of generic species of
biological organisms, is the extension of an innate cognitive module.
Universal taxonomy is a core module, that is, an innately determined
cognitive structure that embodies the naturally selected ontological
commitments of human beings and provides a domain-specific mode of
causally construing the phenomena in its domain (for a more
disembodied view of innate "modes of construal," see Keil 1995). In
particular, the cognitive structure of folk biology specifies that
generic species are the preferred kinds of things that partition the
biological world, that these generic species are composed of causally
related organisms that share the same vitalist (teleo-essentialist)
structure, and that these generic species further group together into
causally related but mutually exclusive groups under groups. In sum,
the generic species is a core concept of the folk-biology module.

Core modules share much with Fodor's (1983) input modules. Both are
presumably naturally selected endowments of the human mind that are
initially activated by a predetermined range of perceptual stimuli.
However, there are differences. Input modules, unlike core modules,
are hermetically closed cognitive structures that have exclusive
access to the mental representations that such input systems produce.
For example, syntactic- recognition schemata and facial-recognition
schemata respectively deal exclusively and entirely with syntactic
recognition and facial recognition. By contrast, core modules have
preferential rather than proprietary access to their domain-specific
representations (Atran 1990:285). For example, core modules for naive
physics, intuitive psychology or folk biology can make use of one
another's inputs and outputs, although each module favors the
processing of a different predetermined range of stimuli.

Moreover, the ability to use a "metarepresentational module," which
takes as inputs the outputs of all other modules, allow changes
(restructurings and extensions) to operate over the initial core
domain as a result of developing interactions with our external
(ambient) and internal (cognitive) environment. Flexibility in core
modules, Sperber (1994) argues, makes evolutionary sense of how humans
so quickly acquire distinct sorts of universal knowledge, which
individuals and cultures can then work on and modify in various ways.
Sperber's discussion also indicates, in principle, how ordinary people
and cognitive scientists can manage the "combinatorial explosion" in
human information without simply making it all grist for an
inscrutable central-processing mill.

A living kind module enables humans to apprehend the biological world
spontaneously as a partitioning into essence-based generic species and
taxonomically related groups of generic species. This directs
attention to interrelated and mutually constraining aspects of the
plant and animal world, such as the diverse and interdependent
functioning of heterogeneous body parts, maturational growth,
inheritance and natural parentage, disease and death. Eventually,
coherent "theories" of these causal interrelations might develop under
particular learning conditions (Carey 1985) or historical
circumstances (Atran 1990). Such systematic elaboration of biological
causality, however, is not immediately observable or accessible.

Core knowledge that is domain-specific should involve dedicated
perceptual-input-analyzers, operating with little interference or
second-guessing from other parts of the human conceptual system (Carey
1996, Gigerenzer in press). What might be the evolutionary algorithm
that activates or triggers the living kind module's selective
attention to generic species? In the absence of experiments or other
reliable data, we can only speculate. Evidence from other core
domains, such as naive physics and intuitive psychology, helps as both
guide and foil to speculation about triggering algorithms for a
living-kind module. For humans as well as animals, there is some
evidence of at least two distinct but hierarchically related
triggering algorithms, each involving a dedicated
perceptual-input-analyzer that attends to a restricted range of
information.

There is an algorithm that attends only to the external movements of
rigid bodies that obey something like the laws of Newtonian mechanics
in a high-friction environment. Thus, infants judge that an object
moving on a plane surface will continue along that surface in a
straight path until it stops, but will not jump and suspend itself in
mid-air (Spelke 1990). There is also an algorithm that attends to the
direction and acceleration of objects not predictable by "naive
mechanics." If the motion pattern of one object on a computer screen
centers on the position of another object, so that the first object
circles around the second object, and speeds up towards or away from
it, then infants judge the first object to be self-propelled or
"animate" (Premack & Premack 1994).

Of course, algorithms for animateness and intentionality can lead to
mistakes. They surely did not evolve in response to selection
pressures involving two-dimensional figures moving across computer
screens. These inhabitants of flatland just happen to fall within the
actual domains to which the modules for animacy and intentionality
spontaneously extend, as opposed to the proper domains for which the
modules evolved (i.e., animate beings and intentional agents). Much as
the actual domain of frog food-getting intelligence involves tongue
flicking at dark points passing along a frog's field of vision,
whereas the proper domain is more about catching flies (Sperber 1994).

Algorithms for animacy and intentionality do not suffice to
discriminate just living kinds, that is, generic species. On the one
hand, they fail to distinguish plants from non-living kinds. Yet
people everywhere distinguish plants into generic species just as they
do animals. An algorithm that cues in primarily on the relative
movement of heterogeneous and diversely connected parts around an
object's center of gravity probably plays an important role in
discerning animals and plants (perhaps first as they move in the wind,
then grow, etc.), although it too may initially err (plastic plants,
perhaps clothes on a line). On the other hand, algorithms for animacy
and intentionality fail to distinguish humans from nonhuman living
kinds, that is, plants and animals.

It is animals and plants that are always individuated in terms of
their unique generic species, whereas humans are individuated as both
individual agents and social actors in accordance with inferred
intentions rather than expected clusters of body parts. People
individuate humans (as opposed to animals) with the additional aid of
a variety of domain-specific "recognizers" for individual human faces,
voices, gestures and gaits, which richly motivate inferences about
motion and intention from rather partial and fleeting perceptual cues
(Fodor 1983, Tooby & Cosmides 1992). Yet no known aboriginal culture -
or any culture not exposed to Aristotle - believes that humans are
animals or that there is an ontological category undifferentiated
between humans and animals.

Let us further speculate about selection pressures involved in our
automatic attention to human individuals versus our automatic
attention to generic species. A characteristic of primates (and some
other vertebrates) is that they are social animals who can distinguish
individuals of their species, unlike termites who cannot (Kummer,
Daston, Gigerenzer & Silk in press). There is evidence that as long as
two million years ago, Homo habilis relied upon nonkin to hunt, gather
and scavenge for subsistence (Isaac 1983). In order to handle the
social contracts required for this mode of subsistence, coalition
forming and cooperation with nonkin were probably required. This
probably entailed a negotiation of intentions with individuals who
could not be identified by indications of blood relationship.

In regard to animals and plants, there is also evidence of varied and
wide-ranging diet and subsistence patterns in hominid social camps at
that time (Bunn 1983). In such a camp, it could be supremely important
to know which individual should be recruited in a food-sharing
coalition if only to avoid "free riders" who take without giving
(Cosmides & Tooby 1989). But it would hardly matter to know the
individual identity of lions which could eat you, nettles which could
sting you, or deer and mangos which you could eat. Knowing not just
the habits of particular species, but making taxonomic inferences
about the habits and relationships of groups of biologically related
species would be likely to increase the effectiveness (benefit) of
such knowledge-based subsistence immeasurably, with little or no added
investment (cost) in time or effort (trial-and-error learning).

The special evolutionary origins of domain-specific cognitive modules
should have special bearings on cultural evolution. One might have
expected the implications of domain-specificity to be compelling for
those who reason in line with Dawkins (1976), viewing the emergence of
culture as a selection process. Unfortunately, aside from notable
exceptions (Sperber 1994; Tooby & Cosmides 1992; cf. Lumsden & Wilson
1981), the focus is primarily on how, for example, "Chinese minds
differ radically from French minds" (Dennett 1995:365; cf.
Cavalli-Sforza & Feldman 1981; Durham 1991). Nevertheless, Dawkins's
idea may be a good idea for the study of human cultures, suitably
modified by the findings and concerns of cognitive anthropology. His
idea is that there may be cultural units that function in social
evolution just as there are biological units that function in
biological evolution. He calls these units of cultural transmission
"memes" - a word that sounds like "gene" and evokes Latin and Greek
words for "imitation." One modification consists in restricting highly
imitative, replicating memes to knowledge produced by core domains,
that is, to memes that have an identifiable syntactic as well as a
semantic aspect. In this respect, folk-biological knowledge is a core
meme.

A core meme, like universal taxonomy, differs from a developing meme,
like the culturally specific elaboration of a scientific research
program, in a number of interrelated ways. An apparent difference is
in the closer resemblance of core memes to genes. First, for core
memes, like genes, there is a strong alignment of syntactic
("genotypic") and semantic ("phenotypic") identity. For example, the
universal structure of folk-biological taxonomy arguably emerges from
a modular cognitive capacity - a mental faculty - that evolved as an
effective means of capturing perceptibly relevant and recurrent
aspects of ancestral hominid environments.

As a result, humans "conceptually perceive" the biological world in
more or less the same way. Processes of perceiving and reasoning about
generic species are intimately connected: they are guided by the same
knowledge system. The folk-biology module focuses attention on
perceptual information that can reveal that an object is a living
kind, or organism, by uniquely assigning it to one or another of the
fundamental partitions of the readily perceptible biological world.
Thus, the key feature of folk biology, belonging to a preferred
taxonomic rank and a causally essential category, is induced from
spatiotemporal analysis via a triggering algorithm that attends to a
limited set of perceptual cues whose presence signals an organism as
belonging to a generic species.

Second, for core memes, conceptual replication involves information
being physically transmitted largely intact from physical vehicle to
physical vehicle without any appreciable sequencing of vehicles. As in
genetic replication, replication of core memes involves fairly
high-fidelity copying and a relatively low rate of mutation and
recombination. Mental representations of generic species, for
instance, are transmitted from brain to brain via public
representations such as uttered names and pointings (Sperber 1985). It
often suffices, however, that a single fragmentary instance of
experience - a naming or sighting by ostension in a natural or
artificial setting - "automatically" triggers the transmission and
projection of that instance into a richly structured taxonomic context
(Atran and Sperber 1991).

By contrast, a developing meme requires institutionalized channeling
of information. For example, specific scientific schools or research
programs involve more or less identifiable communities of scientists,
journals, instruments, laboratories and so forth. Institutionalization
is necessary because the information is harder to learn and keep
straight, but is also more readily transformed and extended into new
or different knowledge. This often requires formal or informal
instruction to sustain the sequencing of information, and to infuse
output with added value by inciting or allowing transformation of
input via interpolation, invention, selection, suppression and so
forth (see Latour 1987 and Hull 1988 for different insights into
institutional constraints).

Third, a core meme does not depend for its survival on the cognitive
division of labor in a society or on durable transmission media. For
example, children can learn about species from written texts, films or
picture books; nevertheless, noninstitutionalized transmission of such
information in an illiterate society is usually quite reliable as long
as there is an unbroken chain of oral communication (within the living
memory of the collective) about events in the natural world.
Developing memes, however, typically mobilize information of such
quantity, diverse quality and expertise that single minds cannot - for
lack of capacity or because of other cognitive demands - keep track of
all that is needed to understand the information and pass it along.
Because scientists can usually only work on bits and pieces of the
information in the field at any particular time and place, but may
also need to consult information elaborated elsewhere or let fallow
for generations (e.g., Mendel's discoveries), durable media are
required for that information to usefully endure.

Fourth, a core meme does not primarily depend on metacognitive
abilities, although it may make use of them (e.g., in stories,
allegories, analogies). For the harder-to-learn beliefs of developing
memes to grow requires the mingling of ideas from different sources,
including different sorts of core memes. For example, numerical and
mechanical knowledge now play important, and perhaps preponderant,
roles in areas of molecular biology. Mingling of ideas implies the
transfer of diverse domain-specific outputs into a domain-neutral
representation. A domain-neutral metarepresentation can then function
as input for further information processing and development.

Fifth, the involvement of core memes in developing metacognitive memes
that ride piggyback on core memes or stem from them, such as totemism
or biological systematics, allows us in principle to distinguish the
convergent evolution of memes across cultures from borrowing,
diffusion and descent. If all memes were purely semantic, such a
distinction might well be practically impossible in the absence of
clear historical traces. One case of convergent evolution is the
spontaneous emergence of totemism - the correspondence of social
groups with generic species - at different times and in different
parts of the world. Why, as Lèvi-Strauss (1963) aptly noted, are
totems so "good to think"? In part, totemism is metacognitive because
it uses representations of generic species to represent groups of
people; however, this pervasive metarepresentational inclination
arguably owes its recurrence to its ability to ride piggyback on
folk-biological taxonomy, which is not primarily or exclusively
metacognitive. Consider:

Generic species and groups of generic species are inherently
well-structured, attention-arresting, memorable and readily
transmissible across minds. As a result, they readily provide
effective pegs on which to attach knowledge and behavior of less
intrinsically well-determined social groups. In this way totemic
groups can also become memorable, attention-arresting and
transmissible across minds. These are the conditions for any meme to
become culturally viable (see Sperber 1996 for a general view of
culture along the lines of an "epidemiology of representations"). A
significant feature of totemism that enhances both memorability and
its capacity to grab attention is that it violates the general
behavior of biological species: members of a totem, unlike members of
a generic species, generally do not interbreed, but only mate with
members of other totems in order to create a system of social
exchange. Notice that this violation of core knowledge is far from
arbitrary. In fact, it is such a pointed violation of human beings'
intuitive ontology that it readily mobilizes most of the assumptions
people ordinarily make about biology in order to better help build
societies around the world (Atran & Sperber 1991).

In the structuring of such metarepresentations, then, the net result
appears close to an optimal balance between memorability,
attention-grabbing power and flexibility in assimilating and adapting
to new and relevant information. This is to assure both ease of
transmissibility and longstanding cultural survival. More generally,
incorporating recurrently emerging themes in religious and symbolic
thought into cognitive science can be pursued as a research program,
which focuses on the transmission metarepresentational elaborations of
intuitive ontologies or core memes (see Boyer 1994 for such a general
framework for the study of religion).

This distinction between convergent and descendant metacognitive memes
is not absolute. Creationism, for example, has both cross-culturally
recurrent themes of supernatural species reification and particular
perspectives on the nature of species that involve outworn scientific
theories as well as specific historical traditions. Here as well,
knowledge of the universal core of such beliefs helps to identify what
is, and what is not, beyond the range of ordinary common sense (Atran
1990). Finally, even aspects of the metarepresentational knowledge
that science produces as ouput can feed back (as input) in subtle and
varied ways into the core module's actual domain: for example,
learning that whales aren't fish and that bats aren't birds. But the
feedback process is also constrained by the intuitive bounds of
domain-specific, common sense (Atran 1987b).

The message here is that evolutionary psychology might profit from a
source barely tapped: the study of cultural transmission. Some bodies
of knowledge have a life of their own, only marginally affected by
social change (e.g., intuitive mechanics, basic color classification,
folk-biological taxonomies); others depend for their transmission, and
hence for their existence, on specific institutions (e.g., totemism,
creationism, evolutionary biology).[10] This suggests that culture is
not an integrated whole, relying for its transmission on
undifferentiated cognitive abilities. But the message is also one of
"charity" concerning the mutual understanding of cultures (Davidson
1984): anthropology is possible because underlying the variety of
cultures are diverse but universal commonalities. This message also
applies to the disunity and comprehensibility of science (part 3).

2. Cultural Elaborations of Universal Taxonomy

Despite the evident primacy of ranked taxonomies in the elaboration of
folk-biological knowledge in general, and the cognitive preference for
generic species in particular, I no longer think that folk taxonomy
defines the inferential character of folk biology as strongly as I
indicated in a previous work, Cognitive Foundations of Natural History
(Atran 1990). Mounting empirical evidence gathered with colleagues
suggests that although universal taxonomic structures universally
constrain and guide inferences about the biological world, different
cultures (and to a lesser extent different individuals within a
culture) show flexibility in which inferential pathways they choose
(for details see Atran 1995, in press; Medin et al. 1996, 1997; Lûpez,
Atran, Coley, Medin & Smith 1997; Coley, Medin, Proffitt, Lynch &
Atran in press). Different tendencies apparently relate to different
cultural criteria of relevance for understanding novelties and
uncertainties in the biological world and in adapting to them.

For example, among the Itzaj Maya, in contrast to the systematic use
of taxonomies by scientists or modern (non-aboriginal) American folk,
understanding ecological relationships seems to play a role on a par
with morphological and underlying biological relationships in
determining how taxa may be causally interrelated. For centuries,
Itzaj have managed to so use their folk-biological structures to
organize and maintain a fairly stable, context-sensitive biological
and ecological order. In a different way, scientists use taxonomies as
heuristics for reaching a more global, ecologically context-free
understanding of biological relationships underlying the diversity of
life. American folk unwittingly pursue a compromise of sorts:
maintaining ecologically valid folk categories, but reasoning about
them as if they were theory-based. Irrelevancy often results.

   2.1. Taxonomy-Based Inference Across Cultures.

To illustrate, consider some recent experimental findings. Our
intention was to see whether and how Americans and Maya reason the
same or differently from their respective taxonomies to determine the
likely distribution of unfamiliar biologically-related properties. Our
strategy was as follows: First we asked individual informants to
perform successive sorting tasks of name cards or colored picture
cards (or specimens in Itzaj pilot studies) in order to elicit
individual taxonomies. Then we used statistical measures to see
whether or not the data justified aggregating the individual
taxonomies for each informant group into a single "cultural model"
that could confidently retrodict most (of the variance in) informant
responses. Finally, we used the aggregated cultural taxonomies to
perform various category-based inference tasks with the same or
different informants. At each stage of the sorting and inference tasks
we asked informants to justify responses. In sum, our techniques
enabled us to describe an aggregate model of taxonomy for each
population in order to determine emergent patterns of cultural
preferences in matters of biological inference.

   2.1.1. An Experimental Method for Generating Taxonomies.

In the sorting tasks, each set of cards represented either all the
generic species of a life form (Itzaj and Michigan mammals) or
intermediate category (Itzaj palms), or a large range of the generic
species of a life form (e.g., all local trees in the Evanston-Chicago
area for people living in the area). The aim was to obtain individual
taxonomies that covered the range of relationships between
intermediate folk taxa, that is, taxonomic relationships between the
generic-species and life-form levels. This was motivated by the fact
that the boundaries of intermediate taxa vary somewhat more across
individuals and cultures than do ranked taxa, and our goal was to
explore as much the differences as the similarities in taxonomy-based
reasoning across cultures. Furthermore, the intermediate level of
taxonomy is where evolutionary relationships are most visibly manifest
and comprehensible (both in the history of science and among educated
lay folk, see Atran 1983), and where ecological relationships are most
manifest for Maya (e.g., in the habits of arboreal mammals on the
fruiting and reproduction of canopy trees). We thought these factors
would increase the possibility of ascertaining whether significant
differences between Americans and Maya relate to different goals for
understanding biological relationships: one weighted by the influence
of science in American culture, and the other weighted by interests of
subsistence and survival in the Maya rainforest.

   2.1.1.1. Methods.

What follows is a brief account of findings in regard to all mammals
represented in the local environments of the Itzaj and Michigan
groups, respectively.[11] For Itzaj we included bats, although Itzaj
do not consider them mammals. For the students we included the
emblematic wolverine, although it has practically disappeared from
Michigan. We asked American informants to sort name cards of all local
mammal generic species into successive piles according to the degree
they "go together by nature." For Itzaj, name cards were Maya words in
Latin letters and informants were asked to successively sort cards
according to the degree to which they "go together as companions"
(uy-et'~ok) of the same "natural lineage" (u-ch'ib'al). When
informants indicated no further desire to successively groups cards
the first piles were restored and the informants were asked to
subdivide the piles until they no longer wished to do so. The
"taxonomic distance" between any two taxa (cards) was then calculated
according to where in the sorting sequence they were first grouped
together. While a majority of Itzaj informants were functionally
illiterate, they had no trouble in manipulating name cards as mnemonic
icons. No differences were observed in handling cards between literate
and illiterate Itzaj, and no statistically significant differences in
results. We chose names cards over pictures or drawings to minimize
stimulus effects and maximize the role of categorical knowledge.

   2.1.1.2. Results: Convergence and Divergence in Intermediate-Level
   classifications.

Results indicate that the individual mammal taxonomies of Itzaj and
students from rural Michigan are all more or less competent
expressions of comparably robust cultural models of the biological
world.[12] To compare the structure and content of cultural models
with one another, and with scientific models, we mathematically
compared the topological relations in the tree structure of each
group's aggregate taxonomy with those of a classic evolutionary
taxonomy, that is, one based on a combination of morphological and
phylogenetic considerations.[13]

There was substantial shared agreement between the aggregated
taxonomies of Itzaj (Figure 3) and Michigan students (Figure 4),
between evolutionary taxonomy (Figure 5) and Itzaj taxonomy, and
between evolutionary taxonomy and the American folk taxonomy.
Agreement between the intermediate folk taxonomies and evolutionary
taxonomy is maximized at around the level of the scientific family,
both for Itzaj and Michigan subjects, indicating an intermediate-level
focus in the folk taxonomies of both cultures. On the whole, taxa
formed at this level are still imageable (e.g., the cat or dog
families).

A closer comparison of the folk groupings in the two cultures,
however, suggests that there are at least some cognitive factors at
work in folk-biological classification that are mitigated or ignored
by science. For example, certain groupings, such as felines + canines,
are common to both Itzaj and Michigan students, although felines and
canines are phylogenetically further from one another than either
family is to other carnivore families (e.g., mustelids, procyonids,
etc.). These groupings of large predators indicate that size and
ferocity or remoteness from humans is a salient classificatory
dimensions in both cultures (cf. Henley 1969, Rips et al. 1973). These
are dimensions that a corresponding evolutionary classification of the
local fauna does not highlight.

An additional nonscientific dimension in Itzaj classification, which
is not present in American classification, relates to ecology. For
example, Itzaj form a group of arboreal animals, including monkeys as
well as tree-dwelling procyonids (kinkajou, cacomistle, raccoon) and
squirrels (a rodent). The ecological nature of this group was
independently confirmed as follows: We asked informants to tell us
which plants are most important for the forest to live. Then, we
aggregated the answers into a cultural model, and for each plant in
the aggregate list we asked which animals most interacted with it
(without ever asking directly which animals interact with one
another). The same group of arboreal animals emerged as a stable
cluster in interactions with plants.

Other factors in the divergence between folk and scientific taxonomies
are related both to science's global perspective in classifying local
biota and to its reliance on biologically "deep," theoretically
weighted properties of internal anatomy and physiology. Thus, the
opossum is the only marsupial in North and Central America. Both Itzaj
and Midwesterners relate the opossum to skunks and porcupines because
it shares with them readily perceptible features of morphology and
behavior. From a scientific vantage, however, the opossum is
taxonomically isolated from all the other locally represented mammals
in a subclass of its own. One factor mitigating the ability of Itzaj
or Midwesterners to appreciate the opossum as scientists do is the
absence of other locally present marsupials to relate the opossum to.
As a result, both Michigan students and Itzaj are apparently unaware
of the deeper biological significance of the opossum's lack of a
placenta.

   2.1.2. Taxonomy-Driven Inductions.

Our inference studies were designed to further explore how the
underlying reasons for these these apparent similarities and
differences in intermediate-level taxonomies might inform
category-based inductions among Maya, lay Americans and scientists. We
tested for three category-based induction phenomena: Taxonomic
Similarity, Taxonomic Typicality and Taxonomic Diversity (cf.
Osherson, Smith, Wilkie, Lûpez & Shafir 1990).

   2.1.2.1 Taxonomic Similarity.

Similarity involves judging whether inference from a given premise
category to a conclusion category is stronger than inference from some
other premise to the same conclusion, where the premise and conclusion
categories are those in the aggregate taxonomic tree. Similarity
predicts that the stronger inference should be the one where the
premise is closest to the conclusion, with "closeness" measured as the
number of nodes in the tree one has to go through to reach the
conclusion category from the premise category. So, suppose that sheep
have some unfamiliar property (e.g., "ulnar arteries") or are
susceptible to an unknown disease ("eta"). Suppose, as an alternative
premise, that cows have a different property ("sesamoid bones") or are
susceptible to a different disease (e.g., "ina"). Following any of the
three taxonomies (Maya, American or evolutionary), one should conclude
that is it more likely that goats have what sheep have than what cows
have, because goats are taxonomically closer to sheep than they are to
cows.

If similarity is a built-in feature of folk taxonomy, then American
and Maya inductions should converge and diverge where their taxonomies
do. They should also resemble and depart from scientific inductions
where their taxonomies do regarding the scientific taxonomy. In fact,
both Americans and Maya chose items like sheep/goat versus cow/goat.
This confirms the convergence of the scientific taxonomy with
reasoning among both Americans and Maya precisely where the structure
of their respective taxonomies should lead us to expect convergence.

Both also chose items like opossum/porcupine versus
squirrel/porcupine, which confirms the expected convergence between
Maya and American classifications, and also the expected divergence of
both groups from scientific classification. Choice of items such as
dog/fox for Americans but cat/fox for Maya confirms that Americans
reason more in line with scientific classifications in such cases than
do Maya. In fact, justifications show that Itzaj recognize numerous
similarities between foxes and dogs (snout, paw, manner of copulation)
but judge that foxes are closer to cats because of interrelated
aspects of size and predatory habits.

   2.1.2.2. Taxonomic Typicality.

The metric for typicality, like the one for similarity, is given by
the taxonomy itself, as the lowest average taxonomic distance. In
other words, the typicality of an item (e.g., a generic species) is
the average taxonomic distance of that item to all other items in the
inclusive category (e.g., life form). Items that are more typical
provide greater coverage of the category than items that are less
typical. For example, Itzaj choose the items jaguar/mammal or mountain
lion/mammal over squirrel/mammal or raccoon/mammal, judging that all
mammals are more like to be susceptible to a disease that jaguars or
mountain lions have than to a disease that squirrels or raccoons have.

This is because Maya consider jaguars and mountain lions more typical
of the mammals than are squirrels and raccoons. In fact, jaguars and
mountain lions are not merely typical for Itzaj because they are more
directly related to other mammals than are squirrels and raccoons;
they also more closely represent an ideal standard of the "true
animal/mammal" (jach b'a'al~che') against which the appearance and
behavior of all other animals may be judged. This is evident from
Itzaj justifications as well as from direct ratings of which mammals
the Itzaj consider to be the "truest."

By contrast, American informants choose the items squirrel/mammal or
raccoon/mammal over bobcat/mammal or lynx/mammal, presumably because
they consider squirrels and raccoons are more typical of mammals for
Americans than are bobcats and lynxes. Note that typicality in these
cases cannot be attributed to frequency of occurrence or encounter.
Our American subjects were all raised in rural Michigan, where the
frequency of encounter with squirrels, raccoons, bobcats and lynxes is
nowadays about as likely as the corresponding Itzaj encounter with
squirrels, raccoons, jaguars and mountain lions. Both the Americans
and Maya were also more or less familiar with all animals in their
respective tasks.

In each case for which we have Itzaj typicality ratings, the "truest"
and most taxonomically-typical taxa are large, perceptually striking,
culturally important and ecologically prominent. The dimensions of
perceptual, ecological and cultural salience all appear necessary to a
determination of typicality, but none alone appears to be sufficient.
For example, jaguars are beautiful and big (but cows are bigger),
their predatory home range (about 50 km2) determines the extent of a
forest section (but why just this animal's home range?), and they are
"lords" of the forest (to which even the spirits pay heed). In other
words, typicality for the Itzaj appears to be an integral part of the
human (culturally-relevant) ecology. Thus, the Itzaj say that wherever
the sound of the jaguar is not heard, there is no longer any "true"
forest, nor any "true" Maya. Nothing of this sort appears to be the
case with American judgments of biological typicality and
typicality-based biological inference. Thus, the wolverine is
emblematic in Michigan, but carries no preferential inductive load.

   2.1.2.3. Taxonomic Diversity.

Like taxonomically defined typicality, diversity is a measure of
category coverage. But a pair of typical items provides less coverage
than, say, a pair containing one item that is typical and another that
is atypical. For example, given that horses and donkeys share some
property, but that horses and gophers share some other property, then
our American subjects judge that all mammals are more likely to have
the property that horses share with gophers than the property that
horses share with donkeys. This is because the average taxonomic
distance of donkeys to other mammals is about the same as that of
horses, so that donkeys add little information that could not be
inferred from horses alone. For example, the distance from horses and
donkeys to cows is uniformly low, whereas the distance to mice is
uniformly high. Now, the distance from horses to cows is low, but so
is the distance from gophers to mice. Thus, information about both
horses and gophers is likely to be more directly informative about
more mammals than information about only horses and donkeys.

Whereas both Americans and Itzaj consistently show similarity and
typicality in taxonomy-based reasoning, the Itzaj do not show
diversity. However, Itzaj noncompliance with diversity-based reasoning
apparently results neither from a failure to understand the principle
of diversity nor from any problems of "computational load," such as
those which seem to affect the inability of young school children to
reason in accordance with diversity (Lûpez, Gelman, Gutheil & Smith
1992). As with the most evident divergences between American and Itzaj
performance on similarity and typicality tasks, divergence on
diversity apparently results from ecological concerns.

The diversity principle corresponds to the fundamental principle of
induction in scientific systematics: a property shared by two
organisms (or taxa) is likely shared by all organisms falling under
the smallest taxon containing the two (Warburton 1967). Thus, American
folk seem to use their biological taxonomies much as scientists do
when given unfamiliar information in order to infer what is likely in
the face of uncertainty: informed that goats and mice share a hitherto
unknown property, they are more likely to project that property to
mammals than if informed that goats and sheep do. By contrast, Itzaj
tend to use similarly structured taxonomies to search for causal
ecological explanations of why unlikely events should occur: for
example, bats may have passed on the property to goats and mice by
biting them, but a property does not need an ecological agent to be
shared by goats and sheep.

In the absence of a theory - or at least the presumption of a theory -
of causal unity underlying disparate species, there is no compelling
reason to consider a property discovered in two distant species as
biologically intrinsic or essential to both. It may make as much or
more sense to consider the counterintuitive presence of a property in
dissimilar species as the likely result of an extrinsic or
ecologically "accidental" cause. Notice that in both the American and
Itzaj cases similarly structured taxonomies provide distance metrics
over which biological induction can take place. For the Americans,
taxonomic distance generally indicates the extent to which underlying
causes are more likely to predict shared biological properties than
are surface relationships. For Itzaj, taxonomic distance offers one
indication of the extent to which ecological agents are likely to be
involved in predicting biological properties that do not conform to
surface relationships.

A priori, either stance might be correct. For example, diseases are
clearly biologically-related; however, distribution of a hitherto
unknown disease among a given animal population could well involve
epidemiological factors that depend on both inherent biological
susceptibility and ecological agency. Equally "appropriate" ecological
strategies may be used to reason about unfamiliar features of anatomy,
physiology and behavior (e.g., in regard to predators or grazers), and
even reproduction and growth (e.g., possible animal hybridizations or
plant graftings).[14]

This does not mean that Itzaj do not understand a diversity principle.
In their justifications, Itzaj clearly reject a context-free use of
the diversity-principle in favor of context-sensitive reasoning about
likely causal connections. In fact, in a series of tasks designed to
assess risk-diversification strategies (e.g., sampling productivity
from one forest plot or several) Itzaj consistently showed an
appreciation of the diversity principle in these other settings. This
suggests that although diversity may be a universal reasoning
heuristic it is not a universal aspect of folk-biological taxonomy.

More generally, what "counts" as a biological cause or property may
differ somewhat for folk, like the Itzaj, who necessarily live in
intimate awareness of their surroundings, and those, like American
folk, whose awareness is less intimate and necessary. For Itzaj,
awareness of biological causes and properties may directly relate to
ecology, whereas for most American folk the ecological ramifications
of biological causes and properties may remain obscure. Historically,
the West's development of a world-wide scientific systematics
explicitly involved disregard of ecological relationships, and of the
colors, smells, sounds, tastes and textures that constitute the most
intimate channels of Maya recognition and access to the surrounding
living world. For example, the smell of animal excrement so crucial to
Maya hunters, or the texture of bark so important to their recognition
of trees in the dark forest understory, simply have no place in a
generalized and decontextualized scientific classification.

   2.1.2.4. Science's Marginal Role for American Folk.

A good candidate for the cultural influence of theory in American folk
biology is science. Yet, the exposure of Michigan students to science
education has little apparent effect on their folk taxonomy. From a
scientific view, student taxonomies are no more accurate than those of
Itzaj. Science's influence is at best marginal. For example, science
may peripherally bear on the differences in the way Itzaj and Michigan
students' categorize bats. Itzaj deem bats to be birds (ch'iich'), not
mammals (b'a'al~che').

Like Midwesterners, Itzaj acknowledge in interviews that there is a
resemblance between bats and small rodents. Because Itzaj classify
bats with birds, they consider the resemblance to be only superficial
and not indicative of a taxonomic relationship. By contrast, Michigan
students "know" from schooling that bats are mammals. But this
knowledge can hardly be taken as evidence for the influence of
scientific theory on folk taxonomy. Despite learning that bats are
mammals, the students go on to relate bats to rats just as Itzaj might
if they did not already "know" that bats are birds. Nevertheless, from
an evolutionary standpoint bats are taxonomically no closer to rats
than to cats. The students, it seems, pay scant attention to the
deeper biological relationships science reveals. In other words, the
primary influence of science education on folk-biological knowledge
may be to fix category labels, which in turn may affect patterns of
attention and induction.

The influence of science education on folk induction may also reflect
less actual knowledge of theory than willing belief that scientific
theory supports folk taxonomy. For example, given that a skunk and
opossum share a deep biological property, Michigan students are less
likely to conclude that all mammals share the property than if it were
shared by a skunk and a coyote. From a scientific standpoint, the
students employ the right reasoning strategy (diversity-based
inference), but reach the wrong conclusion because of a faulty
taxonomy (i.e., the belief that skunks are taxonomically further from
coyotes than from opossums). Yet if told that opossums are
phylogenetically more distant from skunks than coyotes are, the
students readily revise their taxonomy to make the correct inference.
Still, it would be misleading to claim that the students then use
theory to revise their taxonomy, although a revision occurs in
accordance with scientific theory.

   2.1.3. A Failing Compromise.

With their ranked taxonomic structures and essentialist understanding
of species, it would seem that no great cognitive effort is
additionally required for the Itzaj to recursively essentialize the
higher ranks as well, and thereby avail themselves of the full
inductive power ranked taxonomies provide. But contrary to earlier
assumptions (Atran 1990), our studies show this is not the case.
Itzaj, and probably other traditional folk, do not essentialize ranks:
they do not establish causal laws at the intermediate or life-form
levels, and do not presume that higher-order taxa share the kind of
unseen causal unity that their constituent generic species do.

There seems, then, to be a sense to Itzaj "failure" in turning their
folk taxonomies into one of the most powerful inductive tools that
humans may come to possess. To adopt this tool, Itzaj would have to
suspend their primary concern with ecological and morpho-behavioral
relationships in favor of deeper, hidden properties of greater
inductive potential. But the cognitive cost would probably outweigh
the benefit (Sperber & Wilson 1986). For this potential, which science
strives to realize, is to a significant extent irrelevant, or only
indirectly relevant, to local ecological concerns.

Scientists use diversity-based reasoning to generate hypotheses about
global distributions of biological properties so that theory-driven
predictions can be tested against experience and the taxonomic order
subsequently restructured when prediction fails. By contrast, American
folk do not have the biological theories to support diversity-based
reasoning that scientists do. If they did, American folk would not
have the categories they do.

   2.2. The General-Purpose Nature of Folk Taxonomy.

These experimental results in two very different cultures - an
industrial Western society and a small-scale tropical forest society -
indicate that people across cultures organize their local flora and
flora in similarly structured taxonomies. Yet they may reason from
their taxonomies in systematically different ways. These findings,
however, do not uphold the customary distinction in anthropology and
in history and philosophy of biology, between "general-purpose"
scientific classifications that are designed to maximize inductive
potential and "special-purpose" folk-biological classifications
(Gilmour & Walters 1964, Bulmer 1970), which are driven chiefly by
"functional" (Duprè 1981), "utilitarian" (Hunn 1982) or "social"
(Ellen 1993) concerns. On the contrary, like scientific
classifications folk-biological taxonomies appear to be
"general-purpose" systems that maximize inductive potential for
indefinitely many inferences and ends. That potential, however, may be
conceived differently by a small-scale society and a scientifically
oriented community.

For scientific systematics, the goal is to maximize inductive
potential regardless of human interest. The motivating idea is to
understand nature as it is "in itself," independently of the human
observer (as far as possible). For the Itzaj, and arguably for other
small-scale societies, folk-biological taxonomy works to maximize
inductive potential relative to human interests. Here, folk-biological
taxonomy provides a well-structured but adaptable framework . It
allows people to explore the causal relevance to them - including the
ecological relevance - of the natural world, and in indefinitely many
and hitherto unforeseen ways. Maximizing the human relevance of the
local biological world - its categories and generalizable properties
(including those yet undiscovered) - does not mean assigning
predefined purposes or functional signatures to it. Instead, it
implies providing a sound conceptual infrastructure for the widest
range of human adaptation to surrounding environmental conditions,
within the limits of culturally acceptable behavior and understanding.

For scientific systematics, folk biology may represent a ladder to be
discarded after it has been climbed, or at least set aside while
scientists surf the cosmos. But those who lack traditional folk
knowledge, or implicit appreciation of it, may be left in the crack
between science and common sense. For an increasingly urbanized and
formally educated people, who are often unwittingly ruinous of the
environment, no amount of cosmically valid scientific reasoning skill
may be able to compensate the local loss of ecological awareness upon
which human survival may ultimately depend.

3. Science and Common Sense in Systematic Biology

The scenario that I have explored so far comes to this: Some areas of
culture in general, as well as particular scientific fields, are based
in specific cognitive domains that are universal to human
understanding of nature. Concern with elaborating this basis produces
recurrent themes across cultures (e.g., totemism), and its evaluation
constitutes much of the initial phases in the development of a science
(e.g., natural history). The next sections take a closer look at later
phases in the development of systematic biology, where knowledge of
the world comes to transcend the bounds of sense without, however,
completely losing sight.

The experimental evidence reviewed in the previous sections suggests
that people in small-scale, traditional societies do not spontaneously
extend assumptions of an underlying essential nature to taxa at ranks
higher than the generic species. Thus, to infer that a biological
property found in a pair of organisms belonging to two very different
looking species (e.g., a chicken and an eagle) likely belongs to all
organisms in the lowest taxon containing the pair (e.g., bird) may
require a reflective elaboration of causal principles that are not
related to behavior, morphology, or ecological proclivity in any
immediately obvious way. Only this would justify the assumption that
all organisms belonging to a taxon at a given rank share equally some
internal structure regardless of apparent differences between them.

Such predictions lead to errors as well as discoveries. This sets into
motion a "boot-strapping" reorganization of taxa and taxonomic
structure, and of the inductions that the taxonomy supports. For
example, upon discovery that bats bear and nurture their young more
like mammals than birds, it is then reasonable to exclude bats from
bird and include them with mammal. Despite the "boot-strapping"
revision of taxonomy implied here, notice how much did not change:
neither the overall structure of folk taxonomy, nor - in a crucial
sense - even the kinds involved. Bats, birds, whales, mammals and fish
did not just vanish from common sense to arise anew in science. There
was a redistribution of affiliations between antecedently perceived
kinds. What had altered was the construal of the underlying natures of
those kinds, with a redistribution of kinds and a reappraisal of
properties pertinent to reference.

Historically, taxonomy is conservative, but it can be revolutionized.
Even venerable life forms, like tree, are no longer scientifically
valid concepts because they have no genealogical unity (e.g., legumes
are variously trees, vines, bushes, etc.). The same may true of many
longstanding taxa. Phylogenetic theorists question the "reality" of
zoological life forms, such as bird and reptile, and the whole
taxonomic framework that made biology conceivable in the first place.
Thus, if birds descended from dinosaurs, and if crocodiles but not
turtles are also directly related to dinosaurs, then: crocodiles and
birds form a group that excludes turtles; or crocodiles, birds and
turtles form separate groups; or all form one group. In any event, the
traditional separation of bird and reptile is no longer tenable.

Still, even in the midst of their own radical restructuring of
taxonomy, Linnaeus and Darwin would continue to rely on popular
life-forms like tree and bird to collect and understand local species
arrangements, as do botanists and zoologists today. As for ordinary
people, and especially those who live intimately with nature, they can
ignore such ecologically salient kinds only at their peril. That is
why science cannot simply subvert common sense.

   3.1. Aristotelian Essentials.

The boot-strapping enterprise in Western science began with Aristotle,
or at least with the naturalistic tradition in Ancient Greece he
represented. His task was to unite the various foundational forms of
the world - each with their own special underlying nature" (phusis in
the implicit everyday sense) - into an overarching system of "Nature"
(phusis in an explicitly novel metaphysical sense). In practice, this
meant systematically deriving each generic species (atomon eidos) from
the causal principles uniting it to other species of its life form
(megiston genos). It also implied combining the various life forms by
"analogy" (analogian) into an integrated conception of life.
Theophrastus, Aristotle's disciple, conceived of botanical
classification in a similar way.

Aristotelian life forms are distinguished and related through
possession of analogous organs of the same essential function
(locomotion, digestion, reproduction, respiration). For example, bird
wings, quadruped feet and fish fins are analogous organs of
locomotion. The generic species of each life form are then
differentiated by degrees of "more or less" with respect to essential
organs. Thus, all birds have wings for moving about and beaks for
obtaining nutriments. But, whereas the predatory eagle is partially
diagnosed by long and narrow wings and a sharply hooked beak, the
goose - owing to its different mode of life - is partially diagnosed
by a lesser and broader wing span and flatter bill. A principled
classification of biological taxa by "division and assembly" (diaresis
and synagoge) ends when all taxa are defined, with each species
completely diagnosed with respect to every essential organ (Atran
1985b).

In the attempt to causally link up all taxa, and derive them from one
another, Aristotle took the first step in decontextualizing nature
from its ecological setting. For him, birds were not primarily
creatures that live in trees and the air, but causal complexes of
life's essential organs and functions from which generic species
derive. Life forms become causal way stations in the essential
processes that link the animal and plant kingdoms to generic species.
As a result, all higher ranks are now essentialized on a par with
generic species, and the principle of taxonomic diversity becomes the
basis for causal inference in systematics: any biological property
that can be presumed to be related to life's essential organs and
functions, if shared by two generic species, can be expected to be
shared in descending degrees by all organisms in the life form
containing the two.

This first sustained scientific research program failed because it was
still primarily a local effort geared to explaining a familiar order
of things. Aristotle knew of species not present in his own familiar
environment, but he had no idea that there were orders of magnitude of
difference between what was locally apparent and what existed
worldwide. Given the (wrong) assumption that a phenomenal survey of
naturally occurring kinds was practically complete, he hoped to find a
true and consistent system of essential characters by trial and error.
He did not foresee that introduction of exotic forms would undermine
his quest for a discovery of the essential structure of all possible
kinds. But by inquiring into how the apparently diverse natures of
species may be causally related to the nature of life, Aristotle
established the theoretical program of natural history (as biology was
called before evolutionary theory).

   3.2. The Linnaean Hierarchy.

As in any folk inventory, ancient Greeks and Renaissance herbalists
contended with only 500 or 600 local species (Raven et al. 1971).
Preferred taxa often correspond to scientific species (dog, coyote,
lemon tree, orange tree). But frequently a scientific genus has only
one locally occurring species (bear, redwood), which makes species and
genus perceptually coextensive. This occurs regularly with the most
phenomenally salient organisms, including mammals and trees (for
example, in a comparative study, we found that 69% of tree genera in
both the Chicago area - 40 of 58 - and the Itzaj area of the Peten
rainforest - 158 of 229 - are monospecific, see Medin et al. in
press).

Europe's "Age of Exploration," which began during the Renaissance,
presented the explorers with a dazzling array of new species. The
emerging scientific paradigm required that these new forms be ordered
and classified within a global framework that unaided common sense
could no longer provide. This required a further decontextualizing of
nature, which the newly developed arts of block printing and engraving
allowed. In what is widely regarded as the first "true-to-nature"
herbal of the Renaissance (Brunfels 1530-1536), a keen historian of
science notes:

The plant was taken out of the water, and the roots were cleansed.
What therefore we see depicted is a water lily without water - isn't
this a bit paradoxical? All relations between the plant and its
habitat have been broken and concealed (Jacobs 1980:162).

By isolating organisms from local habitats through the sense-neutral
tones of written discourse, a global system of biological comparisons
and contrasts could develop. This meant sacrificing local "virtues" of
folk-biological knowledge, including cultural, ecological and sensory
information.

In the Post-Renaissance, decontextualization of preferred folk taxa
eventually led to their "fissioning" into species (Cesalpino 1583) and
genera (Tournefort 1694). During the initial stages of Europe's global
commercial expansion, the number of species increased an order of
magnitude. Foreign species were habitually joined to the most similar
European species, that is, to the generic type, in a "natural system."
Enlightenment naturalists, like Jungius and Linnaeus, further
separated natural history from its cognitive moorings in human
ecology, banning from botany intuitively "natural" but scientifically
"lubricious" life-forms, such as tree and grass (Linnaeus 1751, sec.
209).

A similar "fissioning" of intermediate folk groupings occurred when
the number of encountered species increased another order of
magnitude, and a "natural method" for organizing plants and animals
into families (Adanson 1763) and orders (Lamarck 1809) emerged as the
basis of modern systematics. Looking to other environments to complete
local gaps at the intermediate level, naturalists sought to discern a
worldwide series that would cover all environments and again reduce
the ever-increasing number of discovered species to a mnemonically
manageable set - this time to a set of basic, family plans.
Higher-order vertebrate life forms were left to provide the initial
framework for biological classes, which only phylogenetic theory would
call into question.

A concept of phylum became distinguished once it was realized that
there is less internal differentiation between all the vertebrate life
forms taken as a whole, than there is within most intermediate
groupings of the phenomenally "residual" life form, insect (bugs,
worms, etc.). This was due to Cuvier (1829), who first reduced
vertebrates to a single "branch" (embranchement). Finally, climbing
the modified ranks of folk biology to survey the diversity of life,
Darwin was able to show how the whole ordering of species could be
transformed into the tree of life - a single emerging Nature governed
by the causal principles of natural selection.

   3.3.Folk Biology's Enduring Embrace.

From Linnaeus to the present day, biological systematics has used
explicit principles and organizing criteria that traditional folk
might consider secondary or might not consider at all (e.g., the
geometrical composition of a plant's flower and fruit structure, or
the numerical breakdown of an animal's blood chemistry). Nevertheless,
as with Linnaeus, the modern systematist initially depends implicitly,
and crucially, on a traditional folk appreciation. As Bartlett
(1936:5) noted with specific reference to the Maya region of Peten
(cf. Diamond 1966 for zoology):

A botanist working in a new tropical area is... confronted with a
multitude of species which are not only new to him, but which flower
and fruit only at some other season than that of his visit, or perhaps
so sporadically that he can hardly hope to find them fertile.
Furthermore, just such plants are likely to be character plants of
[ecological] associations.... [C]onfronted with such a situation, the
botanist will find that his difficulties vanish as if by magic if he
undertakes to learn the flora as the natives know it, using their
plant names, their criteria for identification (which frequently
neglect the fruiting parts entirely), and their terms for habitats and
types of land.

As Linnaeus needed the life form tree and its commons species to
actually do his work, so did Darwin need the life form bird and its
common species. From a strictly cosmic viewpoint, the title of his
great work, On the Origins of Species, is ironic and misleading - much
as if Copernicus had entitled his attack on the geocentric universe,
On the Origins of Sunrise. Of course, in order to attain that cosmic
understanding, Darwin could no more dispense with thinking about
"common species" than Copernicus could avoid thinking about the
sunrise (Wallace 1901:1-2). In fact, not just species, but all levels
of universal folk taxonomy served as indispensable landmarks for
Darwin's awareness of the evolving pathways of diversity: from the
folk-specifics and varietals whose variation humans had learned to
manipulate, to intermediate-level families, and life-form classes,
such as bird, within which the godlier processes of natural selection
might be discerned:

[In the Galapagos Islands] There are twenty-six land birds; of these
twenty-one or perhaps twenty-three are ranked a distinct species, and
would commonly be assumed to have been here created; yet the close
[family] affinity of most of these birds to American species is
manifest in every character, in their habits, gestures, and tones of
voice. So it is with other animals, and with a large proportion of
plants.... Facts such as these, admit of no sort of explanation on the
ordinary view of creation. (Darwin 1872/1883:353-354).

Use of taxonomic hierarchies in systematics today reveals a similar
point. By tabulating the ranges of extant and extinct genera,
families, classes and so on, systematists can provide a usable
compendium of changing diversity throughout the history of life. For
example, by looking at just numbers of families, it is possible to
ascertain that insects form a more diverse group than tetrapods (i.e,
terrestrial vertebrates, including amphibians, birds, mammals and
reptiles). By calculating whether or not the taxonomic diversity in
one group varies over time as a function of the taxonomic diversity in
another group, evidence can be garnered for or against the
evolutionary interdependence of the two groups. Recent comparisons of
the relative numbers of families of insects and flowering plants,
reveal the surprising fact that insects were just as taxonomically
diverse before the emergence of flowering plants as after.
Consequently, evolutionary effects of plant evolution on the adaptive
radiation of insects are probably less profound than previously
thought (Labandeira & Sepkoski 1993). The heuristic value of
(scientifically elaborated) folk-based strategies for cosmic inquiry
is compelling, despite evolutionary theorists being well aware that no
"true" distinctions exist between various taxonomic levels.

Not only do taxonomic structure and species continue to agitate
science - for better or worse - but also the nonintentional and
nonmechanical causal processes that people across the world assume to
underlie the biological world. Vitalism is the folk belief that
biological kinds - and their maintaining parts, properties and
processes - are teleological, and hence not reducible to the
contingent relations that govern inert matter. Its cultural expression
varies (cf. Hatano & Inagaki 1994). Within any given culture people
may have varying interpretations and degrees of attachment to this
belief: some who are religiously inclined may think that a "spiritual"
essence determines biological causality; others of a more scientific
temperament might hold that systems of laws which suffice for physics
and chemistry do not necessarily suffice for biology. Many, if not
most, working biologists (including cognitive scientists) implicitly
retain at least a minimal commitment to vitalism: they acknowledge
that physico-chemical laws should suffice for biology, but suppose
that such laws are not adequate in their current form, and must be
enriched by further laws whose predicates are different from those of
inert physics and chemistry.

It is not evident how a complete elimination of teleological
expressions (concepts defined functionally) from biological theory can
be pursued without forsaking a powerful and fruitful conceptual scheme
for physiology, morphology, disease and evolution. In cognitive
science, a belief that biological systems, such as the mind/brain, are
not wholly reducible to electronic circuitry, like computers, is a
pervasive attitude that implicitly drives considerable polemic, but
also much creative theorizing. Even if this sort of vitalism
represents a lingering folk belief that science may ultimately seek to
discard, it remains an important and perhaps indispensable cognitive
heuristic for regulating scientific inquiry.

   3.4. Are there Folk Theories of Natural Kinds?

So far the line of argument has been that systematic biology and
commonsense folk biology continue to share core-related concepts, such
as the species, taxonomic ranking and teleological causality. Granted,
in science these are used more as heuristics than as ontological
concepts, but their use allows and fosters varied and pervasive
interactions between science and common sense. Still, systematic
biology and folk biology are arguably distinct domains, which are
delimited by different criteria of relevance.

This cognitive division of labor between science and common sense is
not a view favored in current philosophy or psychology (see Duprè 1993
for an exception). More frequent is the view that in matters of
biological systematics, science is continuous with folk biology; only,
science involves a more adequate elaboration of implicit folk meanings
and "theories." Deciding the issue is not so simple - in part because,
as Bertrand Russell lamented: "One of the most difficult matters in
all of controversy is to distinguish disputes about words from
disputes about facts" (1958:114).

Philosophers and psychologists have noted that no principled
distinction between folk and scientific knowledge can be built on
ideas of empirical refutation or confirmation, under-determination or
going beyond appearance or the information given, or even toleration
of internal contradictions and inconsistencies (Kuhn 1962, Feyerabend
1975, Keil & Silberstein 1996). Instead, I want to focus on three
related differences between science and folk systems: integration,
effectiveness and competition. Concerning integration, it does appear
that across all cultures there is some attempt at causal coordination
of a few central aspects of life: bodily functioning and maturational
growth, inheritance and reproduction, disease and death. But the
actual extent of this integration, and the concrete causal mechanisms
that effect it, vary widely in detail and coherency across cultures
(and individuals, judging by informant justifications in the
experimental tasks discussed in the last section).

Although the core concept of a generic species as a teleological agent
may be universal, knowledge of the actual causal chains that linkup
the life properties of a species can involve a host of vitalistic,
mechanical and intentional causes whose mix is largely determined by
social tradition and individual learning experience (e.g., on disease,
see Keil 1994 and Au & Romo 1996 for Americans, and Berlin & Berlin
1996 for Maya). Moreover, few, if any, commonsense accounts of "life"
seek to provide a causal account of the global relationships linking
(e.g., generating) species and groups of species to and from one
another, although there may be various recurrent causal clusters and
family relationships. Aristotle was possibly the first person in the
world to attempt to integrate an entire taxonomic system.[15]

Concerning effectiveness, science's aim is ultimately cosmic in that
it is geared to generating predictions about events that are equally
accurate, correct or true for any observer. By contrast basic
commonsense knowledge, driven by the folk core, has a more terrestrial
aim: namely, to provide an effective understanding of the environment
that allows appropriate responses. From an evolutionary standpoint,
the structure from which we infer an agent's environment must also be
the one that actively determines the agent's behavioral strategies
(congruent actions and responses): "if the resulting actions
anticipate useful future consequences, the agent has an effective
internal model; otherwise it has an ineffective one" that may lead it
to die out (Holland 1995:34). Folk-biological taxonomies provide both
the built-in constraints and flexibility adequate for a wide range of
culturally appropriate responses to various environments. By contrast,
scientific taxonomies are of limited value in everyday life, and some
of the knowledge they elicit (e.g., that tree, bird, sparrow and worm
are not valid taxa) may be inappropriate to a wide range of a person's
life circumstances.

Concerning competition among theories, even in our own culture such
competition only marginally affects the folk-biological core
(Dupreegrave; 1981, Atran 1987b). A tendency towards cultural
conservatism and convergence in folk biology may be a naturally
selected aspect of the functioning of the folk-biology module. As in
the case of language, the syntactic structure is geared to generate
fairly rapid and comprehensive semantic agreement, which would likely
have been crucial to group survival (Pinker & Bloom 1990).[16]
Fundamental conflicts over the meaning or extension of tree, lion and
deer would hardly have encouraged cooperative subsistence behavior.

All scientific theories may be characterized, in principle, in
relation to their competition with other theories (Popper 1972,
Lakatos 1978, Hull 1988). An intended goal of this competition is to
expand the database through better organizing principles. This is the
minimum condition for the accumulation of knowledge that distinguishes
science as a Western tradition from other cultural traditions. For
example, it is only in Europe that a cumulative development of
naturally history occurred that could lead to anything like a science
of biology. Thus, the Chinese, Ottoman, Inca and Aztec empires spanned
many local folk-biological systems. Unlike Europe, however, these
empires never managed to unite the species of different
folk-biological systems into a single classification scheme, much less
into anything like a unified causal framework (Atran 1990).

Finally, consider that a penchant for calling intuitive
data-organizing principles "theories" may stem, in part, from a
peculiar bias in analytic philosophy and cognitive psychology. This
bias consists in using the emergence of scientific knowledge as the
standard by which to evaluate the formation of ordinary knowledge
about the everyday world. From an anthropological vantage, this is
peculiar because it takes as a model of human thought a rather small,
specialized and marginal subset of contemporary thought. It is rather
like taking the peculiar knowledge system of another cultural
tradition, such as Maya cosmography, and using this to model human
thought in general.

This bias to model human cognition on scientific thought is
historically rooted in the tradition of Anglo-American empiricism,
which maintains that science is continuous with common sense, both
ontologically (Russell 1948) and methodologically (Quine 1969). It is
supposedly a natural and more perfect extension of common sense that
purges the latter of its egocentric and contextual biases: for, "it is
the essence of a scientific account of the world to reduce to a
minimum the egocentric bias in [an everyday] assertion" (Russell
1957:386). When faced with a choice between commonsense kinds and
scientific kinds whose referents substantially overlap, people ought
to pick the scientific kind; for, "we should not treat scientists'
criteria as governing a word which has different
application-conditions from the 'ordinary' word" (Putnam 1986:498; cf.
Kripke 1972:315).

The belief that folk taxonomies are approximations to scientific
classifications confounds two appropriate empirical observations and
one inappropriate metaphysical supposition. The observations are that:
the terms for commonsense generic species and the species terms used
in science are often the same; and scientific classification did
initially stem from commonsense classification. The erroneous
supposition is that both terms denote "natural kinds," and that people
will refine their use of natural-kind terms as science improves
because this is an inherent part of understanding what they "mean."
This entails that there is no a priori mental ("syntactic") constraint
on our use or understanding of biological kinds. There is only a
semantic understanding that is determined a posteriori by scientific
discoveries about the correct or true structure of the world. In fact,
neither the terms for generic species nor the species terms used in
science denote natural kinds. Consider:

Mill (1843), who was one of Russell's mentors, introduced the notion
of natural kind in the philosophy of science. Natural kinds were to be
nature's own "limited varieties," and would correspond to the
predicates of scientific laws in what was then thought to be a
determinate Newtonian universe. Counted among the fundamental
ontological kinds of this universe were biological species and the
basic elements of inert substance (e.g., gold, lead).[17]

In evolutionary theory, however, species are not natural kinds.
"Speciation," that is the splitting over time of more or less
reproductively isolated groups, has no fixed beginning and can only be
judged to have occurred to some degree through hindsight. No hard and
fast rule can distinguish a variety or genus from a species in time,
although failure to interbreed is a good rule of thumb for
distinguishing (some) groups of organisms living in close proximity.
No laws of molecular or genetic biology consistently apply to all and
only species. Nor is there evidence for a systematic deferral to
science in matters of everyday biological kinds. This is because the
relevance of biological kinds to folk in everyday life pertains to
their role in making the everyday world comprehensible, not in making
the cosmos at large transparent. When folk assimilate some rather
superficial scientific refinements to gain a bit of new knowledge
(e.g., whales and bats), these usually affect the antecedent folk
system only at the margins.

In sum, a "scientific" notion of the species as a natural kind is not
the ultimate reference for the commonsense meaning of living kind
terms. There is marked discontinuity between evolutionary and
preevolutionary conceptions of species. Indeed, the correct scenario
might be just the reverse. A notion of the species as a natural kind
lingers in the philosophy of science and resolutely persists in
psychology (Schwartz 1979, Rey 1983, Carey 1985, Gelman 1988, Keil
1995), which indicates that certain basic notions in science are as
much hostage to the dictates of common sense as the other way around.
So, to the questions - "what, if not natural kinds, are generic
species?" and "what, if not a theory, are the principles of folk
biology ?" - the answer may be simply "they are what they are." This
is a good prospect for empirical research

CONCLUSION

The uniform structure of taxonomic knowledge, under diverse
socio-cultural learning conditions, arguably results from
domain-specific cognitive processes that are panhuman, although
circumstances trigger and condition the stable structure acquired. No
other cognitive domain is invariably partitioned into foundational
kinds that are so patently clear and distinct. Neither does any other
domain so systematically involve a further ranking of kinds into
inductively sound taxonomies, which express natural relationships that
support indefinitely many inferences.

Although accounts of actual causal mechanisms and relations among taxa
vary across cultures, abstract taxonomic structure is universal and
actual taxonomies are often recognizably ancient and stable. This
suggests that such taxonomies are products of an autonomous, natural
classification scheme of the human mind, which does not depend
directly on an elaborated formal or folk theory. Such taxonomies
plausibly represent "modular" habits of the mind, naturally selected
to capture recurrent habits of the world relevant to hominid survival
in ancestral environments. Once emitted in a cultural environment, the
ideas developed within this universal framework spread rapidly and
enduringly through a population of minds without institutionalized
instruction. They tend to be inordinately stable within a culture, and
remain by and large structurally isomorphic across cultures.

Within this universal framework people develop more variable and
specific causal schema for knowing taxa and linking them together.
This enables people to interpret and anticipate future events in their
environments in locally relevant ways. To be sure, there are universal
presumptions that species-like kinds have underlying causal natures,
and this drives learning. As a result, people across the world
teleologically relate observable morphology, internally directed
growth and transgenerational inheritance to developing ideas about the
causal constitution of generic species. But no culturally elaborated
theory of life's integral properties need causally unite and
differentiate all such kinds by systematic degrees.

Thus, it is not the cultural elaboration of a theory of biological
causality that originally distinguishes people's understanding of the
species concept, taxonomy and teleology, as these apply to (nonhuman)
animals and plants from understanding basic concepts and organization
of inert substances, artifacts or persons. Rather, the spontaneous
arrangement of living things into taxonomies of essential kinds
constitutes a prior set of constraints on any and all possible
theories about the causal relations between living kinds and their
biological properties. This includes evolutionary theories, such as
Darwin's, which ultimately counter this commonsense conception.

From a scientific standpoint, folk-biological concepts such as the
generic species are woefully inadequate for capturing the evolutionary
relationships of species over vast dimensions of time and space -
dimensions that human minds were not directly designed (naturally
selected) to comprehend. All taxa are but individual segments of a
genealogical tree (Ghiselin 1981), whose branchings may never be
clearcut. Only by laborious cultural strategies like those involved in
science can minds accumulate the knowledge to transcend the bounds of
their phenomenal world and grasp nature's subtleties. But this
requires continued access to the intuitive categories that anchor
speculation and allow more sophisticated knowledge to emerge, much as
the universal intuition of solid bodies and contingent movement has
anchored scientific speculation about mass, matter and motion.

This does not mean that folk taxonomy is more or less preferable to
the inferential understanding that links and perhaps ultimately
dissolves taxa into biological theories. This "commonsense" biology
may just have different conditions of relevance than scientific
biology: the one, providing enough built-in structural constraint and
flexibility to allow individuals and cultures to maximize inductive
potential relative to the widest possible range of everyday human
interests in the biological world; and the other, providing new and
various ways of transcending those interests in order to infer the
structure of nature in itself, or at least a nature where humans are
only incidental. Because common sense operates unaware of its limits,
whereas science evolves in different directions and at different rates
to surpass those limits, the boundary between them is not apparent. A
research task of "the anthropology of science" is to comprehend this
division of cognitive labor between science and common sense: to find
the bounds within which reality meets the eye, and to show us where
visibility no longer holds the promise of truth.

NOTES

REFERENCES

Adanson, M. (1763) Familles des plantes, 2 vols. Paris: Vincent.

Anderson, J. (1990) The adaptive character of thought. Hillsdale NJ:
Erlbaum.

Atran, S. (1983) Covert fragmenta and the origins of the botanical
family. Man 18:51-71.

Atran, S. (1985a) The nature of folk-botanical life forms. American
Anthropologist 87:298-315.

Atran, S. (1985b) Pretheoretical aspects of Aristotelian definition
and classification of animals. Studies in History and Philosophy of
Science 16:113-163.

Atran, S. (1987a) Origins of the species and genus concepts. Journal
of the History of Biology 20:195-279.

Atran, S. (1987b) Constraints on the ordinary semantics of living
kinds. Mind and Language 2:27-63.

Atran, S. (1990) Cognitive foundations of natural history: Towards an
anthropology of science. Cambridge, England: Cambridge University
Press.

Atran, S. (1993) Itza Maya tropical agro-forestry. Current
Anthropology 34:633-700.

Atran, S. (1994) Core domains versus scientific theories. In L.
Hirschfeld & S. Gelman (Eds.), Mapping the mind: Domain-specificity in
cognition and culture. NY: Cambridge University Press.

Atran, S. (1995) Classifying nature across cultures. In D. Osherson &
E. Smith (Eds.), Invitation to cognitive science, vol. 3: Thinking.
Cambridge MA: MIT.

Atran, S. (in press) Itzaj Maya folk-biological taxonomy. In D. Medin
& S. Atran (Eds.), Folk biology. Cambridge MA: MIT Press.

Atran, S., Estin, P., Coley, J. & Medin, D. (in press) Generic species
and basic levels: Essence and appearance in folk biology. Journal of
Ethnobiology.

Atran, S. & Medin, D. (1997) Knowledge and action: Cultural models of
nature and resource management in Mesoamerica. In M. Bazerman, D.
Messick, A. Tinbrunsel & K. Wayde-Benzoni (Eds.), Environment, ethics,
and behavior. San Francisco: Jossey-Bass.

Atran, S. & Sperber, D. (1991) Learning without teaching: Its place in
culture. In L. Tolchinsky-Landsmann (Ed.), Culture, schooling and
psychological development. Norwood NJ: Ablex.

Au, T. & Romo, L. (1996) Building a coherent conception of HIV
transmission. In D. Medin (Ed.), The psychology of learning and
motivation, vol. 35. NY: Academic Press.

Bartlett, H. (1936) A method of procedure for field work in tropical
American phytogeography based on a botanical reconnaissance in parts
of British Honduras and the Peten forest of Guatemala. Botany of the
Maya Area, Miscellaneous Papers I. Washington DC: Carnegie Institution
of Washington Publication 461.

Bartlett, H. (1940) History of the generic concept in botany. Bulletin
of the Torrey Botanical Club 47:319-362.

Berlin, B. (1972) Speculations on the growth of ethnobotanical
nomenclature. Language and Society 1:63-98.

Berlin, B. (1992) Ethnobiological classification. Princeton: Princeton
University.

Berlin, B. (in press) One Maya Indian's view of the plant world. In D.
Medin & S. Atran, Folk biology. Cambridge MA: MIT Press.

Berlin, B., Breedlove, D., & Raven, P. (1973) General principles of
classification and nomenclature in folk biology. American
Anthropologist 74:214-242.

Berlin, B., Breedlove, D., & Raven, P. (1974) Principles of Tzeltal
plant classification. NY: Academic Press.

Berlin, E. & Berlin, B. (1996) Medical ethnobiology of the Highland
Maya of Chiapas, Mexico. Princeton: Princeton University Press.

Bock, W. (1973) Philosophical foundations of classical evolutionary
taxonomy. Systematic Zoology 22:275-392.

Boster, J. (1988) Natural sources of internal category structure.
Memory & Cognition 16:258-270.

Boster, J. (1991) The information economy model applied to biological
similarity judgment. In L. Resnick, J. Levine & S. Teasley (Eds.),
PERSPECTIVES ON SOCIALLY SHARED COGNITION. Washington, DC: American
Psychological Association.

Boster, J.; Berlin, B.; & O'Neill, J. (1986) Natural and human sources
of cross-cultural agreement in ornithological classification. AMERICAN
ANTHROPOLOGIST 88:569-583.

Boyer, P. (1994) The naturalness of religious ideas. Berkeley:
University of California Press.

Brown, C. (1984) Language and living things: Uniformities in folk
classification and naming. New Brunswick NJ: Rutgers University Press.

Bulmer, R. (1970) Which came first, the chicken or the egg-head? In J.
Pouillon & P. Maranda (Eds.), Echanges et communications: mèlanges
offerts [daggerdbl] Claude Lèvi-Strauss. The Hague: Mouton.

Bunn, H. (1983) Evidence on the diet and subsistence patterns of
Plio-Pleistocene hominids at Koobi Fora, Kenya, and at Olduvai Gorge,
Tanzania. In J. Clutton-Brock & C. Grigson (Eds.), Animals and
archaeology. London: British Arcaeological Reports.

Cain, A. (1956) The genus in evolutionary taxonomy. Systematic Zoology
5:97-109.

Carey, S. (1985) Conceptual change in childhood. Cambridge MA: MIT
Press.

Carey, S. (1996) Cognitive domains as modes of thought. In D. Olson &
N. Torrance (Eds.), Modes of thought. NY: Cambridge University Press.

Cesalpino, A. (1583) De plantis libri XVI. Florence: Marescot.

Coley, J.; Lynch, E.; Proffitt, J.; Medin, D.; & Atran, S. (in press)
Inductive reasoning in folk-biological thought. In D. Medin & S. Atran
(Eds.), Folk biology. Cambridge MA: MIT Press.

Coley, J., Medin, D. & Atran, S. (in press) Does rank have its
privilege? Inductive inferences in folkbiological taxonomies.
Cognition.

Cosmides, L. & Tooby, J. (1989) Evolutionary psychology and the
generation of culture, part II. Ethology and Sociobiology 10:51-97.

Cuvier, G. (1829) Le rËgne animal, 2nd ed., vol. 1. Paris: Dèterville.

Darwin, C. (1883) On the origins of species by means of natural
selection, 6th ed. NY: Appleton (originally published 1872)

Davidson, D. (1984) On the very idea of a conceptual scheme. In
Inquiries into truth and interpretation. Oxford: Clarendon Press.

Dawkins, R. (1976) The selfish gene. Oxford: Oxford University Press.

Dennett, D. (1995) Darwin's dangerous idea. NY: Simon & Schuster.

Diamond, J. (1966) Zoological classification of a primitive people.
Science 151:1102-1104.

diSessa, A. (1988) Knowledge in pieces. In G. Forman & P. Pufall
(Eds.), Constructivism in the computer age. Hillsdale, NJ: Erlbaum.

diSessa, A. (1996) What do "just plain folks" know about physics? In
D. Olson & N. Torrance (Eds.), The handbook of education and human
development. Oxford: Blackwell.

Diver, C. (1940) The problem of closely related species living in the
same area. In J. Huxley (Ed.), The new systematics. Oxford: Clarendon
Press.

Donnellan, K. (1971) Necessity and criteria. In J. Rosenberg & C.
Travis (Eds.), Readings in the philosophy of language.
Englewood-Cliffs NJ: Prentice-Hall.

Dougherty, J, (1978) Salience and relativity in classification.
American Ethnologist 5:66-80.

Dougherty, J. (1979) Learning names for plants and plants for names.
Anthropological Linguistics 21:298-315.

Duprè, J. (1981) Natural kinds and biological taxa. The Philosophical
Review 90:66-90.

Duprè, J. (1993) The disorder of things. Cambridge MA: Harvard
University Press.

Durham, W. (1991) Coevolution: Genes, culture and human diversity.
Stanford: Stanford University Press.

Dwyer, P. (1976) An analysis of Rofaifo mammal taxonomy. American
Ethnologist 3:425-445.

Feyerabend, P. (1975) Against method. London: New Left Review.

Ellen, R. (1993) The cultural relations of classification. Cambridge:
Cambridge University Press.

Fodor, J. (1983) Modularity of mind. Cambridge MA: MIT Press.

Gelman, R. (1990) First principles organize attention to and learning
about relevant data: Number and the animate-inanimate distinction.
Cognitive Science 14:79-106.

Gelman, S. (1988) The development of induction within natural kind and
artifact categories. Cognitive Psychology 20:65-95.

Gelman, S., Coley, J. & Gottfried, G. (1994) Essentialist beliefs in
children. In L. Hirschfeld & S. Gelman (Eds.), Mapping the mind. NY:
Cambridge University.

Gelman, S. & Wellman, H. (1991) Insides and essences. Cognition
38:214-244.

Gilmour, J. & Walters, S. (1964) Philosophy and classification. In W.
Turrill (Ed.), Vistas in botany, vol. 4: Recent researches in plant
taxonomy. Oxford: Pergamon Press.

Ghiselin, M. (1981) Categories, life, and thinking. Behavioral and
Brain Sciences 4:269-313.

Gigerenzer, G. (in press) The modularity of social intelligence. In A.
Whiten & R. Byrne (Eds.), Machiavellian intelligence II. Cambridge:
Cambridge University Press.

Greene, E. (1983) Landmarks in botany, 2 vol. Stanford: Stanford
University.

Hall, D.G. & Waxman, S. (1993) Assumptions about word meaning:
Individuation and basic-level kinds. Child Development 64:1550-1570.

Hatano, G. & Inagaki, K. (1994) Young children's naive theory of
biology. Cognition 50:171-188.

Hatano, G. & Inagaki, K. (1996) Cognitive and cultural factors in the
acquisition of intuitive biology. In D. Olson & N. Torrance (Eds.),
The handbook of education and human development. Oxford: Blackwell.

Henley, N. (1969) A psychological study of the semantics of animal
terms. Journal of Verbal Learning and Verbal Behavior 8:176-184.

Hickling, A. & Gelman, S. (1995) How does your garden grow? Evidence
of an early conception of plants as biological kinds. Child
Development 66:856-876.

Hull, D. (1988) Science as a process. Chicago: University of Chicago
Press.

Hunn, E. (1976) Toward a perceptual model of folk biological
classification. American Ethnologist 3:508-524.

Hunn, E. (1982) The utilitarian factor in folk biological
clasification. American Anthropologist 84:830-847.

Isaac, G. (1983) Aspects of human evolution. In D. Bendall (Ed.),
Evolution from molecules to men. NY: Cambridge University Press.

Jacobs, M. (1980) Revolutions in plant descriptions. In J. Arends, G.
Boelema, C. de Groot & A. Leeuwenberg (Eds.), Liber gratulatorius in
honorem H.C.D. De Wit. Wageningen: H. Veenman & Zonen.

Keil, F. (1979) Semantic and conceptual development: An ontological
perspective. Cambridge MA: Harvard University Press.

Keil, F. (1994) The birth and nurturance of concepts by domains. In L.
Hirschfeld & S. Gelman (Eds.), Mapping the mind. NY: Cambridge
University Press.

Keil, F. (1995) The growth of understandings of natural kinds. In S.
Sperber, D. Premack & A. Premack (Eds.), Causal Cognition. Oxford:
Clarendon Press.

Keil, F. & Silberstein, C. (1996) Schooling and the acquisition of
theoretical knowledge. In D. Olson & N. Torrance (Eds.), The handbook
of education and human development. Oxford: Blackwell.

Kesby, J. (1979) The Rangi classification of animals and plants. In R.
Reason & D. Ellen (Eds.), Classifications in their social contexts.
NY: Academic Press.

Kripke, S. (1972) Naming and necessity. In. D. Davidson & G. Harman
(Eds.), Semantics of natural language. Dordrecht: Reidel.

Kuhn, T. (1962) The structure of scientific revolutions. Chicago:
University of Chicago Press.

Kummer, H. (1995) Causal knowledge in animals. In S. Sperber, D.
Premack & A. Premack (Eds.), Causal Cognition. Oxford: Clarendon
Press.

Kummer, H., Daston, L., Gigerenzer, G. & Silk, J. (in press) The
social intelligence hypothesis. In P. Weingart, P. Richerson, S.
Mitchell & S. Maasen (Eds.), Human by nature. Hillsdale NJ: Erlbaum.

Labandeira, C. & Sepkoski, J. (1993) Insect diversity in the fossil
record. Science 261:310-315.

Lakatos, I. (1978) The methodology of scientific research programs.
Cambridge: Cambridge University Press.

Lamarck, J. (1809) Philosophie zoologique. Paris: Dentu.

Latour, B. (1987) Science in action. Cambridge MA: Harvard University
Press.

Lèvi-Strauss, C. (1963) The bear and the barber. The Journal of the
Royal Anthropological Institute 93:1-11.

Lèvi-Strauss, C. (1969) The elementary structures of kinship. Boston:
Beacon Press.

Linnaeus, C. (1751) Philosophia botanica. Stockholm: G. Kiesewetter.

Locke, J. (1848/1689) An essay concerning human understanding. London:
Tegg.

Lûpez, A., Atran, S., Coley, J., Medin, D., & Smith, E. (1997) The
tree of life: Universals of folk-biological taxonomies and inductions.
Cognitive Psychology 32:251-295.

Lûpez, A., Gelman, S., Gutheil, G. & Smith, E. (1992) The development
of category-based induction. Child Development 63:1070-1090.

Lumsden, C. & Wilson, E.O. (1981) Genes, mind and culture. Cambridge
MA: Harvard University Press.

Mandler, J., Bauer, P. & McDonough, L. (1991) Separating the sheep
from the goats: Differentiating global categories. Cognitive
Psychology 23:263-298.

Mayr, E. (1969) Principles of systematic zoology. NY: McGraw-Hill.

Medin, D., Lynch, E., Coley, J. & Atran, S. (1996) The basic level and
privilege in relation to goals, theories and similarity. In R.
Michalski & J. Wnek (Eds.). Proceedings of the third international
workshop on multistrategy learning. Palo Alto: American Association
for Artificial Intelligence.

Medin, D., Lynch, E., Coley, J. & Atran, S. (1997) Categorization and
reasoning among tree experts: Do all roads lead to Rome? Cognitive
Psychology 32:49-96.

Mill, J. (1843) A system of logic. London: Longmans, Green.

Millikan, R. (in press) A common structure for concepts of
individuals, stuffs, and real kinds: More mama, more milk, and more
mouse. Behavioral and Brain Sciences 21.

Osherson, D., Smith, E., Wilkie, O., Lûpez, A., & Shafir, E. (1990)
Category-based induction. Psychological Review 97:85-200.

Pinker, S. & Bloom, P. (1990) Natural language and natural selection.
Behavioral and Brain Sciences 13:707-727.

Popper, K. (1972) Objective knowledge. Oxford: Clarendon Press.

Premack, D. (1995) Foreward to Part IV: Causal understanding in naÔve
biology. In D. Sperber, D. Premack & A. Premack (Eds.), Causal
cognition: A multidisciplinary debate. Oxford: Clarendon Press.

Premack, D. & Premack, A. (1994) Moral belief: Form versus content. In
L. Hirschfeld & S. Gelman (Eds.), Mapping the mind. NY: Cambridge
University Press.

Putnam, H. (1986) Meaning holism. In L. Hahn & P. Schlipp (Eds.), The
philosophy of W.V.Quine. La Salle IL: Open Court.

Quine, W. (1969) Natural kinds. In Ontological relativity and other
essays. NY: Columbia University Press.

Raven, P., Berlin, B., & Breedlove, D. (1971) The origins of taxonomy.
Science 174:1210-

Rey, G. (1983) Concepts and stereotypes. Cognition 15:237-262.

Rips, L.; Shoben, E.; & Smith, E. (1973) Semantic distance and the
verification of semantic relations. Journal of Verbal Learning and
Verbal Behavior 12:1-20.

Romney, A.K., Weller, S., & Batchelder, W. (1986) Culture as
consensus: A theory of culture and informant accuracy. American
Anthropologist 88:313-338.

Rosch, E. (1975) Universals and cultural specifics in categorization.
In R. Brislin, S. Bochner & W. Lonner (Eds.), Cross-cultural
perspectives on learning. NY: Halstead.

Rosch, E.,, Mervis, C., Grey, W., Johnson, D., & Boyes-Braem, P.
(1976) Basic objects in natural categories. Cognitive Psychology
8:382-439.

Russell, B. (1948) Human knowledge: Its scope and limits. NY: Simon &
Schuster.

Russell, B. (1957) Mr Strawson on referring. Mind 66:385-389.

Russell, B. (1958) The ABC of relativity. London: George Allen &
Unwin.

Schwartz, S. (1978) Putnam on artifacts. Philosophical Review
87:566-574.

Schwartz, S. (1979). Natural kind terms. Cognition 7:301-15.

Simpson, G. (1961). Principles of animal taxonomy. NY: Columbia
University Press.

Spelke, E. (1990) Principles of object perception. Cognitive Science
14:29-56.

Sperber, D. (1985) Anthropology and psychology. Man 20:73-89.

Sperber, D. (1994) The modularity of thought and the epidemiology of
representations. In L. Hirschfeld & S. Gelman (Eds.), Mapping the
mind. NY: Cambridge University.

Sperber, D. (1996) La contagion des idèes. Paris: Editions Odile
Jacob.

Sperber, D. & Wilson, D. (1986) Relevance. London: Blackwell.

Stevens, P. (1994) Berlin's "Ethnobiological Classification."
Systematic Biology 43:293-295.

Stross, B. (1973) Acquisition of botanical terminology by Tzeltal
children. In M. Edmonson (Ed.) Meaning in Mayan languages. The Hague:
Mouton.

Tanaka, J. & Taylor, M. (1991) Object categories and expertise: Is the
basic level in the eye of the beholder? Cognitive Psychology
23:457-482.

Tooby, J. & Cosmides, L. (1992) The psychological foundations of
culture. In J. Barkow, L. Cosmides & J. Tooby (Eds.), The adapted
mind. NY: Oxford University Press.

Tournefort, J. (1694) Elèmens de botanique. Paris: Imprimerie Royale.

Wallace, A. (1901) Darwinism, 2nd ed. London: Macmillan. (1st ed.
1889)

Warburton, F. (1967) The purposes of classification. Systematic
Zoology 16:241-245.

Zubin, D. & Köpcke, K.-M. (1986) Gender and folk taxonomy. In C. Craig
(Ed.), Noun classes and categorization. Amsterdam: John Benjamins.

Footnotes

[1] The studies reported here were funded by NSF (SBR 93-19798,
94-22587) and France's Ministry of Research and Education (Contrat
CNRS 92-C-0758), with student support from the University of
Michigan's "Culture and Cognition" Program. They were co-directed with
Douglas Medin. Participants in this project on biological knowledge
across cultures include Alejandro Lûpez (Psychology, Max Planck), John
Coley and Elizabeth Lynch (Psychology, Northwestern U.), Ximena Lois
(Linguistics, Crea-Ecole Polytechnique), Valentina Vapnarsky
(Anthropology, Universitè de Paris X), Edward Smith and Paul Estin
(Psychology, U. Michigan), and Brian Smith (Biology, U. Texas,
Arlington). I thank Medin, Dan Sperber, Giyoo Hatano, Susan Carey,
Gerd Gigerenzer and the anonymous referees for comments; thanks also
to Estin and Lûpez for Figures.

[2] Thus, comparing constellations in the cosmologies of Ancient
China, Greece and the Aztec Empire shows little commonality. By
contrast, herbals like the Ancient Chinese ERH YA, Theophrastus's Peri
Puton Istorias, and the Aztec Badianus Codex, share important
features, such as the classification of generic species into tree and
herb life forms (Atran 1990:276).

[3] By contrast, a partitioning of artifacts (including those of
organic origin, such as foods) is neither mutually exclusive nor
composed of inherent natures: some mugs may or may not be cups; an
avocado may be a fruit or vegetable depending on how it is served; a
given object may be a bar stool or waste bin depending on the social
context or perceptual orientation of its user; and so on.

[4]4. It makes no difference whether these groups are named. English
speakers ambiguously use "animal" to refer to at least three distinct
classes of living things: nonhuman animals, animals including humans,
and mammals (the prototypical animals). The term "beast" seems to pick
out nonhuman animals in English, but is seldom used today. "Plant" is
ambiguously used to refer to the plant kingdom, or to members of that
kingdom that are not trees.

[5]5. Life forms vary across cultures. Ancient Hebrew or modern Rangi
(Tanzania) include herpetofauna (reptiles and amphibians) with
insects, worms and other "creeping crawlers" (Kesby 1979), whereas
Itzaj Maya and (until recently) most Western cultures, include
herpetofauna with mammals as "quadrupeds." Itzaj place phenomenally
isolated mammals like the bat with birds, just as Rofaifo (New Guinea)
place phenomenally isolated birds like cassowaries with mammals (Dwyer
1976). Whatever the content of life-form taxa, the life-form level, or
rank, universally partitions the living world into broadly equivalent
divisions.

[6]6. In the logical structure of folk taxonomy, outliers may be
considered monotypic life forms with only one generic species (for a
formalism, see the appendix in Atran 1995).

[7]7. Botanists and ethnobotanists tend to see preferred
folk-biological groups as akin to scientific genera (Bartlett 1940,
Berlin, 1972, Greene 1983). Plant genera especially are often groups
most easily recognized morphologically without technical aids
(Linnaeus 1751). Zoologists and ethnozoologists tend to view them as
more like scientific species, where reproductive and geographical
isolation are more readily identified in terms of behavior (Simpson
1961, Diamond 1966, Bulmer 1970).

[8] In a comparative study of Itzaj Maya and rural Michigan college
students, we found that the great majority of mammal taxa in both
cultures correspond to scientific species, and most also correspond to
monospecific genera: 30 of 40 (75%) basic Michigan mammal terms denote
biological species, of which 21 (70%, or 53% of the total) are
monospecific genera; 36 of 42 (86%) basic Itzaj mammal terms denote
biological species, of which 25 (69%, or 60% of the total) are
monospecific genera (Atran 1995, Lûpez et al. 1997). Studies of trees
in both the Peten rainforest and Chicago area reveal a similar pattern
(Atran 1993; Medin et al. 1997).

[9] Moving vertically within each graph corresponds to changing the
premise while holding the conclusion category constant. This allows us
to test another domain-general model of category-based reasoning: The
Similarity-Coverage Model (Osherson et al. 1990). According to this
model, the closer the premise category is to the conclusion category,
the stronger the induction should be. Our results show only weak
evidence for this general reasoning heuristic, which fails to account
for the various "jumps" in inductive strength that indicate absolute
or relative preference (Atran et al. in press). Note also that we
conducted separate experiments to control for the effects of
linguistic transparency; for example, whether relations between
generic species and life forms were marked (e.g., catfish - fish) or
unmarked (e.g., bass - fish) had no effect on results (Coley, Medin &
Atran in press).

[10] The existence of universal, domain-specific cognitions is not
tied exclusively, or even necessarily, to cross-cultural
pervasiveness. The social subordination of women, for example, appears
in all known cultures (i.e., it is a cultural "universal" in the sense
of Lèvi-Strauss 1969). It could even be argued that this universal has
some biological grounding. There is no reason, however, to attribute
the varied ways people process this pervasive social phenomenon to a
universal cognitive mechanism. Conversely, the ability to understand
and develop mathematics may be rooted in some fairly specific
cognitive mechanisms, with which humans are innately endowed (Gelman
1990). But if so, many cultures do not require that people use this
ability. Nor is it occasioned by every environment.

[11] Each group was tested in its native language (Itzaj and English),
and included a minimum of 6 men and 6 women on each task. The choice
of groups of 12 or more people is based on pilot studies that indicate
this is sufficient to establish a cultural consensus (Atran 1994). No
statistically significant differences between men and women were found
on the tasks reported. The method of successive pile sorts and
taxonomic comparison was pioneered by Boster and his colleagues
(Boster, Berlin & O'Neill 1986; Boster 1991).

[12] For each subject, we have a square symmetric data matrix, with
the number of rows and columns equal to the number of generic species
sorted. Subjects' taxonomic distance matrices were correlated with
each other, yielding a pairwise subject-by-subject correlation matrix
representing the degree to which each subject's taxonomy agreed with
each other subject's taxonomy. Principal component factor analyses
were then performed on the intersubject correlation matrix for each
group of informants to determine whether or not there was a "cultural
consensus" in informant responses. A cultural consensus is plausible
if the factor analysis results in a single factor solution. If a
single dimension underlies patterns of agreement within a domain, then
consensus can be assumed for that domain and the dimension can be
thought of as reflecting the degree to which each subject shares in
the consensual knowledge (Romney, Batchelder & Weller 1986). Consensus
is indicated by a strong single factor solution in which: (1) the
first latent root (eigenvalue) is large compared to the rest, (2) all
scores on the first factor are positive, and (3) the first factor
accounts for most of the variance. To the extent that some individuals
agree more often with the consensus than others, they are considered
more "culturally competent" with respect to the domain in question. An
estimate of individual knowledge levels, or competencies, is given by
each subject's first factor scores. This represents the degree to
which that subject's responses agree with the consensus. That is, the
pattern of correlations among informants should be based entirely on
the extent to which each subject knows the common (culturally
relative) "truth." The mean of all first-factor scores provides an
overall measure of consensus.

[13] Different types of "scientific taxonomy" correlate differently
with folk taxonomy, with cladistic taxonomies (based on strict
phylogentic branching) generally being the least correlated and
phenetic taxonomies (based on relations among observable characters)
being the most. Evolutionary taxonomies represent a compromise of
sorts between cladistics and phenetics.

[14] Apparent lack of taxonomically based diversity is not limited to
Itzaj reasoning about mammals (they show the same pattern when
reasoning about birds and palms, Atran in press), nor is it limited to
nonwestern populations. In another series of studies exploring the
impact of different kinds of expertise on categorization and reasoning
about trees (Medin et al. 1997), we have found that parks and forestry
maintenance workers responded significantly below chance on diversity
items (Coley, Medin, Proffitt, Lynch, Coley & Atran in press). As with
the Itzaj, justifications focused on ecological factors (e.g.
distribution, susceptibility to disease) and associated causal
reasoning. Another American group, consisting of taxonomists, sorted
and reasoned in accordance with scientific classification. These
results confirm the scientific reasoning patterns that were only
inferred from the scientific classification in the mammal studies.
Like American students on the mammal task, the taxonomists also had
overwhlemingly positive responses on the diversity task. Differences
in education did not appear to be significantly correlated with
diversity or lack of diversity in the American populations (note also
that Lopez et al. 1992 found diversity with American ten-year-olds).

[15] The situation is arguably similar for naive physics, not only
between cultures, but within our own culture. DiSessa (1988) speaks of
a "knowledge in pieces" involving concept clusters that reinforce and
help to interpret one another in order to guide people's uninstructed
expectations and explanations about many situations of potential
relevance to them. Although there is appreciable diversity of
expectations and explanations, there are strong tendencies towards the
convergence of concept clusters across individuals (and presumably
across cultures). These are fairly robust, even for people with formal
or scientific education, in part because there is substantial overlap
between scientific (Newtonian) and commonsense physics. The causal
clusters that are formed, however, reflect local family relationships
rather than global coverage: "The impetus theory is, at best, about
tosses and similar phenomena. It does not explain how people think
about objects on tables, or balance scales, or orbits" (diSessa
1996:714).

[16] There is also the cryptic notion of "tacit theory" that
originally came from Chomskian linguistics. Generative linguists
rightly seem to consider this more of a throwaway notion than do some
philosophers. Using "tacit theory" to assimilate universal grammar and
universal taxonomy would wrongly entail assimilating a core module to
an input module, and perhaps also to any complex biological algorithm
(instinct) or automatic organizing process.

[17] Aristotle first proposed that both living and inert kinds had
essential natures. Locke (1848/1689) dubbed these unknowable kinds,
"real kinds," claiming that their underlying natures could never be
wholly fathomed by the mind. Across cultures, it is not clear that
inert substances comprise a cognitive domain that is conceived in
terms of underlying essences or natures. Nor it is obvious what the
basic elements might be, since the Greek earth, air, fire and water
are not universal. The conception of "natural kind," which supposedly
spans all sorts of lawful natural phenomena, may turn out not to be a
psychologically real predicate of ordinary thinking (i.e., a "natural
kind" of cognitive science). It may be simply an epistemic notion
peculiar to a growth stage in Western science and philosophy of
science.