- Introduction

THE COGNITIVE GEOMETRY OF NATURE A CONTEXTUAL APPROACH

Roy Ellen

A published version appears in `Nature and society: anthropological perspectives', eds. P. Descola and G. Palsson. Routledge: London

Introduction


That conceptions of nature vary historically and ethnographically, and are, therefore, themselves intrinsically cultural, is so widely asserted nowadays that it is often assumed to have become a self-evident anthropological truth. Perhaps the best example of this in popular environmentalist discourse, as in some anthropology, is the opposition drawn between the holistic systemic vision of 'traditional', 'tribal' or 'archaic' societies and the dualism of the modern scientific and dominant Judaeo-Christian tradition. How conceptions of nature vary beyond such abstractions is well-demonstrated in individual historical [e.g. Collingwood 1945; Horigan 1988; Thomas 1983; Torrance 1992] and ethnographic studies. In particular, much attention has been given to how these might arise from particular practices of environmental interaction [e.g.Bird-David 1993; Ingold 1992], and how these in turn might sustain, or be sustained by [e.g. Schefold 1988], particular social ideologies.1 As Philippe Descola [1992: 110] puts it, 'each specific form of cultural conceptualisation also introduces sets of rules governing the use and appropriation of nature, evaluations of technical systems, and beliefs about the structure of the cosmos, the hierarchy of being, and the very principles by which living things function'. But empirical demonstrations of such relativity - many of which find their origin in the claim by Leach [Leach 1964: 34-5] that nature is no more than some topological grid imposed upon a continuous world - have led to an almost indignant rejection of the very idea of nature. Indeed, it has become increasingly awkward and misleading to carve out from these implied 'representations' or 'constructions', a conceptual space which is linguistically, cognitively and symbolically coherent. The new consensus has thus given rise to new problems: that of commensurability between different conceptions of nature (including the assumption that our nature always exists as a category comparable to their nature); the implication that each culture has a single ruling (and unambiguous) conception of nature, which it is our task to locate, excavate and describe; and the problem as to how those collective notions of nature which we can claim to exist are 'constructed' or 'negotiated'. I shall argue that we can approach the question of the categorical status of nature in two - superficially antithetical - ways.
The first is a continuation of what the relativists and deconstructionists are arguing; namely that any one population may, depending on the circumstances, generate conceptions of nature which are fuzzy and variable - which may indeed be inconsistent and contradictory - and that such variability may reveal itself at the level of individual praxis and at the level of collective representations, or in some combination of the two. Consequentially, it is difficult to speak of 'societies' and 'cultures' (that is, second-order constructs) as having a single conception of nature, and an exaggeration to claim that it is even true of empirically identifiable local populations. Indeed, it may be claimed that some peoples have no concepts of nature whatsoever.
The second way is to buck this trend and identify a minimum number of underlying assumptions upon which pragmatic schemata and symbolic representations are built, and which ultimately constrain human conceptual permutations [c.f. Boyer 1993].2 If I interpret him correctly, this is what Descola [Descola 1992: 110] means by 'the social objectification of nature [being] implemented through a limited number of operative schemes', and his acknowledgement that 'modes of representation of relations to nature present certain similar characteristics' [ibid. 123]. Such widely-observed 'similar characteristics' may be accounted for if we hypothesize that underlying all models of nature are three cognitive axes or dimensions, which will, when the cultural devices they permit are combined in different ways, generate particular representations, all recognisably transformations of some ur- or proto-nature. The first axis is that which allows us to construe nature inductively in terms of the 'things' which people include within it, and the characteristics assigned to such things. The second is that which allows us to define nature spatially: assigning it to some realm outside humans or their immediate living (cultural) space. The third is that which allows us to define nature in essentialist terms; as some force which is exogenous to human will but which can to varying degrees be controlled. To the extent that these three cognitive axes make an equal contribution to representations, they may be predicted to approach that multi-faceted, ambiguous, but ultimately recognisable idea which we in the West recognise as nature; whereas the more asymmetry is introduced into the model the less familiar the construction becomes. And to the degree that each of these axes dominates a conceptualisation characteristic of a particular context, so it also becomes a 'definition' of nature.
I focus here on examples drawn largely from my own work on the Nuaulu, a people of Seram in eastern Indonesia whose mode of subsistence may be summarily typified as a combination of hunting, sago extraction and swidden cultivation. There are no new data, though the way I have used them is different. Thus, I start by asking how we might begin to identify cultural phenomena which best approximate each of the three cognitive axes specified above, and by exploring the extent to which they permit us to infer the existence of nature as a domain. I then examine how combining cultural ideas derived from each of the axes generates intrinsic ambiguity, and how this is reflected in the variation accompanying different practical and symbolic contexts.

Nature as kinds of 'things'


Let us first examine how far the first axis, the inductivist (nature as 'things') model, assists us in generating an approximation, or one dimension, of some hypothetical Nuaulu conceptualisation of nature. In cultural modifications of this model, particular 'things', by virtue of their resemblance to other things, are - once aggregated - seen as part of nature. Thus, through induction, nature itself becomes a thing which can then be the conceptual starting point of deductive reasoning. It has been suggested [Ingold 1986: 3, 1992: 44] that the fact that we see 'things' at all is what distinguishes Homo sapiens from other animals: birds may perceive functional objects such as anvils or missiles, but only humans can perceive something as abstract as 'a stone'. Hence the human environment consists of neutral objects waiting to be ordered, an orientation which is closely linked to the tendency to view animals and plants as physical objects, things of nature. Such a view is - for example - implicit in Lévi-Strauss's theory of totemism.
If nature is the sum total of its parts then we must begin by examining what these constituents might be, and the most accessible evidence here is ethnobiological. It is now well-established that all peoples work with a concept of natural kind, whether as a 'common sense kind of phenomenal substance', or as an 'ontological entity with an underlying nature' [Atran 1990: 86, 94]. There is some dispute as to whether basic categories always occur at a particular level of abstraction, and concerning the detailed way they map onto phylogenetic templates. The extent to which we can represent and account for more- or less-inclusive categories is contested; but the fact that all human populations engage in such activity is hardly in doubt [Ellen 1993b]. What is more controversial is the claim by Berlin [Berlin 1992] and Boster [Boster] that natural kinds have a different perceptual character to cultural forms, fashioned by selective pressure. As the biological world has radiated, so the human capacity to recognise the basic order in that radiation has co-evolved. Thus, in this view (with which I am in basic sympathy), nature itself - or, at least, 'the biological world' - is the product of human cognitive evolution.
A formal expression for this kind of model would be simple incremental linear aggregation: n1 + n2 + n3 .......... = N, where n is a culturally agreed natural kind, and N nature as a totality. But most peoples that we know of also recognise more inclusive conceptual domains such as 'plant', 'animal', 'rock' and so on, such that an aggregative model of nature should intuitively have a structure something like: animals + plants + other living things + non-living things = nature, which we might express formally as:
(n1 + n2 + n3....=)N + (.....)N + (.....)N + ..... = N.
In order to demonstrate the plausibility of such a model we have to show that (a) each of these generic component parts (N) is recognised, (b) that they share common characteristics, and (c) that there is evidence to show that they are linked together to form a conceptual whole (N); an over-arching category in other words. We can explore (a) and (b) by examining the Nuaulu category which glosses most closely with English ANIMAL. The existence of such a category, what Brent Berlin calls a 'unique beginner', can in principle be inferred from the presence of specific terms whose reference can be shown to be coterminous with the semantic content of the domain, or, where such terms are absent, from various linguistic and cultural markers, or (up to a point) by employing sorting experiments.
There is no one commonly-used or widely-acknowledged Nuaulu term for all animals, though there are three possible, partial and rarely-used candidates [Ellen 1993b: 96-7]. So, in effect, we might say that the category is, terminologically, 'covert', or at least 'semi-covert'. But even in the absence of an unambiguous domain label we may infer the existence of a cognitive prototype marked out from other domains, and also from humans. Thus, when speaking of animals, their qualities and relationships, Nuaulu employ a discourse which can be distinguished in certain small but significant respects from other discourses. It differs both lexically and in terms of appropriate semantic relationships. Thus, there are different words for killing an animal (ihunui) and for killing a human (atoria), for a human voice (mo'nyom) and an animal call (nioke); and for human head hair (hua) and bodily hair, animal feathers or fur (hunue). In some cases the differences amount only to slight phonological shifts, as in anai (human child) and anae (animal young). There is also a specialised lexicon for specific activities relating to animals (e.g. atinai, 'to hunt cuscus'; asakaka, 'to call a cuscus'), and many special anatomical terms (e.g. mata hunua, mollusc and insect antennae; kihene, wings and fins), in addition to about 47 such terms which humans and other animals share. That forms in core animal categories are clearly similar in multiple respects, and often overlap; that the terms are often expressed as contrast sets, while animal partonyms ('head', 'heart', and so on) and other linguistic usages are present, goes some way to imply the existence of a category 'animal' [c.f. Taylor 1990: 47-51}]. Such differences not only help to locate and maintain separate domains, in the interests of effective linguistic communion, but also serve the purposes of symbolic contrast. The core of the domain is clearly regarded as being bound by polythetic affinities, such that the unique beginner is not as arbitrary as some [Hunn 1977: 44] have suggested, although its borders may occasionally be difficult to determine.
We could undertake a similar exercise for the domain 'plant', or indeed for any other conceptual domain which aggregates identifiable parts of the human environment on the basis of similar characteristics or overall form, simply by drawing inferences from linguistic utterance and cultural practice. Once this has been done we might then order the domains themselves into more inclusive contrastive categories, using distinctive features such as those Paul Taylor [Taylor 1990] has suggested for the Tobelo: living--non-living, sexual--non-sexual, breathing--non-breathing and so on.
Such abstract distinctions, are also discernible among the Nuaulu, but, have little practical bearing on their lived culture, are of no prominent symbolic significance, and certainly cannot be held to be a starting point for their classification of the natural world. In this instance, I suspect they are examples of that genre of elicited but quite ungeneralisable contrasts which are happily provided by willing informants, and at which the ethnographer is tempted to grasp in some vain attempt to impose order on what otherwise looks like utter chaos.
In short, the hierarchic conception of nature typified by scientific taxonomy, and its folk-semantic extension which includes at its more-inclusive levels contrasts between unique beginners, and life and non-life, is not one which is readily yielded from the Nuaulu data. But we do not need to rely on such abstractions, irrespective of whether or not they can convincingly be shown to be emically-rooted, in order to model an aggregated or integrated concept of natural things; as conceptual domains can also be linked through their overlap (especially the overlap of their peripheries), by what various authors [e.g. Hays 1976: 502] have described for less inclusive categories as 'chaining', or 'linking'. Thus, if a is linked to b, b to c and c to d then this implies the existence of a 'group' a-b-c-d: an instance of polythetic resemblance. Certain Nuaulu life-forms which are regarded as animals in phylogenetic terms have no obvious affinities with any other category. These include sponges, which are grouped with 'fungi'. Molluscs and starfish, however, are firmly perceived as animals for a combination of behavioural and morphological reasons. There is no sharp division between animals and plants and other domains, either in linguistic or conceptual terms [Lévi-Strauss 1966: 138-9; Morris 1976: 542]. Some invertebrates are ambiguously animals and plants, some plants (e.g. certain fungi and lichen) are ambiguously plant and inert material; different life-forms merging together to strongly permit the inference of a LIFE category, which itself may merge with non-life. Thus, despite the cognitive imperative to distinguish domains in terms of a small number of features or cognitive prototypes, in practice there are always going to be 'problems', some of which may be culturally manifest as anomalies; and it is precisely these which serve to link domains into more inclusive groupings.
If nature is an inventory of things in this sense, then this must be linked to rules about how these things are to be identified and related: that is order, and this order has to find some cultural legitimation. Nuaulu acknowledge that there is (or indeed must be) order in the world, an order which is in general terms comparable to the pre-Darwinian 'great chain of being'. Non-directive evidence for this might first be sought in that part of Nuaulu origin mythology which speaks of the time when the first Matoke (Lord of the Land) descended from the sky and walked throughout the earth where each natural kind was represented only by a single organism - one house, one betel palm, one hornbill, and so on. As the Matoke came by each of these, so he named it, saying: 'This is a snake', 'This is a betel palm', 'This is a hornbill' and so on. And as he did so, the many emerged from the singular. But this is not to say that Nuaulu readily expound, even less agree upon, the principles for order, and certainly not the identification or classification of animals for pragmatic purposes. They will simply admit to not knowing it, or at least large parts of it [Ellen 1993b: 94]. Less modest individuals might dispute amongst themselves, but every disputed classification itself exists within an axiomatic field the coordinates of which are assumed to be quite fixed. And the primus-inter-pares of axioms is that nature itself is finite, and that all animals have names- even if they remain unknown. In Nuaulu theory at least, names are not given arbitrarily, for economy of thought; they reveal part of an order which was laid down at the beginning of the world, but which is only partially known about, and even less understood.
Nature as an inventory of things reaches its apogee in modern Western classifications. The 'great chain of being' has already been mentioned, but it is also there in the concept of species, in the taxonomic schemes of Linnaeus and their Darwinian re-interpretation; it is there in the very idea of 'natural history', embedded within the notion of exhibitions to display nature as these emerged from cabinets of curiosity to become museums of natural history, herbaria, botanical and zoological gardens, during the eighteenth and nineteenth centuries. What is additionally significant here, and which comes through well in static displays, is that these are concerned as much with minerals (which have never lived) as with life; and equally with dead plants and animals, and the ambiguous classificatory material presented in the form of bones, fossils and the mineral extrusions and excretions of living things (e.g. coral). The conceptualisation of nature as a collectivity of things is, therefore, most obvious in the representations generated by western science and those generated by anthropologists investigating the folk classifications of the natural world, where the Western paradigm is the implicit or explicit reference point. It is there also, supremely so, commoditised in both its parts and its entirety, in the slogans of environmentally-inspired marketing and in the politics of ecology and biodiversity.

Nature as space which is not human


Nature is often understood less as some abstracted inventory of its contents (in which items are cognitively detached from their habitat and re-organised according to a limited number of morphological or functional criteria), than in terms of its predominant spatial or phenomenological manifestation. It is this definition which is implicit in many ethnographically-reported instances of the semantic congruence between forest and nature. But for different peoples, using different subsistence strategies, living in different environments, the semantic congruence may be with some alternative topography ('sea' or 'desert', say, or 'mountains'); all of which have in common what is perhaps best (provisionally) construed as the quality of 'wilderness'. But 'wilderness' is the apogee of something closer and more familiar, though different. So, although for the Nuaulu the congruence is archetypically exemplified by 'forest'; more routinely, perhaps, in this spatial sense, nature is 'that which is not of the village', or 'that which is not of the village or gardens'.
The 'natural otherness' of the Nuaulu concept of forest is encountered in its most mundane sense in adjectival qualifiers for animals or plants: as in the contrast between the geckoes imasasae numa ('house') and imasasae ai ukune ('tree top, tree branches, far forest'), for Hemidactylus frenatus and Gekko vittatus; or in the contrast between the murids mnaha numa/niane ('house/village') and mnaha wesie ('forest'), for Mus musculus and Melomys spp. It is also evident in the symbolic organisation of village space [Ellen reductionist], and in the differential attitudes to language and behaviour within and without the village. Thus you may 'joke', mock or use expletives involving certain animals in the village, but you may not do so outside. Among such exclamatory phrases are: ikae nawe, 'long fish', mau (w)anae, 'kitten', asuwani anae, 'young cassowary', and hahu onate, 'large pig'. In the village (and by extension in the gardens as well) these expletives are used extensively in ordinary discourse; in the forest they anger the spirits and bring on heavy rain. While an individual may try to escape the consequences of his utterances, he runs the risk of being swallowed up by the earth [Ellen 1993b: 175-6]. Interestingly, the oath masi mokota, 'let the earth open up', is subject to the same taboo. The same - village (permissible) : forest (prohibited) - rule applies to a wide range of other expressions, among which are included expletives derived from the names of spirits (e.g. painakite raia). This is significant given the classificatory similarities between animals and spirits, which I shall turn to shortly.
But it is insufficient to contrast forest and village in the abstract, since the usual personal experience is that forest, bush, or whatever its semantic approximation might be, surrounds, or encompasses the village, and ultimately the self; and it is in this sense that nature comes closest to what in the western scientific tradition has become 'environment'. Thus, nature is always constructed by reference to the human domain, and is in the last instance informed by ideas and practices concerning 'self' and 'otherness'. This is not merely symbolic analogy, but an homology of experience. This should not worry us since it is a matter of fact that what is experienced and represented around us homologously is at key moments, in significant contexts, symbolically transformed into abstract binary oppositions which permit more formal analogy. Thus the Nuaulu experience of living within houses which are located within village spaces which are in turn located within forest, is easily transformed into various abstract linear oppositions between, say, house and forest or village and forest; which in turn may be drawn into more complex symbolic linkages through analogy [Ellen 1986].

Nature as inner essence


The third dimension of the concept of nature is its sensation as an inner essence or vital energy or force, outside human control. This is the most intangible of the three. We can perceive and touch 'things' and walk through 'spaces', but inner essence is usually only experienced in terms of its sensate consequences, usually through some combination of the first two axes. However, the best physical manifestations of inner essence are those fluids and pulses associated with living things, with bodily function: blood, sweat and tears, semen, breast milk; heart beat, breath, excretion, movement; or more generally in the environment: flow of water, heat and cold, wind, noise, growth. And this is not confined to life narrowly-construed. Thus, Tournefort could identify the act of creation in both seeds and mineral crystals [Atran 1990: 230]. In certain clearly defined circumstances the generic cultural character of essence or energy will be clear; thus with respect to human passion, we often speak of it as 'animal nature'. This is embedded in the Western notion of natural instinct, nature as opposed to nurture; Islamic 'hawa nafsu' as opposed to 'akal' (reason), and the widespread Indonesian idea that the process of socialisation is the progressive controlling of natural forces. But nature within need not primarily be an allusion to animality; the dynamic properties may rather be experienced as consubstantial in any number of different 'kinds of' nature. It need not be a metaphor for the social, and sometimes it might be better to speak of both natural and social phenomena - as say reflected in physical maturation - as wholly comparable outcomes of similar processes [Bloch 1992].
The idea of nature as an essence or force within may everywhere have associations of uncontrollability, but we cannot prescribe in advance whether its cultural expression will be positive, negative or neutral. This will very much depend on the cultural metaphors which draw upon the imagery, and one set of metaphors which do so, and which have been extensively examined in the literature, are those linked to gender [Atkinson and Errington (ed.) 1990; MacCormack and Strathern (eds.) 1980; Valeri 1990]. As it happens, the Nuaulu data fit well with the male is to female as culture is to nature motif, females treated as representing a danger to the male order, the intrusion of nature into culture often physicalised with reference to menstrual blood and the act of childbirth.

Boundary problems and contradictions


Each of the three axes outlined, or if you will dimensions, or definitions, is insufficient in itself to generate or define any one cultural construction of nature: all three are necessary to even begin to map out its underlying geometry. Moreover, my presentation so far has been fundamentally artificial in that I have ignored boundary problems and inner contradictions which arise once we juxtapose two or more of the axes, which is of course how we culturally experience nature. It is true that I have had to anticipate some of these, as no ethnographic data known to me present the conditions in any other way; but I have tried to limit these in the interests of clarity of exposition. It may help in reviewing these issues to imagine the three axes as related in three-dimensional space. I shall first consider the conflation of the first axis (nature as an aggregation of things) with the second (nature as an other space); I shall then consider the conflation of the first with the third (nature as inner essence); and finally, conflation of the second with the third. The justification for this order will, I hope, become apparent. Of course, there are some contexts in which all three axes have a direct bearing on what is going on.
The conflation between 1 and 2 (things and the spatial other) is best exemplified by the universal cultural recognition that humans themselves might possibly be 'things' of nature, comparable to other natural things, that the inventory of nature is not confined to the other (a version of the so-called subject-object problem); and by the recognition that humans physically intrude into the space of nature. We can explore this idea in relation to Nuaulu animal classification.
In general terms, people are regarded by the Nuaulu as being in many respects like animals. People share anatomical and physiological similarities with animals, but more than this myths inform us that animals, like their human counterparts, have societies. In the case of some species, they are represented as reflecting basically human organisation and values [e.g. Ellen 1972: 233]; they are spoken of in the idiom of kinship. Animal societies too are bound by the Patalima-Patasiwa, 'Five group'-'Nine group', division of Seramese peoples (the Nuaulu themselves being Patalima), whilst totemic traditions and a rich mythology underscore the idea that animals may change into humans, and vice versa [Ellen 1993b: 163-76]. In short, humans impose a social classification on the world of animals. Many terms referring to behaviour and appearance which are used for humans are also used for animals. In some instances shared terms may be understood as consubstantial, while in others the allusion - at least - is to a metaphorical extension from humans. The exceptions occur - and we have examined some of these already - where there is no human model, as with 'wing', 'beak', 'tail' and so on. And, of course, none of this prevents Nuaulu from defining the domain of animals essentially in contrast to humans. Taylor [1990: 51] reports that in Tobelorese language and concepts humans are treated quite differently from 'animals'. He regards this as problematic, since 'humans' meet the defining features of the Tobelo category 'fauna'. Rather than being problematic this seems to me to be an understandable feature of all people's conceptual universes. We might ask if humans are 'animals' in British or French folk classification. The answer, of course, is that .I it depends, .R and in this respect Nuaulu beliefs are like those of many other peoples. An extension of the same problem is encountered when we consider the place in any inventory of natural things of domesticated species, or any humanly-modified part of nature [Descola 1992: 111].
The conflation between 1 and 3 (things and essences) is best exemplified in the attribution of essence to particular parts of nature. A widespread version of this idea is associated with animism - a kind of 'social objectification of nature' [Descola 1992: 114], and I have shown elsewhere [Ellen 1988] that the attribution of life to the inanimate (most commonly through anthropomorphism) is basic to all human conceptualisations of the world. It is the continuity of natural kinds, as discussed above, which must, in all cultures, give plausibility to the idea that all nature is animate: animal, vegetable and mineral; the humanly-modified and the humanly-unmodified.
A particular aspect of this continuity is evident in my Nuaulu data dealing with the consubstantiality of spirits and animals. Nuaulu recognise spirit categories in much the same way as they recognise categories of animal; indeed spirits are treated as natural kinds, as equally significant parts of their environment [Ellen 1993b: 176-9]. People claim to hear and 'see' spirits all the time and I have on occasions been present when the alleged discovery of a particular spirit in a tree, or in a bush, has created scenes of some excitement. Some Nuaulu spirit forms appear to describe real animals; for example sinne inae (certain scarab and long-horned beetles, including Oryctes rhinoceros and Mulciper linnaei), (kau) kama nahune (edible long-horned beetles such as Gnoma giraffa and Glenea corona - kama nahune being the spirit of a person killed by falling from a tree in the throes of hunting cuscus), inararai (the frog Litoria amboinensis), and rikune (various kinds of bugs and beetles, including Mictis, Oncomeris and Euphanta). Perhaps we should not be surprised if it is insects which are most likely to be redefined as spirits [c.f. Dentan 1968: 26-7]. Other categories, such as naka which refers to those mythical creatures we call dragons, are used by the Nuaulu to label certain real world animals which they have heard of but never seen, in this case the Komodo 'dragon'. Domains become even more blurred when spirits enter the bodies of animals influencing their behaviour, as when a sakahatene enters the jaws of the death adder nanate (Acanthophis antarcticus). Other spirits are modelled on particular animal prototypes to the extent that experiences of paired entities seem at times to be conflated. Thus, masenu are compared with tuku tuku (prob. Otus magicus), and ahone with sakoa (Ninox squamipila) in their vocalisations. These are both owls and consequently nocturnal, which is itself significant. Some animals are held to be derived from spirits, such as isanone ants from isanone nanie. So not only are there sometimes no simple breaks at domain boundaries in what we might construe as the 'real' world, there are even areas of overlap between the objectively visible and invisible.
The conflation between 2 and 3 (space and essence) is no better summed-up than in the notion of 'the wild', and its cognates. The natural other is not always chaotic or malign and for some peoples it might most faithfully be expressed as some kind of 'culture of the beyond' [Schefold 1988]. However, as far as the Nuaulu are concerned, it is unpredictable, difficult to control and with an fundamentally moral character; there are right and wrong ways in which to engage with forest, which arise in part from the specific social histories of parts of it, but also from its intrinsic mystical properties. This natural 'other' is reflected in the inferential symbolic opposition between 'nature' and 'culture' evident in most ritual, in the specific rituals conducted prior to cultivating forest, in the charms which are used to protect travellers in the forest, in the prohibitions on certain behaviours and utterances while in the forest, in the correct ritual disposal of its products. When humans enter the forest they carry with them what amounts to an 'aura' of culture, and when ritual is conducted in the forest, it is as if islands of culture are created to ensure its efficaciousness. Thus, in Nuaulu male initiation ceremonies platforms are erected which mimic that entity which most epitomizes (indeed physicalizes) culture, namely the house. Individual neophytes at this ceremony are required to stand upon blocks made of five logs as if preventing contamination from the forest. Similar structures are used when performing land-clearing and other routine rituals. The same meeting of nature as inner force and as outer space is reflected through the prism of gender concepts in the symbolic layout of Nuaulu villages. Here, females are associated - through the location of their menstruation and birthing huts - with the outer rim of the village (nearest the forest). What is, therefore, universally significant about this conflationary aspect of nature is that it becomes a condition for knowledge, by controlling the relation between what is taken as internal nature and what is taken as external nature [Strathern 1992: 194], that which is natural and that which is of nature.

Contextual variation


We can see, therefore, that logically the functional association of any two, or all three, of the cognitive axes specified results in conceptual complications which greatly extend and enhance the richness of the symbolic imagery of nature. But what is intriguing about human cognitive and social behaviour generally is that logical inconsistencies can be suppressed in particular circumstances, and different aspects of a multi-dimensional idea privileged at the expense of other aspects in any one context. Let us take a few rather different Nuaulu examples of these: forest, animal spirits and ritual killing.
As we have seen, the archetypal Nuaulu representation of the collective natural other is wesie, uncut primary forest [Ellen 1993a: 138-40]. However, this contrasts in different ways with other land types depending on context. It may contrast with wasi (owned land, which may sometimes display very mature forest growth), emphasising a jural distinction; with nisi (garden land), emphasising human physical interference; or with niane (village), emphasising landforms: empty as opposed to well-timbered space, inhabited (dwelt) as opposed to uninhabited space, untamed as opposed to tamed space, all with various symbolic associations and practical consequences for Nuaulu consumers. Although there are no Nuaulu words for either 'nature' or 'culture', it is in the various and aggregated senses of wesie that the Nuaulu come closest to having such a term, and from which the existence of an abstract covert notion of 'nature' can reasonably be infered [cf e.g. Valeri 1990].
Thus, in particular contexts the meaning of wesie as a natural other may be sharply dichotomized, only to appear in a different guise elsewhere. As Croll and Parkin [1992: 3] argue, most peoples ascribe a somewhat capricious agency to their environment which they are obliged to interpret and negotiate, and which they commonly regard themselves as inseparably part of. In some contexts, even for the Nuaulu, the forest is the people, in the same way as the ancestors are, in a sense extensions of the living. Negotiations and renegotiations take place regarding the meanings of forest and village, cleared and uncleared, cultivated and uncultivated, wild and tame [Croll Parkin 1992: 16]. Oppositions are set up only to be transcended or merged; sometimes forest in male, sometimes female; sometimes portrayed as antagonistic, sometimes life-nurturing. These all provide alternative modes of identification [ibid., 16].
Similarly, the juxtaposition of classifications of spirits and animals not only serves to show the structural similarities and the conceptual bases of categories and their relationship to each other, but also reminds us of an important difference: that spirits are in the normal way experientially incorporeal while animals are first experienced as things, even though the Nuaulu 'know' that spirits have bodies and bodies have spirits. The logic of this in one direction is that the Nuaulu must claim to see spirits for them to exist, and in the other that animals must have spirits because of the prohibitions and beliefs surrounding them. But the notion that animals have spirits is problematic if you have to kill them, and Nuaulu culture provides a very practical response to this, in rituals connected with wooden meat skewers or asumate. When an animal is butchered it is generally skewered on a sharpened wooden stake, and a chip from the pointed end kept and afterwards tied to the pole. This piece is traditionally employed to butcher the killed animal, and represents its spirit; the re-tying to the pole is thus supposed to represent the re-uniting of the soul and body of the animal killed. The purposes of the asumate are: to inform the ancestors that meat has been killed and that they should come and partake of it; to confer prestige, the asumate being placed where everyone can see it; to return the spirit to the cosmos, and to therefore ensure that finite stocks are not depleted and that hunting prospects remain good. Every time a hunter fails to plant an asumate the wild stock of that species is thought to be depleted by a factor of one. Such practical steps, however, if taken to their logical extreme become highly inconvenient in the normal daily round. Far better to rely upon periodic strategic amnesia and operate with two contradictory conceptions of the animal world: one which stresses unity with humankind (and the privilege of taking life for food) and another which stresses the fundamental differences between humans and animals and which legitimates and makes easier their exploitation as food [cf. Wazir-Jahan Karim 1981: 188]. Of course, any cosmology motivated by animism must generate respect for other species, a respect often reinforced by prohibitions of various kinds. But this need not be inconsistent with hunting, and I suspect that there may long have been a contradiction between the doctrine of infinite renewal and the recognition that hunters could exterminate animals locally [Brightman 1987: 137].
My final example conveniently follows on from our consideration of asumate. All Nuaulu killing in the course of hunting takes place outside the village, in the natural other. That killing which takes place as a culmination of hunting not only takes place in nature, it involves a part of nature, and exemplifies the control of natural forces: the domination of culture over nature in nature. By contrast, animal sacrifice takes place in the village (is therefore controlled, cultural, killing), the domination of nature in culture. In the first, the offering is a consequence of a killing for some other purpose (usually for food); in the second the food (where it is consumed) is a consequence of the offering. But meat killed outside the village is not just any kind of meat, it belongs to recognised animal categories, specific natural things. The most important of these are peni (the collective term for pig, deer and cassowary), which is simultaneously natural and cultural. Peni (creatures of the forest) pass, therefore, from a natural (un-controlled) condition into a cultural (controlled) condition, and in an important dual sense sustain the possibility of 'culture', being identified with the descent group and house in which their accumulated mandibles have been stored (culture-within-nature). Animals of sacrifice, by contrast, are of domestic stock (chickens - preferably cockerels) or of the village realm by metaphoric association, as in the case of the cuscus (phalangers), which resemble humans and which, through historical convenience have become a substitute for human heads. Such animals are, therefore, nature-within-culture [Ellen in press a].

Conclusion


Nature is definitely not a basic category. It is more a 'higher order' category in Rappaport's [Rappaport 1971: 33-4] sense, and for many peoples it would appear to have no clearly bounded categorical status at all. As with more inclusive units in the Linnaean hierarchy, non-basic folk categories cannot be objectively defined, and no firm distinction between perceptual and social can be sustained. Indeed, it is inconceivable that classification might proceed in ways which, to follow Geertz, 'externalise culture'. Conceptualisations of nature are not the inventions of individuals (in which case they would more closely reflect cognitive process), but arise through historical contingency, linguistic constraints, metaphorical extension, ritual prohibitions and so on. As parts of belief systems, they are the productions of interactions, accretions, elaborations and condensations. That contradictions and inconsistencies exist is because nature is simultaneously an abstract symbolic and a non-basic cognitive category, variously a model 'of' the world (a representation) and a model 'for' (a plan for action) [Geertz 1966]; and it is precisely this ability to switch between one and the other, to engage with the environment and disengage from it which distinguishes us from non-human primates.
Our understandings of nature are rooted in particular situations which when their common meanings are distilled provide us with something which has a provisional, abstract and emergent quality. Such notions of nature are the consequence of what I have elsewhere [Ellen, 1993: chapter 8) called .I prehension: .R those processes which through various cultural and other constraints give rise to particular classifications, designations and representations. People bring to situations in which classifying activity takes place, and from which verbal statements about classifying behaviour result, information of diverse kinds acquired through both informal and formal socialisation experience, of the world in general and of earlier classifying situations. How they then classify depends upon the interplay of this past knowledge (including prescriptions and preferences with regard to particular cognitive and linguistic idioms) with the material constraints of the classifying situation, the purposes of the classifying act, and upon the inputs of others. inter-penetrating ones.
In addition, it is important to recognise that the processing and storage of information in the brain is imperfect, and communication of that information less perfect still. Paradoxically, there is a connection between this shortcoming and the considerable capacity of the human mind to re-order information in different ways, replacing irrelevant information with that of greater and more immediate utility. As Sperber [Sperber 1985: 31] has remarked, 'mental representations have a basically unstable structure: the normal fate of an idea is to become altered or to merge with other ideas; what is exceptional is the reproduction of an idea'. Following Lévi-Strauss, he insists that any epidemiology of idea
s is, therefore, as much concerned with transformation as persistence. That understandings of nature are messy, cross-cutting and changing is a reflection of this. Concepts are often used, operationalised, without defining them. The efficient practice which precedes theory does not require self-reflection on the operation while performing it, and much of what we learn, in fact, is learning not to think about operations that once needed to be thought about [Medawar 1957: 138].
All of this is to emphasise the contextual, variable and contingent way in which we use those cultural abstractions which in terms of our own emic conventions we find convenient to represent as 'nature'. But, of course, in order for communication to take place classification must have at least some intersubjective structure, agreed cultural rules, some 'doxa' [Bourdieu 1977]. Much of this is possible simply because there is sufficient agreement about cultural conventions, but this is not to endorse a 'a grammar of the variety of ways in which nature is socialised' [Descola, ibid]. Everything I have said here suggests that a grammatical analogy would be quite false, as would be any attempt to confine discussion to a linguistic level of expression. Very few languages have words which easily translate as 'nature', yet some focal notion in terms of the three cognitive dimensions examined here is always present. Sometimes, as in the Nuaulu case, the resemblance to global etic nature is sufficiently close for it to be recognisable; in other cases - as among many hunter-gatherers - the resemblance is much weaker. Language only mediates - and then rather inadequately - between the many cultural appearances and the three underlying cognitive axes which generate the possible coordinates: the objectification of the world, the spatial other and inner essence. And such a model of the cognitive geometry of nature as I have offered here is consistent with recent attempts - such as those of Bloch and Ingold - to go beyond linguistic representation and to situate perception within actions on the world (non-mediated forms of knowledge), and to - paradoxically - resist the imposition of our own nature-culture dualisms on data.

Acknowledgements


Earlier versions of this chapter were presented to a seminar at the Muséum National d'Histoire Naturelle in Paris, sponsored jointly by the CNRS and the Laboratoire d'Ethnobiologie-Biogéographie, and as a Munro Lecture at the University of Edinburgh. I would like to thank Claudine Friedberg, Cecile Barraud and Anthony Cohen for the opportunity to explore the ideas presented here in a revised and shortened form. All of the Nuaulu data referred to has been published before in the works cited, where full acknowledgement of permissions and funding bodies may be found.

Notes

  1. I provide here only a few indicative references. Further examples, together with a more extensive discussion of the 'cultural construction of nature' are to be found in Ellen [Ellen in press b].
  2. In a sense, my aim is to examine the extent to which it is possible to identify a category innocent of morality at a time when the prevailing inclination is to emphasize the intrinsically moral character of nature. One might, of course, demur that ultimately all categories imply rules, and all rules the moral force of 'right' and 'wrong'. In this I would agree.

    References

    Atkinson, J.M. and S. Errington (ed.) (1990) Power and difference: gender in island
    Southeast Asia, Stanford: Stanford University Press.
    Atran, S. (1990) Cognitive foundations of natural history, Cambridge: Cambridge
    University Press.
    Berlin, B. (1992) Ethnobiological classification: principles of categorization of
    plants and animals in traditional societies, Princeton, New Jersey: Princeton University Press.
    Bird-David, N. (1993) Environment, metaphor and culture: hunter-gatherers in Asia,
    Africa, Australia and North America compared, in K. Milton (ed.) Anthropological perspectives on environmentalism (ASA Monograph 31), London: Routledge.
    Bloch, M. (1992) What foes without saying: the conceptualization of Zafimaniry society,
    in A. Kuper (ed.) Conceptualizing society (European Association of Social Anthropologists), London: Routledge.
    Boster, J. (in press) Human cognition as a product and agent of evolution, in R.F.
    Ellen and K. Fukui (eds.) Rethinking nature and culture: ecology, cognition and domestication, London: Berg.
    Bourdieu, P. (1977) Outline of a theory of practice (Cambridge Studies in Social
    Anthropology 16) (trans. Richard Nice), Cambridge: Cambridge University Press.
    Boyer, P. (1993) Cognitive aspects of religious symbolism, Cambridge: Cambridge
    University Press.
    Brightman, R.A. (1987) Conservation and resource depletion: the case of the boreal
    forest Algonquians, in B.M. McCay and J.M. Acheson (eds.) The question of the commons: the culture and ecology of communal resources, Tucson: The University of Arizona Press.
    Collingwood, R.G. (1945) The idea of nature, Oxford: Clarendon Press.
    Croll, E. and D. Parkin (1992) Cultural understandings of the environment, in E.
    Croll and D. Parkin (eds.) Bush base: forest farm: culture, environment and development, London: Routledge