The difficult art of composing worlds (and of replying to objections)

This work is licensed under the Creative Commons | © Philippe Descola. ISSN 2049-1115 (Online). DOI: http://dx.doi.org/10.14318/hau4.3.030


The difficult art of composing worlds (and of replying to objections)

Philippe DESCOLA, Collège de France

Response to Hau Book Symposium on Descola, Philippe. 2013. Beyond nature and culture. Translated by Janet Lloyd with a foreword by Marshall Sahlins. Chicago: University of Chicago Press.

When reading Paul Valéry (1933) many years ago I was quite surprised to learn that the authority of the author does not exist and that “once published, a text is like an appliance which anyone can use as one pleases; it is not sure that its builder uses it better than anyone else,” an idea that Roland Barthes ([1968] 1984) later transformed into a famous manifesto.1 However, when I became an author myself, I realized how true this remark was and how futile it would be to try to redress (what I, as author, would see as) the misguided applications of the scriptural appliances I had produced. A few books later, I am even more entrenched in this conviction as I have seen time and again, not only that any form of rejoinder usually remained to no avail but also, and more interestingly, that the digestion of a text of mine by others often produced unexpected results more stimulating than the one I had intended to generate initially. This is the case in this remarkable batch of comments upon the English edition of my book Beyond nature and culture, comments to which I nevertheless feel compelled to answer—my previous remark notwithstanding—not only because such a reply is explicitly expected of me in several of the comments but also out of respect and gratitude for the time and intellectual energy devoted by each of the symposiasts to taking seriously a bulky volume that some of them probably would never have read of their own initiative. I will not attempt to sum up the argument of the book because several of the commentators have partly done so in their own terms—notably Bruce Kapferer, Marcela Coelho de Souza, and Michael Lambek—and it would be utterly absurd after endorsing Valéry’s proposition that the author “has no authority” to propose another version with the purpose of reestablishing an orthodoxy.

It is not an easy task, though, to address the widely diverse assessments of the book, themselves proceeding from a great variety of theoretical, epistemological, and anthropological standpoints and styles, to the extent that some authors reproach me with holding a position that others regret that I held, and vice versa. For instance, I wonder what Gérard Lenclud, who strongly opposes what he qualifies as ontological relativism, would reply to Christina Toren’s affirmation that each human develops over time a specific schematization of experience, or to Kapferer, who judges on the contrary that my epistemological stance amounts to an a priori universalist fifth ontology. It may be that my own theoretical positions are so clumsily expressed that they lend themselves to a number of contradictory interpretations; or, if I surmise that my readers dutifully apply to my propositions the principle of charity—and I have no reasons to doubt it—that the attempt to navigate between the Charybdis of a broadly defined phenomenological perspective and the Scylla of cognitive realism is bound to leave many critics unsatisfied.

I will attempt to respond to some of the observations by dividing them into two major categories. The first includes those comments that question, albeit in very courteous terms, the soundness of my whole enterprise, either because they detect in it a major epistemological (or logical) flaw or because, implicitly or explicitly, they do not share my view of what the anthropological endeavor is about. Lenclud illustrates the former branch of the alternative, Stefan Helmreich the latter. The second category of comments is much broader and includes all those that raise questions about what appears to be internal inconsistencies of the models I put forth in Beyond nature and culture, either because they do not accommodate certain phenomena (or accommodate them wrongly, or misrepresent them); or because the concepts and methods I use are misguided or inadequate—e.g., ontology, schema, intentionality, cognitive dualism; or because I leave unaccounted for some dimensions of the phenomena I pretend to deal with—ontological combinations, internal differences within ontologies, logical commensurability of the ontological schemas, et cetera. I will not try to answer exhaustively all these insightful comments, for lack of space obviously, but also because I believe that I have already dealt with some of them in other publications, probably too recent (Descola 2013, 2014a, 2014b) or too exotic (Descola 2010) to have attracted the attention of many readers.

What kind of anthropology is scientific?

In his latest book on the epistemology of anthropology, Lenclud has launched an attack on extreme anthropological relativism, that is, the idea that cultures are incommensurate and cannot be translated into one another (Lenclud 2013). Such a view is indeed easily refutable since, to proclaim that a foreign culture has no common premises with that of an observer, the observer must know enough of it to be able to state his claim. To substantiate his refutation of a variety of extreme relativism that nobody is willing to defend, Lenclud is thus forced to make up a conceptual straw man in the guise of an imaginary ontology very loosely derived from my definition of animism, which he then uses to demonstrate that such a construct is untenable, notably because metamorphosis—an uncommon but remarkable experimentum crucis of the conceptual structure of animism—goes against the universal principle of numerical identity. Animism, he argues, must also be more down-to-earth if I was able to gain sufficient knowledge of it to describe how it works. Lenclud’s comment for this Hau Book Symposium, “From one ontology to (an)other,” is a summary of this argument, to which I could feel free not to respond since the hypothetical animism that he dissects bears little relation with the animism I have tried to characterize and to which I and other authors have given ethnographic flesh. Nevertheless, Lenclud raises important epistemological and theoretical questions that need to be discussed because they reflect a basic divide as to what constitutes legitimate anthropological knowledge.

The first of them has to do with no less than “reality.” Lenclud holds a standard—let’s call it Searlian—view of social and cultural diversity, which is that peoples differ in their appreciation of the objective world because they have different perspectives on it: “they see the same things differently.” I contend that this is not the case for there is no such thing as a “thing,” a precut portion of the world that would stand as a given with all its properties readily decipherable by everyone, provided everyone was devoid of cultural prejudice. Let’s take a tree as the “thing.” We know, thanks to ethnobiological works on folk categorization, that the capacity of any human subject to categorize a plant—that is, to produce a judgment of specific identity—is dependent upon his or her familiarity with the plant world; what is perceived by an expert botanist as a forty- to fifty-year-old beech that has suffered from hydric stress and poor drainage due to a clay-laden soil will be just a “tree” for a juvenile city dweller. Is there an eternal essence of the beech, which would define prototypically its “thingness” and that humans would see differently according to their culture? Or is it not more plausible that the plant, or any other percept, is accessible to our knowledge as a set of clues that humans will detect or ignore according to basic inferences that they make about the qualities and types of behavior of objects in the world, inferences that they have learned to form during the process of their socialization? If that is the case, as I surmise, then peoples do not “see the same thing differently,” they actually see different things because the qualities they detect in the same object are dissimilar due to a personal or cultural variability in their attention to perceptual affordances.

But perhaps the example of the tree is too simplistic. For, most often, peoples will not see the “same things” in their environment because the ontological furniture of their worlds will be composed of very different “things.” An Achuar hunter cannot see a quark because a quark does not exist as a “thing” in the natural environment of anyone and is only detectable as an indirect clue thanks to highly complex machinery. It does not mean that the quark does not “exist”; it means that its ontic mode of existence is dependent upon its epistemic mode of existence, and that it thus cannot exist in the ontological furniture that composes the world of an Achuar. Conversely, it is doubtful that a physicist working at the CERN Large Hadron Collider near Geneva will be able to see an Iwianch—an Achuar spirit of the dead—because an Iwianch does no more exist as a “thing” in the environment than a quark does; it, too, is only detectable as a trace, and by the means of a complex set of phenomenological clues that will enable a person who has been trained to identify them to infer its presence. It does not mean that an Achuar, properly trained in physics, would not be able to “see” a quark; or that a physicist, after spending a few years living with the Achuar, would not be able to detect the presence of an Iwianch. It only means that, in normal circumstances, the Achuar and the physicist live in worlds that are different because they are peopled by different beings whose existence is predicated upon different ontological premises. Animism and naturalism are just heuristic models that I derived and systematized from experiences of that kind, not self-contained glass jar into which peoples are locked up.

The problem with an epistemology that is interested exclusively in the conditions of truth of statements is that it is useless, not only for saying anything relevant about ordinary knowledge but also for saying anything relevant about sciences, such as anthropology, which deal mainly with ordinary knowledge. It may be the case that “all humans have a concept of objectivity,” but who makes reflexive use of this concept except philosophers when they legislate on epistemological questions? More relevant than the vericonditionality of a proposition, at least for an anthropologist, appears to be its condition of felicity, for most of the statements that capture the attention of anthropologists are not apodictic, or even constative, but epideictic or performative, at least implicitly. They do not describe “objective reality”; rather they yield clues as to the state of mind with which people apprehend the world and act upon it. If I leave a house with an Achuar on a sunny day and he says that it is raining, I can undoubtedly judge the condition of truth of his statement, albeit disquisitions on this kind of situation are only relevant in the thought experiments upon which analytical philosophy sustains itself. A more common situation is one where, leaving the house, and seeing a bird of a certain species flying toward the east, he will say, “it will rain tomorrow.” And here vericonditionality is senseless because it does not matter much that the proposition will reveal itself true or false, in the same way that it does not matter much that the quark physicist may disprove the existence of Iwianch. In the end, I was not a little surprised by Lenclud’s long developments on questions of epistemological propriety, because they seem unnecessary in view of his concluding admission that the four ontologies are different “sciences of the world,” that is, “theoretical versions of the ways in which human beings schematize their experience.” Indeed they are just that: no more, no less.

What kind of science is anthropology?

I am personally a great fan of STS literature (although easily bored by sci-fi) and a number of my doctoral students actually study scientists and labs of all persuasions in different countries; so I was surprised to read Helmreich’s general statement that “French anthropology” seeks to “discern universals of the human condition” by appeal to “putatively precontact small-scale tribal or village society.” It makes us look like a cohort of Griaulian specters, complete with pith helmets and Kantian prejudices, haunting the gloomy corridors of the musée de l’Homme, or worse still, the kitsch halls of this new temple of primitivism, the musée du quai Branly.2 This nostalgic addiction to a form of neocolonial scientism would explain our indifference to the axes of differences—race, gender, sexuality, ethnicity, nation, class— that have righteously occupied the attention in more progressive anthropological communities.

I cannot speak here for “French anthropology” (roughly a thousand individuals, a majority of them describing themselves as women, and doing fieldwork on every possible topic in every possible part of the world) nor will I attempt to explain why the word “gender” only appears once in Beyond nature and culture,3 since Helmreich himself very fairly did so by referring to my own discussion of the notion in a previous article (Descola 2001). What he failed to see, though, are the more basic reasons why the book makes scant reference to class, gender, race, et cetera. Like most French anthropologists, I am as much interested in accounting for “differences” between social circumstances as are North American and British anthropologists, but I don't see these differences in the same light. I have adopted long ago a basic prescription of structural analysis that acknowledges that sets of phenomena can be brought together, not in spite but in virtue of the differences they exhibit, differences that one then attempts to order and to systematize, for instance in a combinatorial matrix or a group of transformation, that is, in a model that allows a study of the properties of the elements within the set and of their relations. Now the “axes of difference” along which these elements are organized are not the Western vernacular categories that Helmreich mentions—gender, class, et cetera—that have indeed been for quite a while the main descriptive tools of North American anthropology. Rather, I take as axes of differences conceptual systems of contrasts such as the various modes of identification and of relation that I have put forth, as well as the properties they are predicated upon. Imperfect as they may be, these structuring principles are probably less context-dependent than the categories more common in the Anglophone social sciences. For race, gender, ethnicity, nation, class (and one could add society, history, or the state) are categories the genealogy of which can be traced to the Euro-American modernization process of the eighteenth and nineteenth centuries, categories that have evolved from the status of reflexive tools for the objectification of Western historical destiny to the status of all-purpose descriptive labels used by laypersons and social scientists alike. As such, they are both ethnocentric and self-referential in the sense that they express specific contrastive features and contentious issues proper to modern collectives and couched in the language of what I call naturalism.

If there is at least one common purpose in the various approaches that have been subsumed under the label of the “ontological turn,” it is precisely our attempt to do away with those Eurocentric categories and with the colonial project of sucking into our own cosmology peoples who, having lost their land, their dignity, and the control of their work-force, face the added ignominy of having to translate their ways of life into our own way of life and of being grateful to us for providing them the tools to do so. The purpose, then, is to reproblematize the so-called social sciences without constantly paying lip-service to race, class, and gender, and to forge new concepts that would attempt to account for patterns of life, as well as for racism, androcentrism, and economic domination, without being subservient to the ways these processes have been conceptualized in anthropocentric accounts of Western history. Perhaps it is a project that borders on science fiction, as Helmreich argues in his concluding remarks, but one of the best ways to escape from the myopia of presentism is precisely to experiment with utopia.

What kind of phenomena should anthropology deal with?

We may now enter into the discussion of some of the more specific issues raised by the symposiasts. The first one has to do with the question of whether the analytical models I put forth are fine-tuned enough to attend to all the phenomena I claim they can account for. For instance, Stephan Feuchtwang argues, after G. E. R. Lloyd (2012), that analogism may be too broad a category to accommodate on an equal footing the cosmologies of prenaturalist Europe and of China: the former is a world composed of elements in delicate balance while the latter is a world animated by basic processes of change and devoid of preexisting singularities. An empirical answer would require more space and expertise than I can muster, not to mention that Anne-Christine Taylor has already responded cogently to the more theoretical dimension of the question when she discussed Lloyd’s book and his assessment of my attempt to put on the same plane China and Ancient Greece (Taylor 2013).

I will thus content myself here with a more general comment on the question of the appropriate focal lengths adapted to different types of comparison. As Feuchtwang notes, he is himself familiar with a narrower kind of comparatism than the one I engage in, namely that between China and Europe, in itself a long-standing specialty in Western scholarship if one cares to date it back to the Jesuits writings of the eighteenth century. The question is then: at what level does a difference become relevant according to the type of contrast that one wishes to emphasize? It is obvious that there are multiple and basic differences between the cosmologies of China and Ancient Greece—as there are also fundamental disparities between those and other analogist cosmologies, such as those of the Aztecs or of the Chipaya of Bolivia. These disparities bear upon the number and the nature of the beings identified, the forms of relation that they maintain, the types of network they constitute, the keys that insure their interoperability, the way they come (or not) into existence, et cetera. Nevertheless, when compared with cosmologies that correspond to a totemist or an animist mode of identification, the internal differences within analogism tend to fade as they can all be reduced to a general dialectical relation between multiplicity and totalization. All these cosmologies are not identical; but, at the specific scale chosen to apprehend them, they resemble each other in the sense given to resemblance by Lévi-Strauss, that is, as a particular case of difference where the difference approaches zero.

The comparative issue that seems to worry Feuchtwang—for instance when he writes that ontology is inadequate for China since it presupposes the existence of being—is also one of vocabulary. So, at the cost of repeating myself, let me make clear that ontology is not taken here as the science of being expressing itself in Greek—to borrow Heidegger’s qualification of philosophy; an ontology is simply a concrete expression of how a particular world is composed, of what kind of furniture it is made of, according to the general layout specified by a mode of identification (animism, totemism, etc.). By contrast with the latter, a cosmology (like those of China or Ancient Greece) is simply the form of distribution in space of the components of an ontology and the kind of relations that conjoin them. If one were to make an analogy with political systems, the mode of identification would be the general form of government—monarchy, democracy, aristocracy, et cetera. Ontology would be the type of constitution that specifies in each of these regimes the balance of power, the nature of the assemblies, the forms of representation; while cosmology would correspond to the set of legal and regulatory texts that govern the life in common. Seen from that angle, monarchies do form a more relevant unified set, in spite of their internal differences, than the charts governing property rights or taxes.

What kinds of concepts should anthropology make use of?

Another set of comments addresses the relevance of some of the concepts and methods I use by questioning either their philosophical orientation of their epistemological adequacy to the object I employ them for. Kapferer thus challenges the Husserlian regressive experiment upon which I have founded the deployment of the four modes of identification, that is, the idea that a transcendental subject immersed in the world without previous knowledge of it can only avail itself of its body and its intentionality to distinguish between self and nonself. According to him, such a position smacks of atomistic reductionism, even downright Eurocentric individualism, and thus runs counter to the holistic trend that he favors himself and that was illustrated by previous French anthropologists, notably Louis Dumont and Claude Lévi-Strauss, whose project he sees me nevertheless as pursuing in other respects. This is a recurrent criticism, although usually less cogently formulated for lack of philosophical expertise, and it calls for an explication de texte, all the more necessary since Kapferer’s judgment on my intellectual filiation is accurate.

Probably because of my own philosophical training—heavily influenced, as Kapferer shrewdly noticed, by a teratological combination of Marxism, structuralism, and phenomenology—I felt the necessity to establish the ontological matrix of the modes of identification upon a transcendental foundation, much as Lévi-Strauss had done when he anchored the development of the forms of marriage alliance in the primordial scene of the shift from nature to culture. Although developmental psychology also provided resources that allowed one to conceive the duality of accesses to percepts (and in spite of Edmund Husserl’s own stern strictures against “psychologism”), it appeared to me that the phenomenological idea of an antepredicative experience of the world offered a sound substratum for a radical reworking of the concepts and objects of anthropology. Is my hypothetical transcendental subject a crypto-bourgeois individual? I think not for at least two reasons. First, because this is decidedly not the case in Husserlian phenomenology. There is no primacy of the Self because the transcendental subject is relational ab initio thanks to its being an intentional agent, thanks to its capacity to pair with an aliud, i.e., with an alter that is as yet indeterminate.4 By contrast with Descartes’ transcendental subject, who is obliged to seek the ultimate foundation of objectivity in divine truth, Husserl’s transcendental subject accesses objectivity by the means of intersubjectivity. As Paul Ricoeur very aptly puts it in his comment of Cartesian meditations, “Husserl transcends the Ego by the alter ego” (1954: 77, my translation). The second reason is also epistemological but in another sense. In any foundational work in anthropology, it seems reasonable to start from universal features of the human species, which are necessarily lodged in the consciousness, body, brain, capacities, dispositions, et cetera, of an individual. Even in Gibsonian or Varelian approaches, these features are necessary for enaction or for the actualization of affordances (I hope Lenclud will feel reassured). Representations, emotions, intentionality, memory, pain, pleasure are first and foremost experienced by individuals, often in empathy with other individuals; they do not float in ether nor are they flowing out of institutions. I hope that it is clearer now why this egological stance has little relation with Eurocentric possessive individualism as defined by Macpherson or Dumont, for instance.

But there is more. My use of a transcendental subject is in some ways ironical5; but it leads to a sole objective, which in turn is very serious: it generates a device, the ontological matrix, which allows me to treat norms, practices, institutions, collectives, epistemic stances, and the like in an absolutely holistic way, that is, as transformations of each other, and not as the results of the whims and desires of conflicting human individuals (which, by the way, is what history does). So instead of going from wholes to parts through the tortuous paths of historical agency—which I understand is what Kapferer advocates—I try to go from (transcendental) parts to (structural) wholes thanks to an unexpected bypass from egology to ontology.

Toren objects also on epistemological grounds to the way I use a concept, namely that of schema, arguing that I ignored the autopoietic and self-regulating dimensions of this highly complex class of structure as defined by Jean Piaget, and reduced it instead to a stabilized mechanism for integrating the experience and knowledge acquired and developed within a certain milieu. The schema is a tricky concept because, as Piaget insisted, after Kant and Bergson, this mysterious object is not directly perceptible, so that its presence and its nature can only be inferred by what is regarded as its consequences, that is by the repetition of a physical or cognitive action in similar circumstances. However, to view Piaget’s schema as an exclusively “dynamic, self-producing system” that would be basically nonrepresentational appears to me as an oversimplification of his views. For Piaget, schemas organize a variety of functions (action, thought, interpersonal relations, regulation of bodily activities …) and they are thus capable of assimilating information in a variety of ways: reproductive (the repetition of some activity), recognitive (recognizing objects by giving them a meaning) and generalizing (which implies indeed a differentiation according to new situations).6 So, for Piaget at least, the autopoietic process is only a part of the story, and it goes along with mechanisms of assimilation and of repetition of information obtained from the environment.

However, getting Piaget right is not what is at stake here. The main question is whether Toren’s own definition of the schema is appropriate for any anthropological project that would attempt to make sense of “patterns of culture,” to use an old-fashioned terminology. Because, if it were true that “ideas and practice are not transmitted,” if it were true that the only continuity of a human person through time is that of “a self-regulating transformational system,” how then would we go about studying the cultural and social practice of these aggregates of totally idiosyncratic beings who never reproduce patterns of thought and behavior nor maintain an ontological stability through time? Toren’s answer is this: “by demonstrating the historical processes that continue over time to give rise to ontologies.” Very well, and how do we proceed to do that? By studying the way “historical processes” (whatever that is) affect each individual (but this, surely, would be even more reductionist than what I do)? By studying ethnographically the differential ontogeny in a given context? And in the latter case—which is what Toren accomplishes herself excellently—how do we move from the vernacular categories rendered analytical by the ethnographer in a specific place to the vernacular categories rendered analytical by the ethnographer in another place, except by trying to put forth analytical categories that could mediate between the two sets? This is precisely what I have attempted.

What kind of consistency should (my) anthropology aim for?

A last series of comments addresses neither broad methodological and epistemological issues, nor my arguable handling of debatable concepts; more importantly perhaps, they concentrate, from as it were an internal point of view, on certain problems raised by my theoretical models, pointing out that they incompletely achieve what they purport to realize. These perceptive comments mainly come from Lambek and Coelho de Souza and they are harder to answer because they accurately pinpoint difficulties of which I am aware myself and that I have been trying to supersede with various degrees of success. I will restrict myself to discussing two of these questions that are closely interdependent: one is methodological but has conceptual implications—is the ethnography I use a source for the ontological models or an illustration of them?—while the other is conceptual but has methodological implications—how do the models accommodate ontological hybridity?

The first question is thus put by Lambek: “what is the relationship of pure (ideal) types to actual life?” and it goes with his remark that the criteria I used to define the ontological matrix proceed from a priori rational hypotheses while the actual distribution of phenomena resulting from the operation of the matrix is historically-driven. This is probably inevitable. For a variety of reasons, and notably because of my growing annoyance with the kind of subjectivist and overinductivist approach typical of contemporary “ethnographism,” I chose to present the argument of Beyond nature and culture in a classical hypothetico-deductive format: an initial set of hypotheses organized along two major oppositions—one epistemic, with the contrast between interiority and physicality, the other logical with the contrast between identity and difference—are then systematically tested by examining how they account for the various states of a set of phenomena, namely the different ways of expressing continuities and discontinuities between humans and non-humans. This methodological choice was partly inspired by Lévi-Strauss’ famous dictum that “it is not comparison which founds generalization, but the reverse” (Lévi-Strauss 1958: 28, my translation); a remark that has often been found cryptic or paradoxical but that is only meant to emphasize that comparison is not a form of empirical discovery, an intuitive pairing of similarities and differences but a process of ascertaining a posteriori, by using a wide variety of historical material, what is already, if not absolutely known, at least hypothesized. I also confess to a certain esthetic partiality for highly formal modes of presentation such as the one Pierre Bourdieu and Jean-Claude Passeron (1970) adopted in their book on social and ideological reproduction, which is modeled on the geometrical order of Spinoza’s Ethics, itself inspired by Euclid’s Elements. Much as with my toying with a transcendental subject, there was thus an element of irony in mimicking a demonstration more geometrico, although few readers have perceived it.

However, it is obviously only at the cost of an oversimplification that a stark distinction can be made between ethnography and anthropology as opposing, on the one hand, a descriptive, inductive technique dealing with observed facts to, on the other hand, a hypothetico-deductive science dealing with models. For the logical construction of my argument in Beyond nature and culture is not entirely immune from the preliminary procedures that made it possible and that the internal economy of my demonstration, and an esthetic addiction to symmetry, led me to render invisible.7 Actually, it was by a series of inductive generalizations from ethnography, mine and that of others, that the transformational model progressively took shape because, as is the case for all anthropologists with a formal turn of mind, it is only through a long established familiarity with all sorts of ethnographical literatures that the particular social and cultural features that will become relevant within a model can be retrieved from one’s own memory. And one is able to deal expertly with ethnographic information thanks to another kind of expertise that is proper to our trade and is extremely difficult to formalize, namely fieldwork experience, which renders immediately familiar to us the rarely explicit procedures of the objectification process by the means of which other ethnographers have collected, filtered, and presented their data. So Lambek is right: in Beyond nature and culture there is a constant to-and-fro movement between induction and deduction, formal models and historical descriptions, direct and mediate knowledge, which is inevitable in anthropology and makes of this enterprise as much an art as a science.

Through the deceptively simple device of asking why I had dropped from Beyond nature and culture an analysis I had previously propounded of the combination existing among the Brazilian Bororo between animism and totemism, Coelho de Souza poses the most formidable questions of all: how can the orthogonal contrasts of the matrix—which only allow for a system of large indivisible differences—be made to accommodate a composition of distinct ontological principles that necessarily entails a set of narrow differences? Indeed, it is not sufficient to argue, as I did, that structural or historical combinations between ontologies can take the form of a hierarchical encompassment—which, strictly speaking, can only apply to the relation of an element to the whole—or of a Marxian formal subsumption, whereby a dominant ontology would subordinate to its reproduction a dominated one, because these types of articulation are merely accidental while, if I were faithful to my premises (and to a structuralist inspiration), they should have been built as potentialities into the very structure of the initial set of contrasts.

There is no easy answer to that formidable question. The directions I have been following to overcome this difficulty are globally resonant with Coelho de Souza’s own suggestion when she advocates a “reciprocity of asymmetric perspectives”; they imply a possibility of requalifying the terms by allowing them to take a wider variety of positions, in particular by introducing in the modes of identification the perspectival dimension, which I had hitherto confined to the modes of relation. Beyond the flexibility that an interplay between symmetry and asymmetry would introduce, notions of identity and difference in value, function, and context are also called for, much in the light of the complex permutations and internal dynamism that Lévi-Strauss’ canonical formula allows. But the extension of the field of relational positions for the initial terms of the matrix cannot simply take the form of a transcendental deduction—a demonstration based solely on principles; it must make use of empirical deductions—demonstrations based on facts—which is another way to restate my previous insistence on the complementarity between logical consistency and ethnographic relevance.

A way to do that which I have been following these past years is exploring ontological hybridity as expressed in images. Complex combinations between animism and totemism can be found elsewhere than among the Bororo, in particular among native societies of the Northwest Coast such as the Tsimshian. Now, not unlike the dualist Platonician philosophy of the Bororo that J. C. Crocker describes so vividly, and where the domains of animism and totemism do not define classes of beings but chart an opposition between essences and processes, the Tsimshian endow with animist or totemist qualities different types of images—for instance masks or rattles in the first case and heraldic poles or headlets in the latter—that often depict the same referents, say an otter, a raven, or an eagle; these images are thus distinguished not so much by what they figure than by the conditions—spatial, temporal, social, ritual—of their production and their use. The variation in the ontological predication of images then becomes an effect of the inflection that the terms receive when they are displaced in a different pragmatic setting. I hope that the book I am presently writing on the ontology of images will render these processes clearer and provide a more encompassing framework for treating the question of hybridity.

For a short conclusion, I would like to return to Valéry’s remark on the text as an appliance that anyone can use and to which the author has no privileged access for he is not necessarily in the best position to use it efficiently. If I find this remark profoundly true it is not because I want to distance myself from what I wrote—for this I claim full responsibility—nor out of a misguided sense of false modesty; it is because I find fascinating the actual uses that a score of authors have made of the models I put forth in Beyond nature and culture, all the more so because most of them are not anthropologists. Whether philosophers, historians, sociologists, literary scholars, or archaeologists,8 they have sometimes transformed beyond recognition the tools that I had provided, thus giving them a new purchase on a new domain and, perhaps, expanding their life-span. In that respect, as Kapferer aptly puts it, “the proof is in the pudding.” As it always should be.


Apostel, Leo, W. Mays, Albert Morf, and Jean Piaget, with the collaboration of B. Matalon. 1957. Études d’épistémologie génétique IV: Les liaisons analytiques et synthétiques dans les comportements du sujet. Paris: Presses Universitaires de France.

Barthes, Roland. (1968) 1984. “La mort de l’Auteur.” Le bruissement de la langue. Paris: Seuil. Initially published in Manteia.

Baschet, Jérôme, Jean-Claude

Bonne, and Pierre-Olivier

Dittmar. 2012. Le monde roman: Par delà le bien et le mal. Lyon: Editions Arkhé.

Bourdieu, Pierre, and Jean-Claude

Passeron. 1970. La reproduction: Éléments pour une théorie du système d'enseignement. Paris: Éditions de Minuit.

Descola, Philippe. 2001. “The genres of gender: Local models and global paradigms in the comparison of Amazonia and Melanesia.” In Gender in Amazonia and Melanesia: An exploration of the comparative method, edited by Thomas Gregor and Donald Tuzin, 91–114. Berkeley: University of California Press.

———. ed. 2010. La fabrique des images: Visions du monde et formes de la représentation. Paris: Somogy and musée du quai Branly.

———. ed. 2013. The ecology of others. Chicago: Prickly Paradigm Press.

———. ed. 2014a. La composition des mondes: Entretiens avec Pierre Charbonnier. Paris: Flammarion.

———. ed. 2014b. “Modes of being and forms of predication.” Hau: Journal of Ethnographic Theory 4 (1): 271–80.

Descombes, Vincent. 1989. La Philosophie par gros temps. Paris: Editions de Minuit.

Lenclud, Gérard. 2013. L'universalisme ou le pari de la raison. Paris: Hautes Etudes, Gallimard/Seuil.

Lévi-Strauss, Claude. 1958. Anthropologie structurale. Paris: Plon.

Lloyd, G. E. R. 2012. Being, humanity, and understanding: Studies in ancient and modern societies. Oxford: Oxford University Press.

Montesquieu, Baron de. 1950. “Pensées.” In Œuvres complètes, edited by Louis Desgraves Paris: Nagel.

Ricoeur, Paul. 1954. “Étude sur les “Méditations Cartésiennes” de Husserl”. Revue Philosophique de Louvain. Troisième série 52 (33): 75–109.

Serres, Michel. 2009. Écrivains, savants et philosophes font le tour du monde. Paris: Le Pommier, Les Essais.

Taylor, Anne-Christine. (2013). “Distinguishing ontologies. Comment on Being, humanity, and understanding (G. E. R. Lloyd, 2012).” Hau: Journal of Ethnographic Theory 3 (1): 201–4.

Tournay, Virginie. 2014. Penser le changement institutionnel. Paris: Presses universitaires de France.

Valéry, Paul. (1933) 1957–1960. “Au sujet du Cimetière marin.” Œuvres I. Paris: Bibliothèque de la Pléiade, Gallimard.

Wengrow, David. 2014. The origins of monsters: Image and cognition in the first age of mechanical reproduction. Princeton, NJ: Princeton University Press.



Philippe Descola
Laboratoire d’Anthropologie Sociale
Collège de France
52, rue du Cardinal-Lemoine, 75005 Paris, France


1. My translation; the full passage is as follows: “il n'y a pas de vrai sens d'un texte. Pas d'autorité de l'auteur. Quoi qu'il ait voulu dire, il a écrit ce qu'il a écrit. Une fois publié, un texte est comme un appareil dont chacun se peut servir à sa guise et selon ses moyens: il n'est pas sûr que le constructeur en use mieux qu'un autre” (Valéry [1933] 1957–1960: 1507; Barthes [1968] 1984: 61–67).

2. A museum that, by the way, was bold enough to host in 2010–2011 an exhibition that I curated, “La Fabrique des images,” where Yup'ik masks coexisted with, among others, Dutch seventeenth-century landscapes and impressionist paintings, Roman sculptures, and a score of scientific images, including pet-scans of the brain and chronophotographies of E. J. Marey.

3. Apart from the obvious fact that the book under review is a translation and that the English word “gender,” only lately translated in French as “genre” in feminist studies, is in fact a good equivalent of the French “sexe” (hence the title of Simone de Beauvoir’s essay, “Le Deuxième sexe”), a word that for at least three centuries had denoted the social and cultural dimensions of one’s gender identity; see, for instance, the famous (antifeminist) aphorism by Montesquieu “il n'y a plus qu'un sexe, et nous sommes tous femmes par l'esprit” (in Desgraves 1950: pensée n° 1062). The word “sexe” taken in that sense appears 26 times in the French edition of Beyond nature and culture.

4. A useful distinction which I borrow from Vincent Descombes (1989: 89).

5. Manipulating a transcendental Ego is thrilling—it must be like driving a Ferrari.

6. See the synthesis that Piaget provides of these processes in chapter 3 (“Transposition du problème de l'analytique en termes génétiques”) of Apostel et al. (1957).

7. Except in chapter 6, on animism, where I retrace my conceptual itinerary and the reasons that made me evolve from a previous triadic model (animism, totemism, naturalism) to a quadratic one (with the addition of analogism) where all the terms are logically redefined in relation to one another.

8. See Serres (2009); Tournay (2014); Baschet, Bonne, and Dittmar (2012); or Wengrow (2014) for a few, widely different, examples.