An exchange between Philippe Descola and Bruno Karsenti

Translated by John Tresch

The editors of HAR are happy to present a recent lecture by anthropologist Philippe Descola, followed by an exchange with philosopher of the social sciences Bruno Karsenti. Beginning with current and historical relations between philosophy and anthropology in France and beyond, Descola compares strategies of ethnographic generalization including those advocated by Evans-Pritchard and Lévi-Strauss. Highlighting structuralism’s use of deductive logic, he contrasts the models of “transformation” offered by the morphological studies of Goethe and D’Arcy Wentworth Thompson. Karsenti suggests that Descola’s emphasis on “modes of symmetrization” points toward a decentered universalism which gives anthropology unique relevance today, particularly for the environmental crisis (by serendipity, HAR is simultaneoulsy publishing a review of Descola’s newly translated, ecologically-focused book of interviews, The Composition of Worlds). While the data-processing ambitions of the Human Relations Area Files make an appearance, both authors stress the distinctively sensory and experiential basis of anthropology’s philosophical engagements.

The exchange was held on 21 January 2023, hosted by the Société Française de Philosophie at the University of Paris 1, Sorbonne. Many thanks to Philippe Descola and Bruno Karsenti for making this work available to HAR readers.

Contents:

1. Philippe Descola, “Anthropology and Philosophy: How to Symmetrize Ontologies”

2. Response by Bruno Karsenti, “Decentered and Neo-Universalist”

3. Philippe Descola, “The Flavor of the Ordinary”

Anthropology and Philosophy: How to Symmetrize Ontologies

by Philippe Descola

During my last visit to my thesis supervisor, Claude Lévi-Strauss, just before leaving for the field in the Ecuadorian Amazon, I spoke to him at length—probably too much—about the research strategies I was going to employ and the methods I had prepared to obtain the information I needed to carry out my thesis project: to describe and analyze the material and intellectual techniques by means of which the Achuar Indians socialize nature. He listened to me patiently, and ended our conversation with this advice: “Let yourself be carried along by the field!”

What did he mean? How could I possibly incorporate into my conscientious plans what sounded like an invitation to let everything go? It was only after a few months in the field that I began to understand his remark, because, deep down, no one had prepared me for the job of ethnographer. Above all, no one had taught me how to move from the philosophical abstractions I’d been fed to the concrete tasks of participant observation, and from these again to the anthropological generalizations I aspired to obtain. My lecture will reflect on Lévi-Strauss’s elliptical remark.^[1]I already commented on this remark by Lévi-Strauss in another context (see: Descola, “On anthropological knowledge,” Social Anthropology 13, no. 1 (2005): 65-73), as did my wife and fellow anthropologist Anne-Christine Taylor who was also present during this interview with our doctoral supervisor (Anne-Christine Taylor, “Anthropology Comes In When Translation Fails,” Social Anthropology 30, no. 1 (2022): 91-103). One possible definition of anthropology is that it is a way of deriving philosophical lessons from anecdotes; in this sense, my talk will be thoroughly anthropological.

Let’s start by returning to Lévi-Strauss’s advice. The question it raises is roughly as follows: When anthropologists study practices, ways of thinking and institutions that are apparently very different from those in which they grew up, what correspondences exist between the information they gather (statements and actions that are often disconnected from one another), the affinities they feel or sense between the discursive and behavioral style they observe and the modes of conceptualization with which they are familiar or have learned to appreciate, and the greater or lesser degree of reflexivity that is at work in the propositions on which they rely? In short, through what bricolage is anthropological discourse constructed—from the parachuting of an individual into an environment often far removed from their familiar universe, through to the general propositions they believe they are entitled to draw from this ethnographic experience, most often combined with that of colleagues?

The question is all the more acute in the French context since, as we know, a large number of anthropologists since the beginning of the twentieth century were initially trained to be professors of philosophy, a situation that stands out from other major anthropological nations. In a way, “speaking philosophically” comes to us spontaneously, even when—and this is the most usual case—the choice to take up the anthropological vocation was partly fueled by a certain disenchantment with academic philosophy, i.e., with a way of thinking that is above all concerned with the exegesis of its own history, and therefore generally indifferent to questions that have arisen elsewhere, and in terms that, for a long time, it has not sought to understand. Fortunately, the situation is changing, and I interpret this invitation to give a lecture to the Société Française de Philosophie as just one symptom of the interest that philosophers are now showing for anthropology.

I spoke of symmetrization, not symmetry. Indeed, this work of placing things into correspondence is doomed to remain unfinished, since its final product is conditioned by the audience for which it is intended: professional anthropologists, and amateurs of reflective thought trained by two and a half millennia of the European philosophical tradition, who must be addressed in a language they are likely to understand. This incomplete symmetrization can also take very different forms, depending on the types and modalities of transfer between the local ideology or ideologies and that of the analyst. We can distinguish three main types of symmetrization.

The most common, and the oldest, is to develop the conceptual implications of a local institution in such a way that its field of application goes beyond both that of the original institution and the particularities of the region in which it was initially described. In the early days of the discipline, such generalization was undertaken by extending the meaning of an indigenous concept to encompass a host of disparate phenomena whose only common feature was that they did not fit with Western understandings of the kinds of practices that the concept was supposed to explain. “Totem,” “mana,” “taboo,” “shaman,” and “hau” were born in this way. With positive effects, too, since what had previously been seen as laughable or reprehensible superstitions were thereby transformed into philosophical problems or intellectual categories worthy of serious consideration. In more recent times, this type of generalization has been achieved by unfolding the conceptual consequences of an institution, a process, a regime of relations, or an epistemic orientation identified through ethnographic observation. Rather than an unlimited expansion of an initially imprecise meaning, the aim is to deepen and operationalize a very precisely defined meaning. Louis Dumont’s hierarchical encompassment, Marylin Strathern’s to-and-fro between objectification and de-objectification of relationships, and Eduardo Viveiros de Castro’s perspectivism—theoretical constructs initially intended to account for dispositions specific to a particular cultural area—are now classic examples. In any case, the originality of these local models which have become paradigmatic, and the very principle of their constitution, result from the implicit contrast they present with Western ways of perceiving and conceptualizing the field of phenomena these models account for. Here, the generalization of a relative phenomenon relativizes a general principle.

Let’s move on to the second form of symmetrization. This one consists in transforming indigenous thought into a more or less systematized corpus analogous to a philosophical doctrine, at least in its mode of presentation. This too is an old tendency, even older than the previous one, since it was originally characteristic of a certain missionary anthropology. The Historia general de las cosas de la Nueva España compiled in Nahuatl by the Franciscan Bernardino de Sahagún is undoubtedly the earliest example, and the Lettres édifiantes et curieuses of the Jesuits in China is the most famous variant, particularly in view of the important role the latter played in Leibniz’s thinking. A more modern expression of this tendency is Father Placide Tempels’ famous Bantu Philosophy (Tempels, 1959 [1945]), which gave rise to a wide-ranging debate among philosophers from Black Africa— some of whom, like Alexis Kagame and Mubabinge Bilolo, are prepared to find in the traditional discourse of such and such an African civilization the elements of a genuine philosophy, whereas the Beninese philosopher Paulin Hountondji, for example, sees little more in what he disdainfully calls “ethnophilosophies” than classical ethnological studies of African representations of the world and the individual (Kagame 1976; Bilolo 1987; Hountondji 1970).

Although the controversy surrounding alternative metaphysics has mainly raged in Africa, it has also spread to Europe and North America, as witnessed by the debates that greeted the publication in 2009 of Cannibal Metaphysics, the essay in which Eduardo Viveiros de Castro, on the basis of a philosophical reconstruction of Amerindian thought, called into question the forms of objectification of Western epistemology (Viveiros de Castro 2009). This kind of question is not absent from ethnological work proper. Usually, it is visible in the clearly recognizable philosophical colors in which an ethnologist depicts the moral or epistemic dispositions of the society he is studying. Examples abound in France of the influence of philosophy, particularly Husserlian phenomenology, on the training of the first generation of field ethnologists. Whether this influence was direct, as with Leenhardt, or indirect, as with Griaule, it had the effect of proposing an epistemic paradigm that, by running counter to the dominant cognitive realism, seemed to be in agreement with modes of knowledge and presence in the world that ethnographers were discovering far from home. This is one of the reasons why phenomenology was so widely embraced by anthropologists from very different national intellectual traditions, such as Ernesto de Martino in Italy, Marcelo Bórmida in Argentina, and Irving Hallowell in the United States. In recent years, this movement has taken a more decisive turn in English-speaking countries, with the late discovery of Merleau-Ponty and, even more recently, the thought of Deleuze.

That said, although the invocation of philosophical concepts, and above all the authority of certain philosophers, has now become the norm in non-French-speaking ethnological production—in contrast to that of French-speaking ethnology, where on the contrary there is now an inversely proportional relationship between the degree of philosophical training and the quantity philosophical references—this trend should be seen as an ambiguous homage. Philosophical reference remains superficial and serves in reality to cover with an elegant conceptual veil the robust empiricism that is the price of seriously conducted ethnographic investigations. In fact, attempts to provide insight into “altermetaphysics” and to assess their subversive impact on our own way of philosophizing—whether by indigenous authors trained in Western philosophy, or by Western anthropologists who draw lessons from indigenous thought according to the expository canons of a philosophical work—all suffer from the same drawback: they remain an exegesis that wreaks havoc on the pragmatic conditions of enunciation and practice of the statements from which these ethnophilosophies are derived. Without going so far as to say, as Paulin Hountondji does about African ethnophilosophies, that they are “a mere pretext for learned disquisitions between Europeans,” we must nevertheless agree that symmetrization in this case remains imperfect (Hountondgi 1970, 122).

The third form of symmetrization aims neither to generalize the scope of a local principle nor to propose a philosophical counter-model inspired by indigenous thought, but to construct a combinatorial system that accounts for all the states of a set of phenomena by highlighting the systematic differences between its elements. This is a basic principle of structural analysis, well defined by Jean Pouillon when he writes: “Structuralism proper begins when we admit that different sets can be brought together not in spite of, but by virtue of their differences, which we then seek to order” (Pouillon 1975, 15-16). Why is this a symmetrization? Because, in good structural logic, the totalization is never given from the start, as if from some cosmic point of view anthropology could structure the world under its imperial gaze. Instead, totalization results from the always unfinished operation by which cultural traits, norms, institutions and qualities are constituted as variants of one another within a set or a whole (ensemble) which not only can be reconfigured differently if other elements are added, but has no other raison d’être than to subsume the variations of which it is the theater. This type of symmetrization is entirely dependent on the multiple properties that people detect here and there in phenomena, and therefore requires nothing in the way of an overarching perspective other than a bit of erudition about the diversity of the objects it deals with, as in any knowledge enterprise. The fact remains, however, that the symmetrization undertaken by structural combinatorics is just as incomplete as the other two ways of symmetrizing, but for different reasons. It requires a general knowledge about institutions and customs of others that only the West has produced; as a result, it remains dependent on a singular knowledge project, not so much because of its universal ambition—for elsewhere in the world, many knowledge systems have claimed to account for everything—but because of the demand for exhaustive empirical data from around the world on which its universal ambition relies.

It would seem that we are a long way from Lévi-Strauss’s advice on how to conduct fieldwork, with which I introduced my remarks. In fact, not at all. We have just reviewed three ways of symmetrizing knowledge about the other: the generalization of a local concept; the systematization of an indigenous thought promoted as a philosophical counter-model; and the integration of a wide variety of local forms of thought—including Western metaphysics—within a structural combinatorial system in which they are treated as variants of one another. Each of these efforts at symmetrization presupposes a different type of bifurcation taking off from the words the ethnographer hears and the actions he or she witnesses in the field, each a kind of line of legitimate inferences that unfolds in the direction either of exegetical deepening, of conceptual generalization, or of integration into a transformational system.

These modes of bifurcation starting from the experiential data are not philosophically innocent, for they depend closely on the logical operations by means of which anthropologists produce generalizations. The two most common operations can be called ethnographic generalization and ethnological generalization. Ethnographic generalization, the operation on which all anthropologists agree whatever their theoretical differences, is the field survey itself. It is comparative through and through. First, because of the more or less conscious comparison the ethnographer makes between the customs and practices he or she has witnessed and those prevailing in his or her milieu of origin. Second, because of the comparisons he or she is led to make between the way in which key elements of the society he or she is studying are approached in his or her survey and the approaches proposed by other ethnographers of the same society, or those of similar phenomena in different societies. Finally, ethnographic generalization is comparative due to the sorting the ethnographer makes of the observations made, retaining those he or she feels best represent an average state of the phenomena he or she hopes to characterize.

The second stage, ethnological generalization, can be defined as the extension of a schema initially identified by the analyst to account for the structure of a society he or she has studied in depth to other societies in the same cultural area, and possibly to societies with which it shares a family resemblance elsewhere in the world. These two forms of generalization are the same as those by which Evans-Pritchard defined the comparative method in his 1950 Marett Lecture; he called them respectively translation and abstraction (Evans-Pritchard 1950).

The question raised by these kinds of generalization is the following: in a science that places ethnographic inquiry at the heart of its approach to knowledge and as its primary motor, should comparison be seen merely as a sequence of inductive generalizations based on the case initially studied? Or might comparison take other forms than this kind of generalization by successive leaps from case to case—forms that would be less dependent on the simple intuition that there are similarities between, on the one hand, a family of societies or a type of institution and, on the other hand, a starting case arbitrarily chosen at the outset? This question has been widely debated, and this is not the place to go back over the criticisms, both internal and external to anthropology, that have been levelled at the inductive comparison by successive generalizations whose method Evans-Pritchard defined. Suffice it to say that these criticisms are of two kinds: on the one hand, those which point to the inadequacy of the comparative categories used to describe highly heterogeneous social and cultural situations; on the other, those concerning the identification of comparables, i.e. the elements which, in units that are already difficult to define—those abstractions we call societies—could lend themselves to comparison.

A new form of generalization, and more broadly a new way of dealing with social facts, emerged in the 1960s in response to the aporias of inductive generalization. The comparative operation as defined by Evans-Pritchard in the Marett lecture, and as practiced by the majority of anthropologists throughout the world without always realizing it, involves a three-stage rise in generality, starting with particular cases: first, fieldwork, conceived as a necessarily empirical exercise in translation, leading to a second stage, which identifies through abstraction the structure of the society described, enabling a third stage, the comparison of this structure with other structures, identical or different, obtained by the same methods. In this perspective, the second stage consists in matching an empirical substance—a society, a culture—with an underlying form (I borrow the expression from Evans-Pritchard) which the ethnographer isolates by abstraction. But, as is to be expected from an inductive approach, the form that emerges from the study of a society will be specific to the case from which it is abstracted. It will express that society’s presumed structure in a condensed formula. This means that the third stage of comparison, that of comparing the structures or forms of several societies from a typological point of view, has two consequences: either the forms compared will be so idiosyncratic that any comparison will ultimately prove impossible, or one of the particular structures will be declared to be the optimal form against which the others are compared.

This is why the structuralist wave of the 1960s—Lévi-Strauss and Dumont in France, Leach and Needham in the UK—was first and foremost a reaction against the method of generalization theorized by the British functionalists and widely practiced elsewhere. That method was undermined by its ambition of establishing generalities by empirically extracting out of disconnected observations general characteristics based on superficial similarities. Structuralists were at pains to point out that such an attitude rests on a naive realism that sees social and cultural facts as so many spontaneously given objects, or—and this is a correlate—sees societies as composed of naturally discrete elements. For structuralists, by contrast, in the words of Leach, “a society is not an assemblage of things, it is an assemblage of variables” (Leach 1961, 7). What structuralists compare are not pieces of societies, not institutions or practices, but variations within a set of phenomena whose nature the analyst has defined, whose limits he has set, and whose morphological discontinuities he has systematized.

This difference in the conception of the object under study implies a difference in the conception of the comparative method. While the initial phases of comparison are similar for functionalists and structuralists—what Evans-Pritchard describes as translation and abstraction are called ethnography and ethnology by Lévi-Strauss—the final stage—generalization for Evans-Pritchard, anthropology for Lévi-Strauss—differs completely. For Evans-Pritchard, this last stage is merely the culmination of a rise in inductive generality that will allow the social structures revealed by the preceding generalizations to be compared across a vast range of societies. For structuralist anthropology, on the other hand, the final phase is entirely divorced from the preceding ones, in that it is deductive rather than inductive, to use the vocabulary and concepts of John Stuart Mill, whose Treatise on Logic exerted a considerable influence on the French and British social sciences until the 1960s (Mill 1843). For the structuralist, the construction of anthropological results proceeds in three stages: the induction of laws, principles, or hypotheses; the deductive application of these laws, principles, or hypotheses to phenomena; and the verification of the legitimacy of this deduction by means of comparison. To do this, structuralist anthropology clearly dissociates on one hand the induction of abstract hypotheses by generalization from ethnographic data, and on the other, the construction of deductive models which enable these hypotheses to be tested by means of comparison. The fruitfulness of the structural approach therefore lies not in confirming the existence of universals which are supposed to explain the similarities between cultural phenomena—an approach characteristic of cognitive anthropology—but rather in its ability to highlight the invariants that account for a system of differences.

It’s worth recalling that anthropological invariants are not universals. Invariants are rules of combination based on elementary characteristics of the world and on the ways in which humans develop their existence in it, rules that are able to give rise to a finite number of possibilities, only some of which are realized in social life; whereas universals can be seen as cognitive and biophysical properties that have no exceptions, which makes them unproductive on the anthropological level for explaining variation. Invariants, and the models within which they are played out, do not serve to describe empirical situations and their transformations; they serve to highlight formal properties of social life, pushing to its ultimate consequences the necessary distance that exists between the concrete situations observed by the ethnographer and the conceptual vocabulary used to describe them.

In The Art of Anthropology, Alfred Gell observes that anthropologists navigate between two types of relationships: on the one hand, the “internal” relationships which make up the formal terminologies used in the profession, which express the links between terms in a conceptual system, and on the other hand, the relationships he calls “external”: those which ethnographers witness in the field and which connect elements in all sorts of ways—genetic, historical, causal, accidental, institutional, affective, etc. (Gell 1999, 33). “External” relations may be of similarity or difference, but in contrast to the relations that are “internal” to a theoretical system, they connect objects that are independent of each other because there are no necessary links between them. Borrowing Gell’s distinction between two kinds of relations, we can thus say that the inductive approach generalizes external relationships to internal ones (in other words, it transforms observed relations into scientific concepts—for example, ‘kinship’ or ‘hierarchy’), while the deductive method keeps these two sorts of relations separate. In fact, because inductive generalization à la Evans-Pritchard abstracts increasingly refined categories from similarities observed in the widest possible variety of ethnographic cases, it becomes impossible to build an analytically coherent formal system, since relations of similarity that are empirical and contingent cannot be converted into theoretical and structural relations of difference. Structural deduction approaches comparison in a completely different way: instead of ascending in generality from observable resemblances towards increasingly abstract categories that eventually lead to supposed universals, it detects in the particular cases themselves the clues to a closed system of differences, the modelling of which offers a pathway to understanding the formal properties of the set of phenomena we take as our object. It’s no longer a question of moving from a chaotic set of external relations to a system of internal relations by means of successive abstractions, but of discerning beneath the external relations that are presumed to form a system—the bonds of kinship and alliance, for example—the system of internal relations that offers the best possible model to account for the external relations. This movement is captured in Lévi-Strauss’s famous remark: “Comparison is not the basis of generalization, but the opposite” (Lévi-Strauss 1958, 28). In other words, comparison is not the means of making a discovery. It is instead the concrete development of a starting intuition based on the knowledge, latent or dormant as it were, of a large number of cases which were initially understood for their ethnographic flavor. To take a personal example, when I progressively discovered through induction during my fieldwork among the Achuar that they treated plants and animals as persons, and that, as a consequence, our own Western division between nature and society was irrelevant for them, I was provided with an initial nudge which led me to investigate in the ethnographic literature how people elsewhere conceptualize their relationships with non-humans (induction again). Becoming aware of the variety of these relationships thus led me to formulate a general hypothesis to which I shall return later at length: that the diverse types of perceived continuities and discontinuities between humans and non-humans can be organized according to a set of contrasts between interiority and physicality, a (deductive) guiding principle to detect systems of differences in the cultural conceptualization of how humans relate to non-humans.

In such a process, the study of variations—or what Lévi-Strauss calls transformations—becomes fundamental and even makes up the very heart of comparative work. Lévi-Strauss himself repeatedly points out that the notion of transformation is the keystone of the type of structural analysis he practices. While he borrows from linguistics the understanding of “structure” as a system of contrasting oppositions, he gives it an analytical dynamism, through its ability to organize the regular transformations between models of the same group, applying itself to the same set of phenomena. Structure cannot be reduced to a system, to a simple set of elements and the relations that unite them, for in order to be able to speak of structure, he specifies, “it is necessary that invariant relations appear among the elements and the relations of several sets, so that we can pass from one set to another by means of a transformation” (Lévi-Strauss 1988, 159).

Lévi-Strauss envisages this transformation in two very different ways in his work. As he states in his interviews with Didier Eribon, the notion of transformation stems from his wartime reading in the USA of the biologist D’Arcy Wentworth Thompson’s masterwork, On Growth and Form, from which he particularly drew the analysis of the differences between animal forms related by means of geometric transformations (Lévi-Strauss 1988, 158; Thompson 1961). And yet the illustration Lévi-Strauss provides of the fruitfulness of this approach is the study of kinship systems—a surprising choice, in that his use in this domain of transformation, as a variation starting from an original relation, is very different from the notion of a “transformation group” that he would later employ in the analysis of myths. In The Elementary Forms of Kinship (1949), Lévi-Strauss chose as invariant relation the exchange of women, an expression of the principle of reciprocity, itself the positive form of the prohibition of incest. All the forms of matrimonial union he considers in this book are transformations of this original principle; he studies each of these forms according to the increasing degree of complexity they present in relation to the simplest sociological expression of the principle of reciprocity, namely dual organization or the division of a society into moieties. Transformation as understood in this way differs in many respects from transformation as developed by Thompson, as I will explain below. Instead, this tree-like proliferation of marriage forms rather resembles the methodical variations of an Urform, in the sense of Goethe, in that case the exchange of women governed by the principle of reciprocity—whose logical consequences Lévi-Strauss unfolds into so many morphological types of matrimonial union. Just as Goethe cherished the dream of discovering the Urplanfz, the original plant, the prototype from which could be derived by transformation the full set of characteristics of all plant species—not only actual, but also logically possible—in the same manner Lévi-Strauss sees in the principle of reciprocity the originary form of all possible marriage alliances, whose law of development he proposes (Goethe 1790). And just as Goethe set himself in opposition to Linnaeus in botany because he rejected the idea of a static table of attributes, however exhaustive, in favor of the deduction of a principle of transformation of biological forms from an original complex combination, in the same way, Lévi-Strauss distances himself from Radcliffe-Brown in the field of social morphology when he considers as originary the structure that most fully develops the logical possibilities of the principle of reciprocity, rather than taking as his starting point the simplest forms of marriage from which complex forms would be derived.

However, in contrast to this Goethean conception of variation as the development of a single (if complex) prototype—the model of transformation which Lévi-Strauss implements in his studies of social morphology without ever, incidentally, making any direct reference to Goethe—Thompson’s conception of variation emphasizes the geometrical simplicity of the transformational grid that makes it possible to pass from one biological form to another without considering any original form from which the other biological forms might derive. Lévi-Strauss does also adopt this position, but in a distinct morphological domain: that of myths. His other reference to Thompson, in the “Finale” of L’Homme nu, makes explicit the very Thompsonian way in which he uses the notion of “transformation group” in his analysis of myths (Lévi-Strauss 1971, 604-6). In this context, a transformation group is the set of variants of a myth that retain the same structure, even when inverted. It can also be said that a transformation group is formed by the set of myths that are shown to transform one another by borrowing mythemes from each other whose motifs they invert, whose function they permute, or whose relationship between form and content they modify. So it is the set of myths between which these operations are detectable which by right constitutes the transformation group, and this may be made up of the myths of neighboring societies which, due to knowing each other, deliberately make a play of permutations; but it may also be made up of the myths of a much larger set, provided that a transformation between them remains possible.   

These remarks point to two sets of questions raised by the notion of transformation when it is used in the human and social sciences. The first is conceptual in nature, and concerns the distinctions to be made between the two procedures that the term transformation covers. Whether we are dealing with organisms, images, social types, or the semantic units of certain classes of utterances, the transformation of one form into another can be envisaged either more in the terms of a “Goethean regime” or of a “Thompsonian regime.” In the first case, following the method defined by Goethe in his Metamorphosis of Plants, transformation is the development into different forms of an initial plan, itself constructed by comparison of empirical objects belonging to the same set. In the second case, transformation is a deformation by continuous variation within a space of coordinates that are applied to forms already given. The latter method is the one Lévi-Strauss applies in his analysis of myths, while the former is the one he follows in The Elementary Forms of Kinship. The former is also the one that inspired my approach when I set out to organize the various forms of continuity and discontinuity between humans and non-humans on the basis of an original relationship between interiority and physicality. I’ll briefly recall here the guiding principles of this reasoning, in which four ontological types were deduced, starting from the variations authorized by an initial contrast (Descola 2005b).

The starting point for this exercise in structural ontology was the thought experience of a transcendental subject, inspired as much by the Husserlian idea of the pre-predicative conditions of knowledge of the world as by recent work in developmental psychology: I can only detect qualities in an aliud (i.e., an as yet undetermined alter) on the condition that I can recognize in it those qualities by means of which I apprehend myself, which belong both to interiority—mental states, intentionality, reflexivity—and physicality—physical states and processes, sensory-motor patterns, proprioception. This starting point is a hypothetical invariant; it is no more or less plausible as a foundation than the requirement for the exchange of women which Lévi-Strauss saw as the necessary and sufficient principle from which to deduce the different types of marriage systems. But instead of being constructed through the progressive complexification of the logical conditions for unfolding the invariant, as is the case in The Elementary Structures of Kinship, the transformation here is based on unfolding the combinations that the original relationship makes possible. There are four such combinations: either non-humans have an interiority of the same kind as mine, but differ from me, and from each other, in their physical capacities (animism); or, on the contrary, they undergo the same kind of physical determinations as those I experience, but they have no interiority (naturalism); or, again, humans and non-humans share the same group of physical and moral qualities, while being distinguished in bundles from other sets of humans and non-humans who have other physical and moral qualities in common (totemism); or, finally, each existing being is distinguished from the rest by its own combination of physical and moral qualities, which it must then be possible to link to those of others by relations of correspondence (analogism). In this case, as in that of kinship systems, the transformation is the hypothetico-deductive expression of all possible consequences of a core posited as invariant, i.e., a Goethean transformation. The definition of the starting point is crucial, since from it flows the series of transformations that develop its implications in a network of contrasts. Once this foundational invariant relationship has been established—the relationships of identity and difference between interiority and physicality—it must be able to account for all possible combinations in the relationships between humans and non-humans; indeed, this is its only a posteriori justification. The arbitrariness here lies not in the itinerary and modalities the analyst adopts in order to leap from mytheme to mytheme after starting with an initial myth, selected in a contingent manner—as is the case in the Thompsonian transformation employed by Lévi-Strauss in the analysis of myths. It lies instead in the choice of the initial formula, since this conditions the unfolding of the deduction.

How can we reduce this arbitrariness, when the mechanical models that anthropology is reduced to cobbling together, in the absence of an appropriate formal language, turn out to be distressingly crude? Firstly, by ensuring that the logical and experimental foundations on which the original hypothesis is based are solidly established. As far as the matrix of modes of identification in Beyond Nature and Culture is concerned, the starting point was the combination of a philosophical thought experiment with the observation by developmental psychologists that the detection of physical qualities and mental states in objects perceived by humans is probably innate. Admittedly, this initial hypothesis is based on the aptitudes of an abstract human subject, not on a social rule as the Durkheimian tradition would have it—which exposes it to the accusation of incoherence, even duplicity, due to what could be seen as an attempt to bring the wolf of methodological individualism into the structural fold. But it is nothing of the sort. The reason is that social relations do not float in the Empyrean sky of ideas or in the crevices of institutions, but rather they are seen in the interactions between human beings, who do not come into this world like virgin wax waiting to have their properties impressed upon them by the contingent events they pass through. It therefore seems reasonable to base an anthropological invariant on presumed universal characteristics of Homo sapiens, which are necessarily rooted in the consciousness, body, brain, and biophysical capacities of an ordinary human. And that’s why I judged it necessary to root the initial invariant in the dispositions of a transcendental subject.

But perhaps this is an after-the-fact depiction of an act of knowledge which was actually much less systematic than I have painted it here. In addition to the fact that the hypothesis proposed at its source is the product of an act of imagination consisting in placing oneself in a situation which, while not false, is nonetheless fictitious—an experience of imaginative thought based on real life experiences—the deductions that this hypothesis makes possible are not untouched by the prior procedures that made it appear evident, and which the constraints of communicating it, in both written and oral forms, have hidden. For it is clearly through a series of inductive generalizations about ethnography—my own and that of others—that the transformational model of Beyond Nature and Culture gradually took shape. This is nothing out of the ordinary. Indeed, it’s from the ever-expanding repository of ethnological publications that anthropologists extract the elements that enable them to choose this or that fork in the road in relation to the empirical data on which they rely—data that they appropriate in some sense from within, having experienced elsewhere and in other circumstances situations comparable to those recounted. This bifurcation is the ability at the heart of  comparison: to detach oneself from the phenomenal object to which one is paying attention, whether in the field, or as a reader of ethnography, in order to make it exist in the language of relations internal to our conceptual system. And this bifurcation is made possible by a know-how that is all the more difficult to formalize because it relies on the shared mastery of another know-how: that of fieldwork, which immediately makes familiar to the ethnologist the rarely-explained procedures of the movement of objectification by means of which other ethnologists have collected, filtered, and presented data.

This constant to-and-fro between abstraction and description, induction and deduction, direct knowledge and mediated knowledge, local concept and concept with a general claim, explicit comparison and implicit comparison, is probably what makes anthropology a special science—an art of discovery whose flavor of adventure is not only due to the populations among whom we choose to reside. Perhaps this is the philosophical lesson to be drawn from Lévi-Strauss’s remark that we must “let ourselves be carried along by the terrain.”

Decentered and Neo-Universalist

Response by Bruno Karsenti

I’d like to thank Philippe Descola for offering us this lecture, an exposition of what it means to be an anthropologist—or rather of the bifurcations and forks in the road it involves, up to and including the one he himself has embarked upon and continues. Having had the chance to preview this lecture, I was able to gauge both the rigor and the scope of what he offers here, never deviating from his professional orientation: that of an anthropologist speaking to philosophers, and one who doesn’t dress up as a philosopher to do so. Of course, given his philosophical training, which was common among his generation of French anthropologists—the generation that preceded the appearance of an undergraduate anthropology curriculum in French universities— one might think that Philippe Descola might have been tempted to do just that. Paradoxically, however, this was not the case, precisely because this generation was also one of emancipation, of crossing lines in every way possible. In this sense, a background of philosophical training matters only to the extent that it knows what it owes to its break from philosophy rather than to philosophy itself, and takes advantage of it—such that returning to philosophy, doing philosophy on the basis of an anthropological practice that is sent backstage, even for a moment, could only be a step backward.

This is how I understand Descola’s incidental remark about the regular invocation of philosophical concepts and the authority of philosophers, so present in non-French-speaking anthropology today, and the contrast with French anthropology, where there is now, and I quote him, “an inversely proportional relationship between the degree of philosophical training and the quantity philosophical references.” Indeed, today’s meeting is representative of what anthropology, in a certain intellectual tradition, manages to be today—which also gives it, it must be said, a worldwide authority in its disciplinary field. This authority has been particularly evident since the 1960s, through the work of Lévi-Strauss. Today, it continues with the work of Philippe Descola, a leading figure in a genre—which we shall endeavour to describe below—around which a significant part of contemporary anthropology is organized.

But we must also emphasize that this distance in no way detracts from the philosophical value and range of anthropology. Philosophers are interested in what is being done in anthropology, and particularly in what is being done there today, in a shift from structuralism, in the direction of a mapping of what I will call here modes of symmetrization. This interest is due to the range of what anthropology has to offer philosophy, in its capacity to displace and revive traditional philosophical questioning while at the same time altering it—in both senses of the verb alter, i.e., by making it other than it is or thinks it is when it is sustained only by intra-philosophical questioning, and by making it deviate from its exclusively speculative character (thus altering it, in the sense of shedding its philosophical exclusivity).

The statement that sets things in motion comes from a thesis director, Claude Lévi-Strauss, to doctoral student who is not exactly an anthropologist, but an ethnographer-ethnologist, nourished by his readings and ready to set off on his investigation far from his society and culture of origin. To the apprentice ethnographer-ethnologist Philippe Descola, the anthropologist Claude Lévi-Strauss says: “Let yourself be carried along by the field”—a piece of advice tactfully cutting short what must have been perceived as naiveté, in imagining that theoretical questions could pre-order the experience. Perhaps, at the time, Descola was still too much of a philosopher. Perhaps the idea that an empirical science grants to the domain of the sensible and empirical (empirie) the capacity to determine theory—perhaps a certain way of dedicating oneself to the empirical, which is already to configure it theoretically—was still assumed, if only superficially. But it soon becomes clear that at base there’s more to it than that.

Everything turns on the meaning of these verbs: “to carry,” “to let yourself be carried along.” In place of an initial interpretation of simply “letting go”—abandoning oneself to the field—by the end of the argument we come to understand something quite different, almost the opposite: it comes to mean the search for points of anchorage by choosing between possibilities and bifurcations which only the field can offer, and which, to be clearly discerned, demand a continuous and radical reflection at the theoretical level. To put it simply, it is in the practice of anthropology that its theory is found and its philosophical effects unfold. And this is truly a theory through the epistemic choices it makes, and in relation to which the scholar determines himself in the very exercise of his practice.

The history of anthropology, considered as an established science and not simply as a current of thought that took root in the eighteenth century, is a short one. For this science, the ethnographic moment remains foundational. It is the moment of the first correspondences, that is, the working out of the initial generalizations. Of the three modes of generalization that Descola distinguishes at the outset—exegesis through elaborating or condensing indigenous themes or practices, conceptualizing the alternative philosophical order to which these themes and practices are credited, and integrating them as variables within a group of transformations—for him the third is clearly the most consequential. But Descola also insists that, preferable as this form of generalization may be, it suffers from the same deficit as the other two: that of having to deal with an incompleteness, an unfinishedness that is in a sense constitutive of the discipline’s epistemology. At its heart, it could be said that the distinctive style of this epistemology is to develop strategies for reducing incompleteness, in the knowledge that, in the ethnographic moment first of all, there is a gap which reverberates at all levels and which is known to be irreducible. The first error concerning this gap, the first wrong turn, would be precisely to believe that it can be closed. This is undoubtedly the greatest danger, often unconscious, which is faced those who pursue ethnophilosophy, in seeking to transform statements heard in the field into a closed and stable doctrine.

Descola captures the entire set of strategies for reducing this gap under the generic category of “symmetrization”—which, I must point out right away, marks a certain shift in problematic, in relation to Levi-Strauss’s anthropology.

It could be said that the concept of symmetrization plays out on two levels in this lecture: first as a generic concept, and second as a specific, even singularizing concept—in the sense that by gradually refining its meaning and implications it allows Descola to characterize the turn he himself has taken within structuralism. The concept or idea of symmetrization, is, I believe, the best way for us to grasp the philosophical effects of this lecture’s reflection on anthropology. Therefore I will frame my questions by following this thread.

We must begin by making clear that symmetrization is the hidden spring, more or less conscious, of the correspondences that the ethnographer-ethnologist seeks to establish in the field, and that this spring is the one that truly carries him through to the anthropological stage. This is the first “carrying along” effected by fieldwork, and it is necessarily a comparative one. This is the inductive starting point which no anthropologist can escape. It is impossible not to make comparisons—between his own ways of thinking and those of his interlocutors, between what he knows of ethnological works centered on the same cultural area, or on the same type of reality as the one he attends to but in other cultural areas, or again in the phenomenal sorting he undertakes, starting from experiential data, and establishing controlled correspondences. This presupposes that we reflect on one hand about how “comparables” are determined, and the other hand about the relationship under which a given comparison can be considered valid. 

To speak of symmetrization draws out the consequences of this reflection, and, in a way, takes it to its logical conclusion. This assumes that the relationship under which we make comparisons tends towards a situation of equilibrium and reversibility—so that it could be taken up again in the direction opposite to the one the anthropologist followed in the first approach, starting from the site to which the field site is compared. This is not a comprehensive approach, proceeding by imaginative variations—as if the aim were to project oneself into the other’s point of view in order to occupy, phenomenologically, his perceptive position. Instead it’s a matter of intellectually constructing a model that can be applied in several directions, where the relationship between the interpreter’s interpretation and the interpretation of the interpreted tends toward equivalence. All this is done in the knowledge that such adjustment cannot be complete, that symmetrization is necessarily unfinished—due to the very fact that the interpreter is not the interpreted, and the site of fieldwork is for him a field, not a social and natural milieu experienced through familiarity, indigeneity, and belonging.

Can symmetrization, understood this way, be undertaken while remaining at the level of only inductive logic? Descola’s answer is negative. Anthropology moved forward in the 1960s in what has been broadly defined as the structural turn (which even in France has several versions, although the aura of Lévi-Strauss somewhat blinds us to the varieties of structural reasoning) by defining itself as a social science that is partially but decisively deductive. In this it was inspired by linguistics, while separating itself from sociology, and more generally from social sciences whose epistemology remains essentially historical (or Weberian, as Jean-Claude Passeron would say) and which are never, even in their most elaborate attempts at modeling, understood as hypothetico-deductive sciences.

So we have a discipline—the only one, along with linguistics (though we might also ask about economics)—that can decisively call itself hypothetico-deductive, precisely by virtue of its properly empirical character. This is what, epistemologically, invites a first philosophical interrogation. But the interrogation becomes even more precise: What Descola shows is that deductivism is not a unitary path, and that it too has its bifurcations. And it is by taking one of these paths, by once again making a choice within the bifurcations through which one may be “carried along,” that anthropology determines itself more specifically as the symmetrization of ontologies, as it is defined here.

To grasp the scope of this latest gesture, which began in the 1990s—and which draws on several currents of thought, notably the results of adopting the principle of symmetry within the anthropology of science, especially under the impetus of Bruno Latour—we must first emphasize the status it grants anthropology, which makes it stand out among all the social sciences. Descola notes this discreetly in his lecture, but it seems to me that two points need to be underlined more strongly if the discussion is to engage philosophers. On the one hand, the option taken by structuralism amounts to reinvesting the universality of human phenomena in an unprecedented way (at the greatest distance both from the old philosophical anthropology, and from physical anthropology, which is being renewed today by being bundled up with biology and neuroscience). On the other hand, it initiates a gesture which has not been sufficiently noticed—and which only becomes fully apparent in its latest developments, precisely through accentuating the concept of symmetrization—which is to make the interpreter, or the one who compares, an integral part of the comparison itself: a relative figure, inscribed within the model he himself has constructed.

The tension between these two points must not be neglected. From the first, it follows that the incomplete anthropological symmetrization must be read as a constantly relaunched drive towards an exhaustive knowledge of human facts in their entirety, a gesture that is unambiguously, and even proudly, acknowledged as that of Western reason or rationalism. But at the same time, by inscribing itself within the system of variations that transformational logic aims to construct, it manages to break with the ethnocentrism still borne by earlier epistemologies—above all, that of functionalism. This double feature combines rationalism with a de-centering of knowledge to an unprecedented degree, sketching an epistemic profile that is particularly fertile for tackling the current challenge of the transnational opening of modes of thought and social practices, as well as the critique of reductive forms of universalism that goes along with it.

This is undoubtedly the reason why today, as I suggested at the outset, it is anthropology out of all the social sciences that has the privilege of most strongly calling out for philosophical interrogation. Its philosophical effect could be described as follows: anthropology requires or teaches philosophy to disentangle itself from the idea that universalism can in any way have the value of a postulate—not in order to renounce it, but in order to impose on it the requirement of being tested in action: in other words, to subordinate the solidification of the universal to the ability to trace and expand systems of variation, within which the cultural form that carries it learns to place itself. 

Comparison, or comparative logic, is the royal road of this non-dogmatic neo-universalism, relativized through its own operations of relaunched totalization. This depends on the fact that certain invariants, on which systems of differences are built, have been formally identified. In this sense, comparison is founded on generalization, but of a kind that functionalism could never have imagined. A deductive generalization, born not of translation or abstraction, but of a synthesis in which the data are defined as variables in advance. That is why this generalization is centered not on universals, but on invariants—specific to the group of transformations we extend to its logical limits. And in one of these groups, the person making the comparison is also included, as one of the figures determined to make universalism itself their specific horizon of thought.

It is undoubtedly on this point that philosophy is invited to measure the scope of the anthropological gesture and to translate it back into its own system of questioning. In the period that began in the ’60s and ’70s—I’m thinking in particular of political philosophy, which in many respects in France can be said to have been the organizing axis of philosophy as a whole, a situation whose effects are still being felt—philosophy also sought a critical rethinking of universalism. It often formulated this as an intensive universalism, ever more capable of integrating differences and looking back on itself as a situated point of view, capable of de-centering and self-criticism. What anthropology’s epistemological reflection makes possible, having reached a certain stage in the history of its discipline, is to trace an operative figure for recasting universalism which philosophy could not access on its own, and which philosophy must now interpret—that is, from which it sees that it must learn something essential for philosophizing in the present.

And what is this operation? First, philosophy must learn the self-confidence of rationalism, in one of its realizations—in which it reaches a stage of formal thought in exactly the same movement as it becomes empirically anchored, and reveals itself to be better able to extricate itself from the illusions of the universalist postulate. This confidence arises from a resolute investment in what the social sciences have progressively bequeathed to philosophical reflection: namely a radicalization of investigations of formal and material causalities in the explication of human phenomena, rather than those of efficient and final causalities, which were prevalent in an earlier era (and to which naturalist and physicalist anthropology remains riveted, as does the naive culturalism found in philosophical anthropologies). This confidence increasingly acknowledges the constitutive incompleteness of the act of research, and is dedicated to reconstructing, relaunching, and reopening its totalizing operations—organized around the construction of transformation groups of variable amplitude. What remains of universalism is only the direction, not the target—an intention of universalism, much more than an intensive universalism, therefore. And a universalism in intention which is paradoxically founded on its own relativization.

But to reach this point, structuralism itself had to undergo a shift. The emphasis here is on an alternative present in Lévi-Strauss, between The Elementary Structures of Kinship of 1949 and the approach to myths achieved in Mythologies. This is an alternative embedded in hypothetico-deductive generalization. It doesn’t give up on the leap of hypothesis, but instead recognizes that this leap will be understood differently depending on whether, in one case, structure is the originary form from which the maximum number of possible variations can be derived (as with Goethe’s Urform); or whether, in the second case, structure belongs to the transformation grid itself, and can only be read in the regular deformation of one figure into another (on the model of D’Arcy Thompson’s geometric transformations). Why is this alternative ultimately decisive for pursuing the structuralist project, and for tilting it more clearly towards an anthropology of modes of symmetrization?

Here again, it is only by analyzing the way in which the anthropologist forges or chooses conceptual tools that an answer can be given, and the philosophical effects of his gesture can be measured.

It seems to me that the heuristic advantage of Thompson’s form, or structure, over Goethe’s form or structure, is that it responds more consistently to anthropology’s immersion in the field, without giving up ground and even pushing it further, by expanding its formal operations—and thus deepening its hypothetico-deductive vocation. Why? Because transformation is not, in this case, the transformation of an abstractly isolable original form—the principle of reciprocity, to take the “Goethean” example—but the passage from one form to another in a space of determined coordinates, inseparable from the set of forms already given. To stay with Lévi-Strauss’s “Thompsonian” example, in this case it would be myths actually recounted, which turn out to be transformable into one another, and which therefore belong to the same group.

In this way it is only in the Thompsonian case that we can truly speak of symmetrization—and therefore it is only there that neo-universalism, in intention and not in intensification, can obtain. Because the invariant, in this case, is not a universal of a higher degree—such as the prohibition of incest combined with the principle of reciprocity—but an invariant in the strict sense, i.e., the point we hypothetically give ourselves, to trigger deduction within a set of variables that the investigation began by constituting, and which it only gradually expands, through continuous variations, by adding new data to that which has already been processed. In the same way, including the perspective of the one who makes the comparison within the model they set into motion is also taken a step further, as this is deduced in turn as a specific deformation within a given transformation grid. (Within Descola’s framework, one consequence is that naturalism is one of the ontologies under consideration.)

Finally, I’d like to emphasize two points, which are important from a philosophical point of view, and which, by the meaning they take on there, may well have a knock-on effect on anthropology itself as it is practiced.

The first, as I have just noted, is the status conferred on the “operator” that the anthropologist of modes of symmetrization must give himself in order to conduct both his investigation and his interpretation. The relationship between interiority and physicality that Descola has established, beyond the nature-culture divide, in order to identify the modes of symmetrization at work between social forms of existence and the living environment, is not the principle of reciprocity linked to the prohibition of incest (which, for its part, stages this divide, by presenting itself as the decisive step away from nature and into culture). But, paradoxically, this version of structuralism reintroduces a minimal phenomenological base. Admittedly, this base is not a metaphysical foundation. But if it is not, and if it has only a strictly epistemological—or methodological—meaning, the question arises as to whether it is susceptible to modification, due to subsequent discoveries which may arise within the picture it will have made possible. Or, to put it another way: if the hypothetico-deductive becomes stricter, it’s on condition that another hypothesis may come along to replace the one that has been applied to the current terrain it enables us to treat. This is undoubtedly the result of our making an even more radical distinction between invariant and universal, even more strongly than in the Goethean scheme. Therefore the “meta-invariant” (for Descola, the relation between interiority and physicality, taken as a starting point, within a naturalist perspective) cannot avoid being subjected to the test of the variations it has itself made legible.

In this sense, I wonder whether Descola’s decision to refer the starting invariant (or “meta-invariant”) to the hypothesis of a transcendental subject equipped with basic dispositions—the detection of physical qualities on the outside and mental states on the inside—is not in some way a step backward from this procedure’s full implications. In philosophical terms, it seems that this invariant can be described as a principle not of reciprocity, but of identification and counter-identification. This is why, although based on a human disposition, it does not a priori separate humans and non-humans, or so-called cultural forms from so-called natural forms, which can be distinguished only in one very particular mode of identification and counter-identification, that of naturalism, which is never more than one way (though it is dominant for us) of symmetrizing. In fact, identification has two faces: it identifies and it identifies itself. It is identifying and reflexive. What the anthropology of modes of symmetrization makes visible are social or socio-historical variations, which are so many variations in the equilibrium to which the double gesture involved in identification is susceptible. (It also allows us to escape from a confusion between society and its environment, which takes place against a backdrop of entanglement and variable exchanges between society and the living milieu, or between relations of production and productive forces, whose variations cannot be predetermined, but the detection of which must always be suspended during fieldwork, and perhaps most visibly in “the anthropology of nature.”)

This way of extending the anthropological operation, in my mind, would restore a certain value to the Durkheimian inspiration that runs through the history of anthropology, for which Lévi-Strauss was a relay, and whose significance I know Philippe Descola appreciates. However, he dismisses it in passing in his lecture, for reasons I understand: the insistence on social rules—and hence the hypostasis of “society”—which would exist in the Empyrean, detached from practices and interactions. I won’t dwell on the fact that Durkheim and the Durkheimians can be read in a very different way. Instead, what I want to emphasize in conclusion is simply that the anthropology of modes of symmetrization, if it makes room for the principle of identification and counter-identification, restates anew the way in which society and world interweave and reflect each other in various ways in order to constitute themselves, allowing for regular identifications and counter-identifications among the individuals and sub-groups that take shape within them.

At a time when this inter-implication and internal reflection is emerging as the major political issue of the present on a global scale, anthropology is certainly at the forefront of what we need to think about in order to act more effectively. In short, it has a singularly stimulating effect on our political philosophy, calling on it to renew itself intensely. From my point of view, this leads us to a better understanding of what we mean today by society, and toward a level of integration and interdependence between society and environment that extends in both directions, and thus requires us to find a new language. In Philippe Descola’s enquiries, perhaps a new container , neither analogist nor naturalist, neither animist nor totemist, can be found.

The flavor of the ordinary

Philippe Descola responding to Bruno Karsenti

First of all, I’d like to thank Bruno Karsenti for agreeing to comment on my lecture. I have just experienced, as I imagine you all have, the appropriateness of the choice I made in asking him to comment. He has both deepened certain elements I proposed in the lecture and raised some extremely important questions which I won’t be able to answer exhaustively, though I will try to shed some light on them.

The first question you asked basically concerns the epistemological status of the first formation of hypothetico-deductive hypotheses, based on what I called at one point “dormant knowledge.” This lets us come back to the status of Thompsonian transformation, particularly in Lévi-Strauss’s analysis of myths, which is founded on prior knowledge and on uncommon intellectual qualities. Lévi-Strauss used to say that he had put thousands of myths on filing cards. But even thousands of filing cards don’t enable someone to move to this extraordinary enterprise of jumping from one mytheme to another all around the planet, and particularly in the Americas, unless they have thousands of myths in memory, independent of filing cards. Somewhere he says that he lived immersed in myths for many years, in daily contact with them—and this implies a very special kind of exercise. An undertaking like Mythologiques is unrepeatable not only because the analyst’s choices were in fact in line with reality through these successive leaps from mytheme to mytheme and were based on highly personal and completely contingent intuitions but also because it requires mastery of this considerable mass of information. The hypotheses we use to build models of invariants must be based on prior knowledge. For Lévi-Strauss, it was a very great, very solid ethnographic knowledge and familiarity of the Americas in particular, which is all the more remarkable given that in his time the South American ethnographic materials on which he worked were often of pathetic quality and very incomplete. The situation was quite different for North America, thanks to the remarkable work carried out by the Bureau of American Ethnology since the late nineteenth century. To this fragmentary information, which he drew from the most diverse sources, Lévi-Strauss added an encyclopedic knowledge of the peculiarities of American flora and fauna, particularly in the tropics, astronomical and climatic cycles, and the characteristics of ecosystems, which enabled him to make in a flash the connection between the color of a bird, a constellation, and the qualities or uses of the wood of a certain tree, which itself would be the basis for an inference leading him far away by ricochet to another myth, with other properties, potentially analyzable as a formal transformation of the previous one, and so on infinitely, since “the world of myths is round.” Armed with this double knowledge—of thousands of pages of strange tales and a thorough knowledge of the physical environment in which these played out—Lévi-Strauss was thus in a position to bring together episodes from stories that seemed to have nothing in common and actual and proven aspects of the world. In addition to requiring a phenomenal memory (despite the famous index cards), these resonances, which are the elementary brushstrokes of transformation in the analysis of myths, can also be seen as the hypothetico-deductive operation in its primary form. But this is not reproducible, since any interpretative path from mytheme to mytheme, from sense property to sense property, even if the analyst should have knowledge and intellectual qualities equivalent to those of Lévi-Strauss, will chart its own sui generis course. This is the limit of Thompsonian transformation applied to myths: it is not an ordinary scientific method, for it opens up as many possible alternatives as the imagination allows, none of which can be said to be superior to the others.

Let us add that the structural analysis of myths differs from the Thompsonian method on one crucial point that Lévi-Strauss himself underlined. As D’Arcy Thompson conceded, there are specific and generic differences among the organisms whose transformations we analyze by continuous variation, and these are linked to discontinuities in the genetic code. But the transformation group that serves as a framework for an arrangement of mythemes makes a virtual environment within which the analyst makes ideal cuts. In the first case, the principle of discontinuity is inherent to reality, while in the second it is imposed on it by the mind of the analyst, for the variants of a myth do not in themselves carry a principle of discontinuity—least if we take them in their paradigmatic aspect, which is the only one that counts here. According to Lévi-Strauss, the actualization of certain possible forms of the myth is the result of an external principle, as the mind introduces discrete changes in the form of contrastive intellectual relationships—a kind of transcendental subject which the analyst, as he rediscovers the formal intuitions on which this subject once played, strives to incarnate in his own person by playing with contradiction, inversion or symmetry. (Incidentally, this way of conceiving mythic inventiveness contradicts Paul Ricoeur’s assertion that the structural method is a “Kantianism without a transcendental subject.”) The transformation that Lévi-Strauss activates in the analysis of myths is therefore not the same as that used by Thompson to move from the form of one organism to another. Lévi-Strauss’s version of transformation introduces differences by means of contrastive operations upon a set of phenomenal properties relating to the same aspects of the world; while the Thompson’s version of transformation introduces continuity by means of deformations performed upon a set of phenomenologically dissimilar forms, between which affinities are postulated.

Moreover, if it is impossible to reproduce experimentally, the very particular form of hypothetico-deductive operation that the analysis of myths puts into motion is also empirical in another sense. Anthropology is a deductive discipline based on the empirical, not necessarily that of the field, but of the experience and the taste we acquire in the field for the flavor of the ordinary. And you have to be able to nurture this taste for the ordinary by reading ethnography. You can’t do comparative anthropology as Lévi-Strauss did or as I have done without a deep-rooted appetite for ethnography, for the pleasure of the singular, which is a bit of a paradox. It’s the capacity to project yourself into very different ways of world-making, and to experience a pleasure that is both aesthetic and almost sensual, from which precisely to extract data. In France at the Laboratoire d’Anthropologie Sociale, Lévi-Strauss was the promoter of an enormous file produced by Yale University, the only one of its kind in Europe, called the Human Relations Area Files. It enables ethnographic data to be cross-referenced, made discrete, always with the problem of knowing what are the comparables. But Lévi-Strauss made relatively little use of it, because he already had inside him this dormant knowledge of diversity which is the driving force behind the creation of hypotheses. I find myself in the same situation, not out of imitation but because I also love more than anything the flavor of reality that comes from reading ethnography. This question of the empirical and theoretical components of transformation is a very important point, I thank you for emphasizing it.

There are still many things you’ve said that I’d like to pick up on, and in particular this point somewhat related to what I just mentioned, which is the self-confidence of rationalism, grounded in empiricism. I think this is very true. That’s why I distance myself from ethno-philosophical endeavors, because it seems to me that there is a tradition of European rationalism that is anything but exhausted. This is one of the things of modernity to be saved, which nonetheless makes it possible to imagine less Eurocentric and less anthropocentric ways to account for the world’s diversity of lifeways. In a way, anthropology could only be an invention of naturalism, not simply because of colonialism, but also because of the injunction (which Todorov for example brought to light in The Conquest of America), which was to constantly link knowledge, the submission of objects by thought, to the objectification of others through colonial domination. Anthropology is an heir to this, and so either it disappears, or it must confront this heritage. It must confront the fact that it is a product of Europe’s colonial enterprise, and admit that part of its fecundity comes precisely from a kind of drive for knowledge which is characteristic of the relationship with others that developed in Europe in the Renaissance.

Another important question you’ve raised is that of the closure of my portrayal of the four modes of identification, and the possibility or even the necessity that naturalism be transformed, for political reasons, to escape the catastrophe we’re currently witnessing, which is not simply ecological. It’s a very important question for which I don’t really have an answer. Basically, what I call modes of identification are prerequisites for the constitution of ontologies. And among these plural ontologies, there are very few that are pure, many of them are combinations. This is what I set out to explore in my recent book, Les Formes du visible, when I sought to show, for example, that the Tsimshian people of Canada’s Pacific Coast made images, some of which were animist, while others were totemist. Some of these combinations may be structural, while others are the effect of historical circumstances. You quoted our friend Bruno Latour about symmetrization, and this is a discussion he and I had for years: Have we ever been modern? If we have never been modern, what are we? Are we analogists who have somewhat transformed themselves? And if so, is the form of analogism which belongs to what I call naturalism, itself susceptible to transformation into something else?

This is a question for which I don’t have an answer, except that—and this is why I’ve taken such an interest in images— it seems to me that images in some way herald, or are explicit prefigurations of new forms of “worlding” (mondiation), that is, ways of detecting and highlighting certain elements of the world, and of organizing it differently. Such is the case with images in fifteenth-century Europe, which heralded naturalism long before naturalism became explicit in the philosophical texts of the seventeenth century. I believe that our current situation has been fairly well described by certain contemporary artists—while we intellectuals, philosophers and anthropologists, struggle to grasp this situation—who are already in configurations that bear no relation to those familiar to us. I’m thinking in particular of an artist who really struck me when I saw his installation in Arles, at the LUMA Fondation: Pierre Huyghe. Some of you may have seen this installation called UUmwelt, which is basically the creation, on a reduced scale of course, of a world in which the ontological distinctions with which we are familiar have been completely exploded. There is no longer any nature, no longer any culture, no longer any representation of the former by the latter. There are tools, biological objects. There are machines, there are swarms of bees. He has created a completely original Umwelt or environment. He is not the only one to do this: think of Ólafur Elíasson, Hicham Berrada, or the films of Lucien Castaing-Taylor and Verena Paravel. I think we have to be humble in this area and look to the artists and try to imagine where we’re heading. You mentioned political philosophy, which is one of the reasons why I wrote my latest book with Alessandro Pignocchi—to propose cosmopolitical perspectives that are a little different from those that are classically ours due to the duality we’ve inherited: liberalism on the one hand, and Marxism on the other (Descola and Pignocchi 2022).

A final word on Durkheim, which is a debate we’ve been having for some time now. To come back to Bruno Latour, I don’t have the loathing for Durkheim that he had—I think the three of us have even discussed it. Fortunately, it wasn’t Gabriel Tarde who won the debate with Durkheim at the start of sociology, otherwise we’d be in a very bad position…. In fact, this allowed us to come back to Durkheim, possibly to envisage him in the light of Tarde—whereas with Tarde alone, I don’t know where we would have ended up.

For me, the problem with Durkheim is what I’d call a sociocentric impasse: that basically, by definition, the unit of analysis is social institutions. Now, what I’m interested in, and this is why I’ve spoken of an antecedent situation (or “antipredicative” in Husserl’s sense), are those moments of emergence when people select relevant elements from their environment to compose original worlds with them, out of which, in my view, social institutions flow. Social institutions are products of these arrangements or dispositifs, very much upstream of the collective life which then solidifies. This is why I believe, even so, that we have to turn our backs a bit on Durkheim—to overcome the sociocentrism of the social sciences, and of anthropology in particular.

Works Cited

Bilolo, Mubabinge. 1987. “De l’ambigüité de philosophies africaines de la personnalité, de l’identité et de l’authenticité. Une relecture critique de la Négritude et de l’Authenticité.” In Das Andere und das Denken der Verschiedenheit, edited by Heinz Kimmerle, 315-335. Amsterdam: B. R. Grüner.

Descola, Philippe. 2005a. “On anthropological knowledge.” Social Anthropology 13, no. 1: 65-73.

Descola, Philippe, 2005b. Par-delà nature et culture. Paris: Gallimard.

Descola, Philippe and Alessandro Pignocchi. 2022. Ethnographies des mondes à venir. Paris: Seuil.

Evans-Pritchard, Edward E. 1950. “Social anthropology: past and present; the Marett Lecture.” Man 50: 118-124.

Gell, Alfred. 1999. The Art of Anthropology: Essays and diagrams. London: The Athlone Press.

Goethe, Johann Wolfgang von. 1790. Versuch die Metamorphose der Pflanzen zu erklären. Gotha: C. W. Ettinger.

Hountondji, Paulin J. 1970. “Comments on Contemporary African Philosophy.” Diogenes 71: 109-130.

Kagame, Alexis. 1976. La philosophie bantu comparée. Paris: Présence Africaine.

Leach, Edmund R. 1961. Rethinking Anthropology. London: The Athlone Press.

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

Lévi-Strauss, Claude. 1971. Mythologiques IV: L’homme nu. Paris: Plon.^[2]Lévi-Strauss’s four-volume series Mythologiques begins with The Raw and the Cooked translated by John and Doreen Weightman (New York: Harper & Row, 1969) and continued until The Naked Man translated by John and Doreen Weightman (New York: Harper & Row, 1981).

Lévi-Strauss, Claude and Didier Eribon. 1988. De près et de loin. Paris: Editions Odile Jacob.

Mill, John Stuart. 1843. A System of Logic, Ratiocinative and Inductive: Being a Connected View of the Principles of Evidence and the Methods of Scientific Investigation. London: John W. Parker.

Pouillon, Jean. 1975. Fétiches sans fétichisme. Paris: François Maspero.

Taylor, Anne-Christine. 2022. “Anthropology Comes In When Translation Fails.” Social Anthropology 30, no. 1: 91-103.

Tempels, Placide. 1959 [1945]. Bantu philosophy. Paris: Presence Africaine.

Thompson, D’Arcy Wentworth. 1961 [1917]. On Growth and Form. Cambridge: Cambridge University Press.

Viveiros de Castro, Eduardo. 2009. Métaphysiques cannibales. Translated by Oiara Bonilla. Paris: PUF.^[3]Appearing in English in 2014 as Cannibal Metaphysics. Translated by Peter Skafish. Minneapolis: University of Minnesota Press.