Cognitive Foundations of Natural History: Towards an Anthropology of Science 
by Scott Atran.
Cambridge, 360 pp., £35, August 1990, 0 521 37293 3
Show More
Show More

‘War,’ said an old peacenik poster with the words scrawled across a child’s drawing of a tree, ‘is harmful to children and other living things.’ This subtle and sophisticated book has a little of that same power to shock by innocence. It is about how children think of living things, less a matter of what they learn than of what human nature teaches about nature. That is a pleasing picture but not a popular one. Atran is not in search of opponents, but here are a few of the things he is up against ... Every people classifies in its own ways: it is cultural imperialism to expect our sortings of life to be duplicated by others (standard all-purpose cultural relativism). Aristotle was a disaster for biology: he taught that plants and animals should be defined by a priori essences that not only devalued observation but also made it impossible to attend to variations among the species, thereby impeding evolutionary thought for two millennia (textbook history of biology). The taxonomies of natural history are an artwork of the Enlightenment; the Renaissance was mired in a wonderful world of resemblance, natural magic and the doctrine of signatures, whereby plants were the signs of minerals which were the signs of stars (François Jacob, Michel Foucault). The semantics of natural kind terms is such that, when we speak of living things, we are referring, whether we know it or not, to the fundamental kinds at which science aims (Saul Kripke and, sometimes, Hilary Putnam). We have an obligation to integrate our commonsense categories into the best knowledge available; when there is conflict, common sense yields to knowledge (most good scientists, starting with Aristotle).

That is a diverse collection of foes. Aside from the concluding scientism, I don’t suppose that any one person is deeply committed to more than one of those doctrines. I can’t imagine Foucault having cared for rigid designators; the doctrine of signatures is not Kripke’s cup of tea. Do the enemies have anything in common? It is to Atran’s credit that his unified theory makes out that all these baldly summarised assertions are false.

Here’s how it goes. Children have an innate disposition to classify the plants and creatures of the ecosystem. Hence all peoples recognise pretty much the same classes of living things in their locality. We are predisposed to name about six hundred different kinds of life, and to arrange them in a simple taxonomic structure common to all societies, all languages. We come into the world expecting that ‘nature has some basic kinds.’ Classification, moreover, brings with it instant generalisations; we expect all instances of a kind to behave in much the same ways – if you’ve seen one, you’ve seen them all. Everyone acts as if ‘basic kinds have basic natures.’

As for Aristotle, the only part of his biology that was a priori was in effect innate. That is, he took for granted the local classification of six hundred or so living things suitable for his part of the Mediterranean. He wrongly thought that it was almost exhaustive. He aimed at a unified theory of Nature to make sense of his instinctive taxonomy. It would explain why things are what they are: ‘the natures of kinds are to be understood by the place of the kinds in Nature.’ (These mnemonic sentences are mine, not Atran’s.) Individuals are fulfilled only when they reproduce their kind. They should develop as fully as possible according to a template suitable for that kind. Nothing in Aristotle’s theory makes the characteristics of a species eternal or independent of habitat. If the surroundings change, the kinds would have to adapt to fulfil themselves as well as they could in altered circumstances.

Immutable species have a late Renaissance origin. Voyages had begun to bring back a lot of new plants. Cesalpino (1519-1603) had to re-systematise Aristotelian taxonomy to accommodate new kinds. Kinds, once sorted instinctively, now were so plentiful that they needed a general principle to put them in order. He fixed on the seeds, fruits or flowers as the keys to placing plants on a universal taxonomic tree. Local and exotic plants were thus arranged on a timeless scheme. The species were defined by how they reproduced, which in turn entails what Ernst Mayr in the 1960s named the ‘species concept’: a species is a set of individuals that can interbreed.

From here on Atran provides a more standard tale in which the ‘system’ begun by Cesalpino is completed by Linnaeus, and is challenged by the ‘method’ which objects that it is artificial to characterise the species by just one characteristic, fructification. Theoretical and metaphysical doctrines matter to subsequent developments, but so does the constant pressure of exploration bringing back new plants and to some extent new animals, and then new skeletons, fossils. In any one locality, six hundred species are enough to get around with. When we have tens of thousands, a new principle is needed. The genus, which in Aristotle had simply been higher than a species in a definition by division, now occupied central stage, and we moved up to about six hundred genera. As the number of genera grows out of control we need the family, the class, the order. There is no big ‘epistemological break’ marking the transition from natural history to biology. Instead, there are stabs at the increasing complexity of the known world of nature. Our built-in capacity to sort plants and animals according to how they look and move and grow becomes increasingly irrelevant. By the time of Cuvier (1769-1832) our attachment to surface phenomena was fully replaced by anatomy and the function of parts. A ‘scientific breakaway’ from common sense was in place. Darwin lay in waiting.

But Atran thinks we still want our commonsense sortings. They co-exist with scientific ones, and for many purposes are much more useful. Virtually all peoples who have seen trees group some plants as trees – that is part of our basic built-in taxonomising. Trees do not form a botanic kind at all but remain a perfectly good kind in nature. New Scientist is not promoting bad science with its current poster offer ‘The Tree’. As the ad says, ‘trees are essential to the stability of our environment and our future well-being.’ Atran has a provocative thesis about how common-sense classifications and those of systematics differ, but first we should attend to the grander themes announced in the title: does natural history want ‘cognitive foundations’? And what is an anthropology of science?

Cognitive science lies close by Atran’s side, guiding his claim that we have an innate capacity for sorting plants and animals. The cognitive revolution, as it is now called, took off from Noam Chomsky’s thesis that there is a universal grammar, somehow innate, that makes it possible for infants to catch on to any language spoken round them almost without trying. This idea, once so radical, is now celebrated as fact and as a comprehensive model for the human mind. Where we might have had a post-war electronic jungle of information-processing, artificial intelligence, cybernetics etc, Chomsky indelibly stamped the field with his rationalist and egalitarian conviction that we study the mind, something ‘species-specific’ and trans-cultural.

Trans-cultural, yes, but not monolithic. We have many competencies. Most cognitivists now hold that far from there being one grammatical machine in the head, there are a lot of distinct ‘modules’ that we bring into play in acquiring and using grammar. The same is true for perception, pattern recognition, problem-solving, generalising – lots of different innate systems. Don’t expect an all-purpose organ for classifying things and experiences in general, but rather engines designed for specifics. Atran’s thesis about a built-in cognitive skill for sorting the living beings fits nicely into this tradition. But you don’t have to recite the cognitivist catechism to gain admission to his book. You can also think he is wrong-headed in his application of the cognitivist programme, and still immensely admire his work.

He is moving ‘towards an anthropology of science’. His final chapter hints at what that might be. It is not enthnography. Atran does tell us a good deal about peoples in various parts of the world, and how they sort plants and animals. He was a pupil of Margaret Mead’s. So he is an anthropologist in a standard sense of the word. But there is an older meaning in which anthropology is humanology. It starts with the firm conviction that there are general, complex and important truths about all of us and our place in nature. Chomsky ensured that cognitive science would be part of humanology, the general theory of mental abilities specific and common to all members of our species. So Atran’s anthropology of science has nothing to do with those fascinating (and very popular) ‘anthropological’ studies of the laboratory. Nor does he describe the social arrangements or systems of meanings within which scientists thrive. He wants instead to answer a bold and possibly nonsensical question: what is it about us and our universe that makes science possible?

With so rich a roster of topics I have to pass by Atran’s insights into the history of systematics. That means ignoring half the book. I’ll retrace more slowly the path from cognitive foundations to anthropology of science. During the heyday of cultural relativisms the idea of shared classifications was distinctly ‘out’. It has been increasingly ‘in’ for the past decade or more. It leans on two kinds of support: comparative studies of languages in the ecosystems where they are used, and observations, usually made at home, of children learning from scratch to make distinctions, There are, it is urged, a set of ‘basic-level’ categories distinguished and organised by all peoples. Children catch on to them at an early age. There may he subordinate and super-ordinate classes of, for example, animals that are distinguished locally or for scientific purposes, but basic-level groupings are found around the world.

Such opinions ride well with Chomsky’s beliefs about universal human competences. They are consistent with the existence of the bizarre categorisations with which cultural anthropologists fascinate us. Basic-level categories can be used as centres for metaphorical extension and bases for myth. The way a metaphor works or the direction taken by a myth may he embedded in many layers of local meaning and practice. Just because the basic-level classes are so salient, we expect them to serve the most powerful metaphors, the predominant myths. The raven is a key bird, and also a key figure, for a people living in coastal British Columbia. Their word for ‘raven’ may denote not only birds but also youths who have had certain experiences, or a powerful intruder, or an agent in creation – but it also serves to pick out ravens, and the class of birds that it picks is co-extensive with anyone else’s ‘raven’.

Atran’s picture is as follows. Until recently, all peoples have lived in close contact with a lot of plants, animals, fishes, insects, birds, reptiles. Every society groups living beings that are roughly our size into big clumps such as insects, fish etc. Plants are classed as trees, shrubs, vines, herbs, grasses. The kinds of plants which arc singled out and given names depends on what plants are around. But when the same salient plant or what we call a close relative of it is found in two ecosystems, the peoples of both regions will each have a name for that kind, and moreover, give it the same place in their larger taxonomy (a tree is a tree, and a vine is usually a vine), At least this much structure is demanded for these thoughts to fit in with cognitive science. Innate ideas are not enough. They have to be organised (grammar is the model).

‘Nature has some basic kinds’ that people recognise. Atran’s more controversial thesis is a converse: ‘basic kinds have basic natures.’ That means that people expect all the things of a kind to be much the same in a significant number of ways. We expect what goes for one good example goes, for the most part, for all. Tasty, long-lived, runs fast, cures the runs, blooms every century, leaps for flies in the evening, burns well, mighty oaks from little acorns grow. To see something as of a kind is to be willing to jump to some conclusions. If today’s snails eat my cabbages, so will tomorrow’s snails. Atran goes one step further. He thinks we innately expect living things to have ‘underlying’ natures – essences, for short. That is certainly a handy road into Aristotle.

We are predisposed to think kinds of living beings have essences. What could that mean? Atran makes a contrast with artifacts. Artifacts do not fit into any tidy tree-like taxonomies. They are characterised by their functions, what we made them for. They don’t have natures. For example, there is nothing that you can expect of a comb, insofar as it is a comb, except that you use it to comb your hair. You can have expectations of it as a discrete ‘thing’, as a material object, or as made of plastic, but not as a comb. Atran rejects experimental results claiming to show that children make no ‘essential’ distinction between artifacts and plants or animals. Good: like much developmental psychology, the experiments seem confused and the interpretations are dubious. Let’s suppose that children do act as if basic kinds of living things have natures, and that they don’t react to artifacts the same way. How could that show that we innately expect living beings to have ‘underlying’ natures?

Philosophers coined the term ‘natural kind’. Those who think that the idea is useful tend to have one of two accounts of natural kinds: things of a kind have an inner constitution which makes them the way they are; or things of a kind have certain laws of nature that are true about them, and true of no other kind. Atran thinks we are innately philosophers of the first kind. I cannot imagine grounds for this Opinion, because both views are too subtle to attribute to anyone except philosophers. By the age of six children spontaneously discuss many of the problems of philosophy (the origin of the universe, free will etc), but the very idea of a natural kind takes a lot of teaching. It derives, possibly, from innate dispositions to recognise some things as ‘the same’ and others as different – and such dispositions are no different from our inclination to jump to conclusions about some kinds of thing, and not about others. But more than sameness – an inherited theory about sameness – is too much to expect of the science of genetics.

I regret my scepticism here, for I very much like Atran’s idea that common-sense kinds are different from those of systematics. It is all very well to notice that we cling to ‘tree’ even though there is no such botanical sort. That is not a difference in principle but in convenience. Atran’s stronger claim is that commonsense kinds are committed to ‘underlying’ natures, while the kinds used in systematics have given up on fundamental internal constitutions (species don’t have unique genetic codes), and instead rely on law-like connections between kinds and changes in kinds.

Compared to the claim about the innateness of essence, the alleged innateness of sorting and of elementary taxonomy is dull, almost empty. I recall a published discussion of the Sixties between Chomsky and the nominalist and sceptical philosopher Nelson Goodman. Goodman found nothing in favour of innatism: we need notice nothing more than the remarkable ability of children for ‘groping and grasping’. That’s how I feel here. Our regional species are distinct, and trees are different in many ways from vines: does a child in a region need more than that to form expectations about plants? Yes, it needs to see that they are plants, to know what kinds of expectations are germane. But why can’t that just be, as Mill urged long ago, a higher-level induction?

It is part of Atran’s position that living things are in a class of their own. So they are. He rightly contrasts artifacts. His examples are a bit clumsy and bulky – cars, robots, wheelchairs, tables. What about leather and linen? There are quite a few things you can expect of linen because it is linen, because it is of that kind. It fits into a small taxonomy, linen being a kind of cloth. It has not so rich a texture as the flax plant, but we may have more of a continuum than Atran’s slender diet of examples suggests. He barely attends to the minerals, whose classificatory scheme is not universal. What of geographical features, headlands and valleys and streams? Thunder, lightning, rain? I bet those are just as universally noticed as kinds of plant; do we need a climate-kind-recognition-module in the head to explain this? He mentions parts of the body, such as wings, but not flesh and blood and bone. He winces a bit when experimenters confront children with agricultural quasi-artifacts like apples, but does not pause to consider the domesticated plants and animals.

I sense, in skimming through Atran’s examples, that he is fascinated by peoples in the tropics who cultivate little but have an extraordinary pharmacopoeia. It is their romantic kinds that we are to contrast with machine-made artifacts. I suddenly found myself wanting a treatise on natural kinds written by a feminist. There is a persuasive school of pre-history in the making, according to which early societies were so structured that women created agriculture, being able to tell which plants could be domesticated and turned into food. That might start us thinking about the kinds of classification that are peculiarly human, and force us to remember that it is domestic plants and animals that so many peoples think of us natural kinds. I’ve assiduously avoided the term Atran most uses: he calls our ‘innate’ classification system ‘folk-biology’. I don’t care for the word ‘folk’. But maybe there is a place for it. We seem to be told, in this book, about the classifications favoured by menfolk, and we should hear a little more from the womenfolk.

For me, the appeal of innatism is just where most people would find it most improbable: Atran’s vision of an anthropology of science. He expresses no qualms about science, the theoretical and experimental investigation and organisation of very general facts. Our findings are discoveries about how the world is (no cultural relativist claptrap for Atran). We have been very good at finding out ever so many things, and bid fair to go on doing so. How can we do it? There might be no answer, or there might be only little bits of answers about this or that. If there were any general sort of answer to the question, it would need two parts. One would be local history. What Atran counts as science began around the Mediterranean, took a new turn in Western Europe and now thrives on the Pacific Rim. But scientific knowledge is, thinks Atran, a property of human beings. Pythagoras’s theorem, the first three minutes of the universe, the origin of species – these can be accessed by humans of any origin.

Individual sciences have flourished under peculiar social and institutional conditions. No account of the very possibility of science could ignore local accounts of how a science came into being, who did what, and why it persisted and flourished, or hybernated or died. Those are the proper topics of history and sociology of science, including the ‘anthropology’ of the laboratory. Atran gives his own chunks of local history in his sketches of Aristotle, Cesalpino or Linnaeus and their times. On the other hand, the abilities needed to understand and advance the various schemes are abilities that run across our species. They are different abilities. The naturalist hones one type of skill, the laboratory experimentalist another. None of these abilities will grip unless there is something to grip onto, a symbiosis between the ability and the recalcitrant world around us.

The naturalist needs to be sorting kinds in nature, by and large. The experimentalist relies on the fact that on a good day some of the apparatus works, and accordingly trims ideas of what it is for something to work at all. Chomsky alludes to certain abilities that are preconditions for the possibilities of mathematics but seems too cautious, at present, to say what they are. Atran’s hunches formed by cognitive science make him look for particular human abilities that are preconditions for particular branches of knowledge. That is why he can see his cognitive foundations for natural history as one step towards an anthropology of science. Innate dispositions to sort are the preconditions for the possibility of systematics. Other innate abilities would be preconditions for other departments of knowledge. And I believe Atran is on to something here, but he’s clearly nervous because of his view about ‘the scientific breakaway’. Does science break away from those earlier conditions, with which it needs to have only a loose relationship? Thus he wonders whether ancient and perhaps innate ideas about motion are simply replaced by mechanics. He even seems to be wondering whether there is some kind of competence called ‘scientific’, a predisposition that requires a certain historical conjuncture to trigger it. That, I think, is a bad way to go. Now is a time for hard thinking about what makes our particular sciences possible, what is the background of agreement (to recall Wittgenstein) that may be peculiar to this or that style of reasoning, rather than some global, and vacuous, talk about science in general.

Send Letters To:

The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN

Please include name, address, and a telephone number.

Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.

Sign up to our newsletter

For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.

Newsletter Preferences