The grid and the tree

This work is licensed under the Creative Commons | © Philippe Descola. ISSN 2049-1115 (Online). DOI: http://dx.doi.org/10.14318/hau4.1.015


The grid and the tree

Reply to Marshall Sahlins’ comment

Philippe DESCOLA, Collège de France


It was an immense privilege for me to have Marshall Sahlins contribute a foreword to the English edition of Beyond nature and culture (Descola 2013), and it has perhaps been an even greater honor to see him comment on the argument of the book by proposing to write another version of it. Among Marshall’s many celebrated aphorisms there is one the lucid cruelty of which I particularly enjoy: “A scholar is only sure of two things, that he is going to die and that he will be proved wrong; it is best when the former happens before the latter.” But, as Marshall himself seems to imply in his comment, there may be a happier alternative: a theoretical construct may appear compelling enough to provide the basis for a series of reformulation by others that will vindicate the basic intuition it was based upon. There is no need to die, then; not yet.

The gist of Marshall’s argument is that the default ontological inference of all humans as to the properties of many objects in their environment is anthropomorphism, of which the closest systematization is animism. All the other ontologies, even what I call naturalism, are but variations of that widespread propensity to personify beings and phenomena. I cannot enter here in the heated debates raging in anthropology and the cognitive sciences upon whether or not anthropomorphism is a universal disposition, by what mechanisms it is activated, and whether or not it provides a sound basis for inferences about the agency of nonhuman beings. Suffice it to say that I do not preclude the possibility that there may be specific cognitive dispositions operating in the formation of ontological judgments of the kind I have sketched in Beyond nature and culture, a point I have started lately to explore with a team of developmental psychologists. However, my main objective is not to unearth cognitive universals, but to provide anthropological models (i.e. able to deal with ethnographical and historical data) which eventually may, or may not, be rooted in psychological dispositions. This is why I prefer to draw a distinction, as I have done in the book, between anthropomorphism and anthropocentrism, the former being a tendency to detect human personhood in nonhumans, a tendency on the cognitive nature of which I can say little as an anthropologist, while the latter is an anthropological feature—humankind as the highest form of being and the template for judging all others—the existence of which can be ascertained in some cultural settings. In that sense, animism is not even anthropocentric, for in animist collectives nonhumans share the same condition as humans, the only privilege that the latter claim for themselves being that they engage with nonhumans in relations predicated upon norms of behavior that are common among humans. Animism is thus “anthropogenic” rather than anthropocentric, in that it derives from interactions between humans all that is necessary to make it possible for nonhumans to be treated as humans. By contrast, naturalism is properly anthropocentric in that its spokespersons define nonhumans tautologically, by their lack of humanity, and claim that humankind and its attributes represent the paragon of moral dignity that other existing beings lack.

Once he has set up animism as the default ontological principle, Marshall then proceeds to show, through a wealth of ethnographic examples, first, that totemism and analogism are but variations of the animist scheme and, second, that each ontological mode may also encompass elements of the others in a subordinate position. A discussion of each case would take us too far, and I will only tackle those that stem from what appears to be a confusion of categories on Marshall’s part between three different analytical levels: that of the modes of identification, that of the modes of aggregation, and that of the modes of relation. To cut a long theory short, modes of identification are the building blocks from which ontologies are made, the basic inferences regarding the types of beings the world is furnished with and what they afford to humans in terms of possible interagency. From these elementary features—hypostasized as animism, naturalism, totemism, and analogism—specific modes of aggregation of the different kinds of beings are derived: animist tribe-species are thus human and nonhuman monospecific collectives each defined by their distinctive “physical dispositions” (which include, among humans, such traits as languages, weapons, or ornaments); naturalist collectives are “societies”; totemic collectives are hybrid aggregates of humans and nonhumans sharing a set of moral and physical qualities; analogist collectives are cosmic aggregates of interlocking mixed segments, and so on. Now, the reason for my giving an analytical preeminence to ontological specifications (the modes of identification) over modes of aggregation was to get away from the usual sociocentric interpretation which qualifies cosmological features as an outcome of how humans organize themselves in specific polities. The whole point of bringing forward an ontological perspective is precisely to avoid stating what Marshall writes apropos totemism, that “nonhuman persons, as species-beings, are substantively identified with different human collectives.” In spite of Marshall’s claims to the contrary, this amounts to a proposition wherein natural species elevated to the status of persons, on the one hand, and clans, on the other, are analytically separated and ontologically preexist their articulation, thus opening the way for the implication either that the former are a mental model for the latter (Lévi-Strauss), or that the latter are templates projected onto the former (Durkheim). There are no “human collectives” in totemism—even less a Scottish version of them—but hybrid multispecies groupings wherein humans strive, through complex rituals, to disentangle themselves from the mass of beings with whom they share an origin and an identity and to carve out some functional mechanisms for their specifically human life concerns. The dazzling complexity of the latter has given food for speculation to a score of social anthropologists who, being faithful to their trade, could not avoid treating them as autonomous, multipurpose social devices.

The third level, that of modes of relation, refers to encompassing schemes defined by a dominant kind of behavior toward humans and nonhumans alike, schemes which can be specified because they are institutionalized in norms and obligations, and often ritually enacted in a condensed form. To take a simple example, the ontological specifications of animism in Amazonia imply a single mode of aggregation—monospecific collectives acting as tribes-species all endowed with humanlike sociocultural features—but they also afford the possibility of three modes of relation within and between collectives: exchange, predation, and gift. These schemes characterize the differential ethos of different collectives, all marked by relations of asymmetry or dissymmetry between equipollent terms, whether human or nonhuman, while at the same time encompassing restricted expressions of other modes of relation in subordinate positions. This is why human collectives which favor either exchange, or predation, or sharing may also accommodate an asymmetrical relation of protection such as the one which parents extend over their children or which spirit masters extend over game animals. However, although protection implies a disparity between the protector and the protected—and thus contrasts with the three dominant modes of relations which are grounded upon an equivalence of the terms—it is nonetheless not a relation of hierarchy, as Marshall puts it. For hierarchy is a device that operates at another level, in analogist systems, to structure discontinuous elements—humans, deities, qualities, organs, positions, sites, propensities, and so on—into a series and along a graded scale. Hierarchy in that sense is not expressed primarily as the domination of one element over a set of others (although it may often entail effective supremacy); it is, rather, one of the several implements devised by analogist systems to organize and render manageable long chains of singularities. Hierarchy thus typically operates at the ontological level, and to take it as a mode of relation—as in the control extended over game animals by spirit masters—amounts to a confusion of categories.

I now turn briefly to the very interesting counterexamples that Marshall has fished out from the depths of his fantastic erudition to argue that what I qualify as totemism or analogism are nothing but local variations of animism. I do not have the space here for a detailed confrontation of ethnographic data, so I will restrain my answers to a few remarks. According to Marshall, Australian totemism can be construed as segmentary animism, mainly because some nonhumans have a personal relationship to humans (e.g. the sun and the moon, who appear as humans among the Aranda), and also because there are cases of perspectivism (in Aranda myths, again) where humans and animals perceive the same thing differently. On the latter point I confess that I now have doubts as to whether perspectivism is indeed, as I surmised in Beyond nature and culture, a particular case of animism ingeniously exploiting the possibilities opened up by the difference in physicalities upon which this ontological regime is founded. The huge success enjoyed by the notion ever since Eduardo Viveiros de Castro (1996) launched it in the academic sphere has triggered the documentation of countless cases of perspectivism in every imaginable setting, so that the phenomenon may be after all, if not a universal feature, at least a product of a mental exercise common enough everywhere humans hunt and fish: that is, where they need to see themselves through the eyes of their prey. Thus, a case of perspectivism in the central Australian desert would not be surprising.

Aside from that, I am still adamant that the kind of relations that Australian Aborigines maintain with nonhumans of every denomination is very different from the one that prevails in Amazonia, in the Arctic, or in Northern Siberia. For brevity’s sake, I will use, just this once, an argument of authority. Nicolas Peterson (2011) has recently severely criticized anthropologists writing on Aboriginal people’s relation to their environment—including colleagues working, as he does, in the central part of the continent—for being exceedingly influenced by the likes of Descola and Viveiros de Castro when portraying the people they study as animists who endow plants and animals with personhood. Academic life is full of paradoxes, for the arguments he sets forth vindicate my approach of totemism, in particular his conclusion that “the evidence that normatively Warlpiri, or other desert Aboriginal people believe or believed that the world is full of persons, only some of whom are humans or that plants and trees have sensibilities, is lacking” (177). As for the fascinating Manambu material, I am forced to make another candid confession, this one more damning: I am baffled by the ethnography of Papua New Guinea, for not only does one find there clear instances of animism (peoples of the Great Papuan Plateau, for instance) and totemism (the Sepik indeed) distributed in a relatively restricted area (at least compared with animist Amazonia or analogist central America), but this ontodiversity is also combined with cases where dimensions of animist, totemic, and analogist collectives seem to coexist. This opens up the possibility of treating the island and its neighbors as a vast group of transformation where a basic set of ontological features are expressed in complex combinations, much as Pierre Lemonnier (1990) did with his comparative study of the systems of exchange and political authority in the Highlands, a vast program that may be possible one day when anthropologists awaken from their monographic slumber.

As for the Polynesian version of analogism summed up by Marshall from Valerio Valeri’s superb monograph (1985), it appears to me as a perfect example of that ontological regime and I fail to see any trace of animism in it. As in every analogist cosmology, the Hawaiian world is parceled into complementary domains, each subsuming a multitude of entities, qualities, and processes, including different varieties of humans. Each of these domains falls under the jurisdiction of multifaceted cosmocratic deities who can hardly be described as human-like persons since they are perceived as being embodied in the infinitely diverse concrete manifestations of their powerful agency. Although Valeri qualifies this distributed embodiment as a metamorphosis, it is certainly not a metamorphosis of the animist type, that is, a shift of perspective from physicality to interiority; it is, rather, a standard manifestation of the analogist all-pervasive process of expressive causality (to borrow one of Leibniz’s concepts describing the workings of his own analogist cosmology) whereby the components of a system are immanent to it and thus only acquire form, meaning, and function as a result of their relation to the structural whole that they jointly conform. This overdetermination of the parts by the whole (of the ontology of the domain by the principle that expresses itself in each of its manifestations) is a far cry from the random intersubjective identifications with individualized agencies from various provenances which characterizes animist ontologies. It seems to me that Hawaiian deities are no more animist “persons” than other analogist principles of totalization—the Christian God, Pharaoh, the Inka—who, although they may have a human embodiment, are mainly focal points of ontological reverberation, and thus not very amenable to direct intercourse. And even if Lono, in one of its guises, may eventually appear as Captain Cook, the domains that this deity and its fellow ontocratic generators encompass are not per se anthropomorphic, but composed of all that is necessary to furnish a world, including British naval officers and the ships that bear them.

As regards naturalism, finally, and the common use in the media, underscored by Marshall, of the rhetorical device of personification which might be seen as a degenerate form of anthropomorphism, I am reminded of sarcastic remarks made by Charlus in Le Temps retrouvé. The baron disliked the pedantic and chauvinistic articles on the war that Norpois published in the press, particularly for the latter’s immoderate use of the trope of personification, which Charlus viewed as a sort of self-realizing wishful thinking that verges on magic. How despairingly trivial by contrast with the hallucinatory metaphysics of animism!

A last word on Marshall’s figure (I won’t dare to comment upon Mauro Barbosa’s intricate Galois lattices). It struck me that while I organized my ontological grid as a structural group of transformation of the Goethean type (a development of a set of binary contrasts which contains initially all the potentialities that the morphogenetic process unfolds), he resorted to a tree-like diagram where ontological options are the product of a series of bifurcations following a logical, if not historical, evolutionary line of descent. It is probably an indication of our respective positions somewhere in-between the two poles of pure structuralism (nothing but contrasts) and pure historicism (nothing but events).


Descola, Philippe. 2013. Beyond nature and culture. Translated by Janet Lloyd. Chicago: University of Chicago Press.

Lemonnier, Pierre. 1990. Guerres et festins: Paix, échanges et compétition dans les Highlands de Nouvelle-Guinée. Paris: Éditions de la Maison des Sciences de l’Homme.

Peterson, Nicolas. 2011. “Is the Aboriginal landscape sentient? Animism, the new animism and the Warlpiri.” Oceania 81: 167–79.

Valeri, Valerio. 1985. Kingship and sacrifice: Ritual and society in ancient Hawaii. Chicago: University of Chicago Press.

Viveiros de Castro, Eduardo. 1996. “Os pronomes cosmológicos e o perspectivismo ameríndio.” Mana 2 (2): 115–44.


Philippe DESCOLA is chair of the “anthropology of nature” at the Collège de France and professor at the École des Hautes en Sciences Sociales, Paris. Following extensive research in Amazonia, Descola developed a comparative approach to relations between humans and nonhumans, and now studies the anthropology of images, non-Western and Western.

Philippe Descola
Collège de France
52, rue du Cardinal-Lemoine
F-75005, Paris, France