Anthropology without Belief: An Anti-representationalist Ontological Turn

1. Introduction: Belief and the Problem of Alterity

While the ontological turn in anthropology has many interconnected themes, its rejection of the role of “belief” in ethnography is one of the most philosophically striking, and perhaps the most profound. “Belief” has been the central concept in the anthropological conceptualization of human difference since E. B. Tylor’s ([1871] 1889) Primitive Culture inaugurated that project in its modern form. In their recent book, The Ontological Turn: An Anthropological Exposition (Holbraad and Pedersen 2017), Martin Holbraad and Morten Pedersen contend that eschewing belief “provides perhaps the most incisive way of expressing the essential move of the ontological turn—its bottom line, so to speak” (2017,188). Because it rejects the category of belief, the ontological turn has the potential to transform the anthropological project of understanding the cognitive differences among humans.

While the claim is provocative, contributors to the ontological turn have been unclear about why “belief” should be eliminated and what it means to do anthropology without “belief.” This essay will argue that the rejection of “belief” is best understood as part of the ontological turn’s broader rejection of representationalism. Representationalism is a philosophical picture of mind and language that has dominated European thought since the 17th century. The signature contribution of some of the 20th century’s major figures, including Heidegger, Dewey, Wittgenstein, and Davidson, has been to try to think about mind and language without the representational categories of “ideas” or “beliefs.” As the eclectic list of canonical figures suggests, anti-representationalism is no more univocal than the ontological turn itself. The first task, then, is to get a handle on what “anti-representationalism” might mean in anthropology.

2. Belief and Representation in Anthropology

For the first 100 years of its existence, the culture concept was anthropology’s primary tool for understanding human difference. Each term of E. B. Tylor’s famous definition established a dimension of human alterity: “Culture is that complex whole which includes knowledge, belief, art, morals, law, custom, and any other capabilities and habits acquired by man as a member of society” ([1871] 1889, 1). Knowledge and belief characterize cognitive differences, while art, morals, law, and custom capture valuational differences. Tylor, of course, was a progressivist, so cognitive difference included the rational march of science. Europeans knew that the earth orbits the sun, others did not yet know this. Yet, even for Tylor, knowledge is too factive to account for human difference on its own. It entails that difference is mere ignorance, and it thereby elides many of the differences that exist. Fully capturing human cognitive difference requires characterizing content without appeal to truth or fact. Within the European experience, Catholics and Protestants differed in a cognitively interesting way, and calling the schism a difference in “belief” is an ecumenical characterization. Turning from knowledge to belief focuses the difference on the content of what is believed and away from the questions of truth or justification. Tylor could consistently say that while some Europeans have believed that witches are human women with special powers, the Azande believed that witches are spiritual emanations arising from malevolent thoughts, and he did not believe that witches existed at all. A difference in belief, conceived as part of cultural difference, was the primary way in which anthropologists of the late 19th and 20th centuries analyzed cognitive alterity.

Representationalism constituted the philosophical background to Tylor’s conception of belief, knowledge, and cognition more generally. The primary philosophical movements of the modern period, including empiricism, rationalism, and idealism, shared a common picture of mind and language. The mind is a repository for contentful states—ideas—that are the direct objects of conscious reflection. The semantic structure of propositions reflected the structure of ideas, and ideas purported to mirror the world. Representationalism gives modern philosophy its distinctive epistemological and metaphysical questions. I know (am directly aware of) the representations in my mind, but how do I know that my ideas accurately represent non-mental reality, or the minds of others? Is it the case that the structure of reality matches the structure of our ideas? Idealism is the limit case of these answers: we know our representations, but as Berkeley said, an idea can only represent another idea. It is representations all the way down. On the other side of the debate, realists take the representational purport of thought and belief seriously. When a representation is successful (true), according to the realist, mental states correspond to features of a non-mental, external world.

The common conception of a belief among western intellectuals has two key features that tie it inextricably to the representationalist project. A belief is a combination of (a) a determinate, introspectable content and (b) a distinctive kind of attitude toward that content. Beliefs are introspectable in the sense that the believer knows in a direct way what she or he believes and has a kind of first-person authority about what she or he believes. “Content” is what distinguishes one belief from another; content is the difference between believing that it is raining and believing that it is snowing. The attitude of belief distinguishes believing that it is raining from wanting it to rain. Since one knows whether one wants it to rain or believes that it is raining, a believer must be aware of this attitude along with the content. The common conception of beliefs is a characteristic expression of representationalism. The mind, on the representationalist picture, is fundamentally a producer and consumer of representations. Belief contents are paradigmatic representations, and they are distinct from and independent of what they are about. Beliefs, including perceptual beliefs, are the primary mode through which we can cognize the “external” world.

In the 19th century, representationalism figured decisively in the emergence of anthropology and the other social sciences. The scientific revolution raised deep questions about how we are to understand ourselves. Successful explanations of mechanical, electrical, luminary, astronomical, and chemical phenomena raised the question of whether human life would succumb to scientific treatment. Kant argued that it would not. The biological and human sciences had a non-mechanical, teleological character requiring a form of judgment distinct from that used to understand mechanical systems. In the 19th century, while Darwin was leading biology away from teleology, scholars like Dilthey were leading historians, sociologists, and anthropologists back to Kantian arguments. The human sciences needed distinct forms of inquiry because the human capacity to imagine what is not (or not yet) the case, to conceptualize what cannot be observed, and to reflect on our own capacities for thought and perception mark us off from other kinds of being. We are the animals for whom our own existence is a problem, Erich Fromm would later say. Our capacity to form and communicate representations entails that the study of humans must be deeply distinct from the study of non-humans. Representationalism thus composed one of the last bulwarks in the defense of human exceptionalism.

Anthropology inherited its commitment to representationalism from the 19th-century debates over the possibility of a human science. I would contend that representationalism was constitutive of early and mid-century conceptualizations of culture. While an exhaustive defense of this claim is impossible here, evidence can be found in the writing of David Bidney. Writing in the 1940s, Bidney was one of the first philosophers with a broad and deep knowledge of anthropology. Considering the different conceptualizations of culture in play at the time, he distinguished between “idealist” and “realist” conceptions of culture. He meant “idealism” in the sense that Berkeley used it: mind is ontologically primary. Culture, on an idealist conception, is a realm of representations, composed of meanings, symbols, and beliefs. The realist alternative held that “culture consists of the body of material artifacts and non-material customs and ideals” (Bidney 1944, 31). From his citations, it is clear that he had the Boasians in mind as paradigmatic realists. For idealists, culture shares the ontology of the mind, whereas for the realists culture is composed of both material and ideational traits. The distinction, then, is fundamentally an ontological one.

Bidney’s distinction between idealist and realist conceptions of culture captures one of the important watersheds in anthropological thinking and shows how it results from shared commitments to representationalism. For the idealists, a representational understanding of individual mental states is pushed up to the social level. Collective representations have all the features of individual thoughts and beliefs: representational purport (including intensions and extensions), semantic compositionality, entailment relationships, and the capacity to be true or false. By calling such views “idealist,” Bidney is drawing attention to the fact that the content of collective representations is internally determined. For the idealists of the modern period in philosophy, like Berkeley, the content of an idea— roughly what we would now call a concept—is determined by other ideas. Complex ideas can be decomposed into simpler ones. Simple ideas are not given content by their capacity to refer to material objects or some other relationship external to ideas. Rather, they have their content immediately, by one’s direct consciousness of the idea. Anthropologists were not committed to the empiricist analysis of simple and complex ideas, but they did adopt the analogous doctrine of semantic holism, wherein the content of any particular cultural representation was given by its relation to others. Cultures could thereby take on the character of total “world views,” and relativism (like idealism before it) could become almost irresistible.

While Bidney’s “realists” were as committed to the representationalist framework as were the idealists, there were two important differences between them. First, the realists treated material objects as parts or elements of culture. Culture was not only ideas, but things. Second, they saw cultures as less cohesive, coherent, and internally consistent than the idealists. Cultures, for the Boasians, were collections of traits. Their questions were less about how the traits hung together (though they were also committed to holism) than about how traits diffused among cultures. Representationalism emerges in trait theories insofar as some of the traits are contentful, and they keep their content as they are exchanged among individuals or groups. Stories, myths, symbols, and other things that might be called “beliefs” were among the traits that defined a culture at a time. To travel among cultures, a story or symbolic complex has to be re-identified in terms of its content. Its meaning might shift as it encounters other traits (including material traits), but content constitutes identity. Where the idealists envisioned a culture as a continent of representation, the realists envisioned it as an archipelago: islands of meaning.

While realist conceptions of culture were still common when Bidney was writing, post-war anthropology on both sides of the Atlantic quickly turned to idealist conceptions of culture. Structuralism provided sophisticated ways of relating the content of representations. While the analysis was different, Geertz’s interpretive treatment of culture as “thick description” was equally idealist. Thickly described cultural events express conceptual relationships embedded in speech and reflected in action. Structuralism and interpretivism presuppose that cultural analysis proceeds without attention to individual psychology. They thus stood together against componential analysis and early forms of cognitive anthropology. Cognitive anthropology, in its 1960s and 1970s renditions, took culture to be what an individual needs to know in order to behave as a competent member of a society. While there were some contrary tendencies—Marvin Harris’ work comes to mind—it is fair to say that until the 1980s, conceptualizations of culture were primarily idealist.

Critiquing representationalism and trying to think our way out of its commitments has been a major thrust of 20th-century scholarship. Literary studies have celebrated the death of the author, science studies have tried to adopt the perspective of the objects, while philosophy has flirted with behaviorism, eliminative materialism, Heidegger’s dasein, and varieties of pragmatism. Within anthropology, one of the primary anti-representational movements has been the dismantling of the culture concept, with Writing Culture (Clifford and Marcus 1986) as its iconic center. The multi-stranded critique focused on the reification of culture. All idealist conceptions treat culture as a system of meaningful units shared by members. The writing culture critique cast doubt on the epistemological, metaphysical, and political credentials of the idealist conception of culture.

The writing culture critique was an important attack on representationalism. In its wake, some approaches to ethnography and cultural anthropology have adopted an anti-representationalist stance. Practice theory, for example, has its roots in Wittgenstein’s anti-representationalist meditations. However, not all post-critique anthropological theories have adopted a full-throated anti-representational stance. Sperber’s “epidemiological” approach (Sperber 1996) rejected the idea that culture is shared, but kept the notion that the mind produces and consumes representations. Where does the ontological turn fit? The answer seems to depend on which authors one chooses as emblematic of the ontological turn. Many authors associated with the ontological turn, such as Edward Viveiros de Castro or Philippe Descola, have strong structuralist affinities (Turner 2009). If these authors are emphasized, the ontological turn seems to remain squarely within the representationalist camp. On the other hand, some important contributions have emphasized the role of things in the analysis of ethnographic phenomena (Henare, Holbraad, and Wastell 2007). These authors seem to be adopting an anti-representational stance.

Because “belief” is such a centrally representational idiom, the rejection of belief by proponents of the ontological turn is diagnostic of its ambitions. Should “belief” be rejected in such a way as to preserve some form of representationalism? Or does the rejection of belief entail a more thoroughgoing repudiation of anthropology’s representationalist origins? And if the latter path is taken, how is human difference to be understood?

3. Against Belief

In the characterization and analysis of human difference, “belief” has two related functions. First, it permits the description of belief contents (ideas, concepts, percepts, propositions) without commitment to the truth or falsity of what is believed. In this way, “believes that . . .” is different from so-called “achievement verbs” like “sees that . . .” or “knows that . . . ” Second, and because it has this first function, belief permits the anthropologist to describe a representational scheme different from her own without inconsistency. It is perfectly consistent to say that Alice believes that shamans undergo metamorphosis and I do not believe that they do.

It is just this apparent usefulness of the belief idiom that Viveiros de Castro calls into question. In his discussion of Amerindian metamorphosis, and the trope of bodies as clothing for the soul, he critiques Irving Hallowell’s account of the Ojibwa (Hallowell 1955, 1960). On Viveiros de Castro’s reading, Hallowell attributes the clothing trope to postcontact indigenous re-interpretations. Faced with skeptical colonists, the Ojibwa came to understand metamorphosis as not an essential change, but a change in appearance. He rejects Hallowell’s supposition, arguing that he “could not grasp the force of the indigenous idiom . . . because of his implicit belief that metamorphosis is in fact impossible, or rather, that it could only be a belief, a representation of the Ojibwa” (Viveiros de Castro 2012, 137). Treating cognitive difference in terms of a difference in belief, Viveiros de Castro contends, distorts those differences and fails to take them seriously. However, if the belief idiom is not deployed, the ethnographer must report that when the shaman undergoes metamorphosis, he actually enters the spirit world. Viveiros de Castro and other proponents of the ontological turn seem to embrace this consequence in the name of a full-blooded reflexivity. Anthropology, Viveiros de Castro argues, has treated cultures as many, but nature as one. He wants to invert this order: culture is one, and natures are many.

Viveiros de Castro, of course, revels in striking assertions. But why should we reject the idiom of belief in anthropological discourse? Holbraad and Pedersen unpack Viveiros de Castro’s idea in two arguments (Holbraad and Pedersen 2017, 184-93). Holbraad and Pedersen begin their account of de Castro’s thought by noting that the ethnographic encounter begins as a problem of translation. It is not just the translation of one language into another, but the translation of an alien field of concepts. Sometimes, these differ in their commitments about what sort of things exist. In these cases of ontological difference, especially, it has felt natural to anthropologists to invoke the notion of “belief.” Viveiros de Castro identifies this initial framing as problematic: it is why Hallowell “could not grasp the force of the indigenous idiom.” The difficulty is that characterizing cultural differences in terms of “belief” sets up an epistemological asymmetry. The ethnographer’s problem is now one of explaining why “they” believe differently than “we” do (Holbraad and Pedersen 2017, 192). Starting from a description of local beliefs results in the one-nature-many-cultures idea that Viveiros de Castro wants to invert.

If we refrain from framing the ethnographic encounter with ontological difference in terms of belief, how is it to be framed? In answer to this question, Holbraad and Pedersen introduce the distinction between intension and extension. The distinction is quite familiar from linguistics and philosophy of language. The extension of a concept is the set of things of which it is true. ¹ A translation that preserved extension would be one where the two words picked out the same set of objects. “The intension of an expression,” on the other hand, “comprises the criteria (i.e., the sufficient and/or necessary conditions) for determining its extension” (Holbraad and Pedersen 2017, 188). Given this conceptualization of the intension/extension distinction, translation that preserved intension would necessarily preserve extension. But not vice versa: since the same set of things can be identified with distinct criteria, translation might preserve extension without preserving intension.

To characterize the ethnographic encounter in terms of a difference in belief, according to Holbraad and Pedersen, is to prioritize extension over intension. When extensionality is prioritized, ethnographic subjects are understood to have beliefs about the same things recognized by the ethnographer, but they either specify unusual sets of familiar things or they specify familiar things in unusual ways. Anthropological functionalism notoriously relied on such interpretations. A functionalist might say that Ojibwa talk of bearwalking really refers to practices that help maintain social status and power, not to a true metamorphosis. There is a natural sense in which such interpretations do not “take seriously” the alternative conceptualizations. Our ethnographic interlocutors insist that shamans undergo metamorphosis. These are prima facie differences in extension. The ethnographer cannot accept the existence of metamorphosis, and so treats her interlocutors as specifying a familiar thing (e.g., mechanisms for maintaining social order) in an unfamiliar way (as a “shamanistic metamorphosis”).

The alternative to prioritizing extension over intension is to prioritize intension over extension. The ethnographer is faced with a prima facie conflict between what she is committed to and what her interlocutors are committed to. Prioritizing intension resolves the conflict by revising the intension of the concept the anthropologist is using in her initial translation. Shamans are not just socially powerful individuals, but spirit beings as well. The ethnographic imperative to understand others in their own terms, or as Malinowski put it, “to grasp the native’s point of view, his relation to life, to realize his vision of his world” (Malinowski 1922, 19), seems to demand such a strategy. But, since intension determines extension, revising intensions means shifting ontological commitments. And of course, the intensions of concepts are interlaced; changing the intension of one concept will require adjustments to an indeterminate number of others. Working her way into the conceptual field of a group, the ethnographer discovers not different ways of conceptualizing the world, but different worlds.

Summing up Holbraad and Pedersen’s first argument against belief, to characterize the ethnographic encounter in terms of a difference in belief prioritizes extension over intension. It treats one’s interlocutors as speaking about familiar things in unfamiliar ways and thereby fails to take them seriously. Rejecting the idiom of belief permits the prioritization of intension. Intension determines extension, so understanding one’s interlocutors means taking their existence claims seriously. The Ojibwa do not (just) believe that shamans are spirit beings who undergo metamorphosis; shamans are spirit beings who undergo metamorphosis.

Holbraad and Pedersen have a further argument that treating difference in terms of “belief” fails to take the ethnographic interlocutors seriously. “Belief,” they write, “is what we call a view to which we are reluctant to commit—a view we can’t quite bring ourselves to take entirely seriously” (Holbraad and Pedersen 2017, 192). To distance oneself in the way suggested by “they believe that . . .” is to project difference as error. It treats what is really the anthropologist’s failure to properly describe or understand her interlocutors as a mistake on their part. It silences and thereby dominates their voice in a politically unacceptable way.

If the idiom of belief is avoided, then something else needs to function as an index of ontological difference. After all, anthropologists have the promiscuous habit of investigating many ethnographic worlds. It would be absurd to simply replace one ontology with another like a fickle teenager. To resolve this issue, Holbraad and Pedersen adopt the metaphor of “modulation” to express the way in which the anthropologist lives in multiple worlds. To modulate, in music, is to change keys. Different keys do not contradict one another, nor does playing in one key preclude playing in another. But while playing in a key, one must commit to it. “Modulation” lets the different ontologies be marked as different, without either collapsing into incoherence or creating unbridgeable gaps.

Holbraad and Pedersen’s version of the ontological turn has three elements: reflexivity, conceptualization, and experimentation. The arguments against belief support that part of the project involving reflexivity and conceptualization. Because intension determines extension, being open to the ways in which her interlocutors conceptualize their world requires the ethnographer to recognize that their world is different from the one she has assumed. The way that the Ojibwa understand shamanistic metamorphosis challenges anthropological preconceptions about what persons are. Reflexivity then turns this challenge back onto anthropology:

. . . the ontological here is meant as a call to keep open the question of what phenomena might comprise a given ethnographic field and how anthropological concepts have to be modulated or transformed the better analytically to articulate them. (Holbraad and Pedersen 2017, 11)

Ethnographic encounters thus provide material for new anthropological conceptions of what things are.

Whatever the virtues of Holbraad and Pedersen’s larger vision for anthropology—a question to which we will return—each of their arguments against belief has a critical weakness. With respect to the first argument, Holbraad and Pedersen are correct to think that framing the ethnographic encounter in extensional terms is troublesome. However, invoking the notion of belief does not create an extensional framework; quite the opposite. It has been a commonplace of semantics since Frege that the phrases within the scope of “believes that . . .” cannot be understood extensionally. The predicates “triangle” and “three sided figure” have the same extension but different intensions. Sentences (1) and (2), then, are examples of sentences that are extensionally equivalent but intensionally not equivalent.

But one may believe (1) without believing (2). Hence, the corresponding sentences within the scope of “believes that . . .” are neither intensionally nor extensionally equivalent:

Semantically, to put a sentence within the scope of “believes that . . .” is precisely to cut the extension away and to focus solely on the intension, that is, the content of the belief. Therefore, everything that Holbraad and Pedersen say about treating “translations between native and anthropological conceptualizations as a matter of modulating intensions rather than fixing extensions” (Holbraad and Pedersen 2017, 188) applies if native conceptualizations are treated as the intensions of their beliefs.

Where the first argument suffered from an undernourished semantics of “belief,” their second argument suffers from an emaciated pragmatics. Holbraad and Pedersen are correct to say that “belief” can pragmatically indicate a lack of commitment on the part of the speaker. Notice, however, this primarily occurs in the first person, and in contexts where the conversational expectation is for me to say the sentence plainly, rather than adding the belief operator. So, if I choose to say “I believe that I turned off the gas” rather than “I turned off the gas,” I signal an uncertainty or lack of commitment about whether the gas is off. However, when belief is attributed to others, in either the second or third person, matters are different. Consider:

Holbraad and Pedersen argue that (6) signals uncertainty about (5). It is true that by uttering (6), the speaker is not committed to (5). But that is a different matter than being “reluctant to commit” to (5). If Holbraad and Pedersen were correct that doubt about the proposition believed is the pragmatic default of third person belief attribution, it would take a special pragmatic circumstance to attribute true beliefs to others. But it does not: “Jane believes she turned off the gas and I agree” requires no special pragmatic circumstance to be understood. Hence, attributing a belief to another person need not suggest a suspicion that the belief is false. Therefore, while treating a failure to understand as an error on the part of one’s interlocutors is an important methodological and political error, it does not have its source in the ethnographic attribution of beliefs.

Rather than fully breaking away from the ethnographic attribution of belief, Holbraad and Pedersen’s view eschews the word but deploys an equivalent notion. Different ontologies recognized by the anthropologist have to be marked in some way as distinct from each other. “Modulation” is a nice metaphor, but the “they believe that . . .” locution has all the same formal features: it partitions ontologies so that their commitments do not conflict, and it allows the ethnographer to characterize the intensions of her interlocutor’s concepts independently of their extension. That modulation is formally equivalent to belief is not surprising, nor is it inconsistent with Holbraad and Pedersen’s larger view. While they acknowledge the anti-representationalist moments of the ontological turn, their own view stops short of a full-blown rejection. The “central concern” of their version of the ontological turn is “conceptual definition”:

Simply: instead of treating all the things that your informants say of, do to and with things as modes of representing the things in question, treat them as modes of defining what those things are. (Holbraad and Pedersen 2017, 213 emphasis in original)

If conceptual definition is the central concern of the ontological turn, the rejection of representationalism can at most be half-way, since definition and concept are prima facie representational idioms. To say that “Cuban diviners do not believe that powder is power, but rather define it as such” (Holbraad and Pedersen 2017, 222, emphasis in original) merely swaps one form of representation for another.

Holbraad and Pedersen’s version of the ontological turn, then, does not break free of the representationalist paradigm. This, in itself, is not a knock-down argument; representationalism is an enduring paradigm for a reason. However, their version of the turn does not truly reject the centrality of belief to ethnographic description, even if it eschews the word and, perhaps, some of its politically inapposite connotations. If we seek an ontological turn where the rejection of belief is truly “the bottom line,” we must begin our accounting with a different ledger.

4. An Alternative Origin Story for the Ontological Turn

In their genealogy of the ontological turn, Holbraad and Pedersen look to figures who have deep structuralist roots. A structuralist analysis needs a field of ideas, which is why structuralists traditionally presuppose an idealist (in Bidney’s sense) conception of culture. To be sure, Holbraad and Pedersen’s ontological turn goes beyond the play of ideas, but it does so by committing to the reality of what is represented. However, neo-structuralism is not the only path forward from the tradition. Rodney Needham engaged in extensive defenses of structuralism throughout his career, and in so doing found a different—and more profound—reason to question the idiom of belief.

In Structure and Sentiment (1962), Needham took up the question of whether social institutions were best explained psychologically, or whether explanation could proceed entirely at the social level. Specifically, he compared Levi-Strauss’s structuralist explanation of unilateral cross-cousin marriage with the psychological explanation proposed by Homans and Schneider (1955). Not surprisingly, he sided with Levi-Strauss, and in so doing argued that appeal to individual psychological states is inappropriate in social analysis. In itself, this is an unsurprising conclusion. It serves to reinforce the point that the field of representations subject to a structuralist analysis are collective representations, not individual beliefs.

In Belief, Language, and Experience (1972), Needham turned to the next question: is “belief” appropriate at all as a term of ethnographic description? Needham’s meditation begins with the closing sentences of Evans-Pritchard’s (1956, 322) Nuer Religion:

Though prayer and sacrifice are exterior actions, Nuer religion is ultimately an interior state. This state is externalized in rites which we can observe, but their meaning depends finally on an awareness of God and that men are dependent on him and must be resigned to his will. At this point the theologian takes over from the anthropologist.

Evans-Pritchard wrote these words after 300 pages describing the Nuer’s system of belief. He clearly had no qualms about working out the content of Nuer religion, considered as a collective representation; he had no hesitation in asserting that “the Nuer” believe in this or that. He does hesitate when it comes to individual belief. What is the relationship between the “interior state” of individual belief and the collective “beliefs” described in Nuer Religion?

To conceive of belief as an “interior state” is to deploy what Section 2 called the “common conception” of belief, and what contemporary philosophers call the “folk psychological” conception of belief. Needham’s argument proceeds along both anthropological and philosophical lines. On the anthropological side of the argument, Needham looks to the translations of the English word “belief,” and finds that not all languages have a corresponding term. While ethnographers report on the beliefs of their interlocutors, their interlocutors do not always report on each other in this way. The notion of “belief,” then, is culturally contingent. It is just one of many possible ways to think about ourselves. Needham’s move in this part of the argument is consonant with the reflexivity of the ontological turn: others’ ontologies give us a critical perspective on our own.

The mere fact that there are alternatives to a concept or theory is, of course, not a reason to reject it. To complete the argument, he turns to our own ontology through a survey of philosophical work on the concept of belief. Here, he discovers that the two features of the ordinary concept (cf. Section 2) will not suffice as an analysis. He finds no agreement among the philosophers on what the core criteria for the concept might be. More profoundly, he fails to find a compelling phenomenological basis for the philosophers’ disagreements. The content and attitude that define belief do not seem to be directly present to awareness. He concludes,

The specific argument of the investigation that I have undertaken here is that the notion of belief is not appropriate to an empirical philosophy of mind or to an exact account of human motives and conduct. Belief is not a discriminable experience, it does not constitute a natural resemblance among men, and it does not belong to “the common behaviour of mankind.” (Needham 1972, 188)

The common conception of belief does not correspond to anything. The idiom of “belief,” on Needham’s view, is a culturally specific category for classifying mental states. Its particular meaning for us is a product of our history and social development.

While Needham’s conclusion that the concept of belief is “not appropriate to an empirical philosophy of mind” is controversial, were he to conduct his survey today, he would find many philosophers sympathetic to his view. Stich (1983), Churchland (1986), and Schiffer (1987) argue for similar conclusions, as does Davidson (2001) in his later years. And the case against the folk-psychological conception belief is further buttressed by advances in neuroscience. However, fully defending such a thesis is beyond the scope of this essay. The point here is that there is ample material in anthropology, psychology, and philosophy for a direct argument that beliefs, conceptualized in representationalist terms, simply do not exist. That is the bottom line on an entirely different kind of ledger.

It is interesting to note that, unlike Holbraad and Pedersen, Needham does not recommend eliminating the word “belief” from our technical vocabulary and replacing it with something more suitable. Needham (1972, 191) remarks,

What we must do . . . is to give up the received idea that this verbal concept [of belief] corresponds to a distinct and natural capacity that is shared by all human beings. If we are persuaded to do this, then there remains no further occasion to consider any improvement in the use of the word “belief” as a term of ethnographical report.

The idea that there is no room for improvement comes from the deep Wittgensteinian roots of Needham’s analysis. “Belief” is part of a language game we use for getting along, and as such it is perfectly in order. The problems arise when we take our words to be carving nature at its joints. ² When doing ethnography or social theory, we need to be aware that we are using one language game to describe the rules of another. Those language games need not correspond, nor need they refer to the same underlying reality. Like other views of cross-cultural understanding influenced by Wittgenstein, such as Winch’s and Geertz’s, the result is a behavioral (but not behaviorist) understanding of individual belief. What we need to know about other people is on the surface, not hidden deeply in their minds.

Needham’s Wittgensteinian commitments permit him to reconcile the structuralist’s dependence on collective representations with the lack of a natural and universal category of belief. Cultural analysis depends on patterns of behavior exhibited by the whole group. Even though he was a structuralist, this view puts Needham squarely into the company of Winch and Geertz. All three take for granted a concept of culture that treats systems of thought as social facts. The behavioral foundation of these social facts permits all three to sidestep a Durkheimian ontology. Ontologically, there is nothing more to social representations than patterns of behavior.

Unfortunately, it is just here that Needham’s position has a fatal flaw. On a Wittgensteinian view, meaning—whether the meaning of words or the content of belief—is a property of the community. It is a pattern of use that different individuals can participate in. Neeedham’s critique of belief was aimed at the idea that it was a particular, introspectable internal state. Since content cannot be introspectable, anything that we might identify as a belief would have to get its content from the broader community. Ethnographic description captures these patterns with what Geertz called “thick description.” The attribution of beliefs to individuals is thus derivative of attribution of collective representations. Therefore, while Needham rejects the representationalist concept of individual belief, he remained committed to an idealist concept of culture that could not be entirely shorn of its representationalist commitments.

As discussed in Section 2, the concept of culture as a collective representation was roundly criticized in the decades since Needham, Geertz, and Winch wrote. Geertz was more often the explicit target of such criticism, but structuralists like Needham’s were vulnerable as well. There were, as is well known, a cluster of arguments against the classic culture concept (see Risjord 2012, for a review). One particularly damning one, for our purposes, had been foreshadowed by the Boasian critique of the functionalists. In the 1930s, Lowie and Radin argued that cultures were much less consistent and coherent than their portrayal in the ethnographic literature. In the 1970s and 1980s, this line of critique argued that treating cultures as homogeneous made it impossible to see the deep differences that often existed among people who “shared” a culture. Resistance to and manipulation of collective representations was often more important than conformity.

The critique of culture of the 1970s and 1980s is devastating to the Wittgensteinian picture of culture and belief. Without an underlying unity, it is impossible to characterize the patterns of behavior and speech that are the basis of thick descriptions. And the importance of individual difference means that belief cannot be understood in terms of a conformity to the community meanings. The concept of culture derived from the Wittgensteinian picture of meaning is an “idealist” culture concept in Bidney’s terms. Structures (Needham) or thick descriptions (Geertz) are systems of meaning. The breakdown of these systems is the breakdown of the idea that culture can be understood in representationalist terms.

If his arguments are successful, Needham’s work shows that a structuralist analysis not only may do without appeal to individual belief, beliefs (as commonly conceived) do not exist. If the rejection of belief is truly the “essential ‘move’ of the ontological turn,” then anthropological theorizing that proceeds on this basis is deserving of the name. However, if this essential move is part and parcel of a full-throated rejection of representationalism, then structuralism cannot be maintained, at least in anything like its classical form. It requires an idealist version of the culture concept, and that conception of culture must succumb to the Writing Culture critique.

5. Ontology as Ecology

The upshot of the foregoing discussion of Needham is that there is a twofold challenge to understanding human cognitive alterity. Rejecting representationalism requires successors to both the common conception of belief and the idealist conception of culture. The concept of belief functioned to characterize what others think without commitment to the truth or falsity of those thoughts. The notion of belief, however, captures individual differences, whereas the phenomenon of human cognitive difference occurs at the level of groups as well. The classical concept of culture filled this role.

In the aftermath of Writing Culture, practice theory emerged as a popular successor to idealist conceptions of culture. Bourdieu (1977, 1990) is perhaps the most prominent theorist, though Ortner (1984) is also an important source. Like Needham and Geertz, Bourdieu has roots in Wittgenstein. There are, however, crucial differences that help practice theory avoid the arguments that scuttled the idealist conception of culture. The central difference can be found in an ambiguity in the Wittgensteinian slogan “the meaning of a word is its use in the language” (Wittgenstein 1953, Sec. 43). Early interpreters treated “use” as describing current and past uses of a word. This requires the kind of consistency, coherence, and boundedness to language communities that makes it vulnerable to the argument presented in Section 4. Practice theory takes “use” to be future-oriented and prescriptive. Practices are actively created and maintained by agents. They thereby give up the idea of consistency and coherence among the practices prevalent in a social group. Stressing agency, their analyses often focus on the way that agents either maintain or undermine cultural norms through their interactions with each other. These differences make practice theory immune to at least some of the central arguments of the Writing Culture critique.

Practice theory alone, however, is not a form of ontological anthropology. Nothing in practice theory requires one to reject the idiom of belief. Indeed, many practice theoretic analyses retain a very traditional notion of agency and the attendant representational picture of the mind. In addition, in its standard formulations, it preserves a strict culture-nature dichotomy that is emblematic of idealist conceptions of culture. Practices are juxtaposed to the natural world, which is treated as fixed independently of practice. To serve as part of an anti-representational, ontological anthropology, practice theory needs to efface the culture-nature dichotomy.

Needham’s conclusion that nothing corresponds to the common conception of belief has been developed in recent philosophy of mind and language. One popular idea, semantic externalism, holds that content of belief and speech is determined partly by the objects they are about. Our capacity to refer to a kind is independent of the representational devices we use for the reference (Burge 1979; Kripke 1972; Putnam 1974). We can pick out a substance, say gold or water, independently of the intension of the words “gold” and “water.” Because the contents of beliefs are not determined by processes internal to the mind, they are not present to consciousness, nor are they introspectable. Semantic externalism is thus a start on an anti-representationalist successor to the common conception of belief.

While semantic externalism is a start, it cannot be satisfactory. It provides a very poor basis for theorizing human cognitive difference. Semantic externalists do not typically problematize the objects that partly constitute belief. These are taken as given independently of any human activity or thought. Like the interpretations that prioritize extension over intension, the best that such views can do is to regard humans as speaking about familiar objects in unfamiliar ways. Indeed, some externalists would follow Davidson in arguing that large-scale cognitive difference is impossible.

Active externalism makes a better candidate for an anti-representational conception of belief than semantic externalism (Clark 1998, 2008). It contends that human cognitive processes are ecological. Our cognitive system recruits and depends on aspects of the environment. Memory is a particularly appealing cognitive system for active externalist treatment because of the variety of technologies for memory enhancement humans have invented, from cuneiform tablets to the Internet. Information processing systems, too, can easily be seen in active externalist terms: think of astrolabes, stonehenge, or slide rules. It is important to separate two possible views here. One might take the claim as a fairly common-sense point that external objects cause beliefs or that we use information processing tools like calculators as evidence for belief. The active externalist wants to make a stronger claim. The cognitive process (of memory, belief, affect, or action) is composed of the environment and the brain in dynamic interaction.

Understood in its strong form, active externalism about cognition is a radical revision of our ordinary conceptions of belief, memory, and action. As noted in Section 2, the conception of belief we have inherited from modern philosophy holds beliefs to be internal states, introspectable, and have a semantic structure that mirrors states of affairs in a way that makes them capable of truth or falsity. To see the contrast with representationalism, consider the way that an ancient observatory structure like stonehenge would be part of the process of belief formation on an active externalist view. The structure is not consulted for evidence about the current phase of the planting cycle (assuming that stonehenge was used in this way). Rather, positioning oneself bodily (perhaps by participating in the appropriate ritual activities) within the structure makes the time apparent. The belief about the time is not an internal state and its truth or falsity is not a function of its semantic structure. It is not the product of introspection, though once the information has been encoded in the brain, it can be retrieved by memory processes (that may themselves depend on the environment). Active externalism is thus a strongly anti-representationalist conception of cognition.

The combination of practice theory with active externalism breaks down the conceptual barrier between nature on one hand and the mind or culture on the other hand. Objects are actively made to be components of our thought. It is crucial not to conceptualize objects as given independently from practices. What an object is for us depends on how it is taken up in our practices. This salutary move creates a puzzling circularity. Our embodied cognitive coupling with objects is a kind of practice. This means that objects are partly constitutive of practices. How can objects also be constituted by practices? The answer is that there are two sorts of constitution involved. Practices should be understood as having non-human parts; they are not just the reproduction of behaviors. They are the reproduction of behaviors, objects, and their interaction. On the other hand, objects are given their conditions of identity by human interaction with them. Practices determine when two objects are tokens of the same type or the same object at different times.

While he has not adopted the mantle of ontological anthropology, Tim Ingold is the anthropologist whose essays both work within and work out the foundational ideas sketched in this section. He adds a Heideggerian dimension to the philosophical background and a Gibsonian analysis of perception. The essays of The Perception of the Environment (Ingold 2000) are a paradigm of anthropological theorizing that rejects the traditional conceptions of belief and culture and self-consciously deploys a full-throated anti-representational perspective on human activity.

The fictional genealogy of the ontological turn developed in this section and the last, then, runs as follows. The philosophical origins lie in the anti-representational reflections of Wittgenstein and Heidegger. These philosophical ideas are picked up in the work of Needham, and refined by the Writing Culture critique. Successors to the traditional conceptualizations of belief and culture are developed in Clark’s active externalism and Bourdieu’s practice theory. Ingold combines these ideas with Gibson’s ecological account of perception and develops it in the context of rich ethnographic materials. While these authors are not the typical heroes of the ontological turn, I submit that their work constitutes a philosophically robust foundation for ethnographic description and anthropological theorizing that is entirely deserving of the name.

6. Conclusion: An Ontological Approach to Human Difference

The discussion so far has been relentlessly abstract. One might wonder whether the program sketched above can really provide satisfying and interesting accounts of human cognitive difference. By way of conclusion, then, I want to contrast Ingold’s (2000, Chapter 6) analysis of Ojibwa shamanistic metamorphosis with Viveiros de Castro’s (2012).

Viveiros de Castro treats metamorphosis as a philosophical problem in the sense that it presents an intellectual puzzle or paradox: what would it mean for a human being to transform him- or herself into an animal? What conceptualizations of person, body, and animal are required for such a transformation to be regarded as not only possible, but actual? He describes his method in these terms:

a small sample of texts . . . are used as a springboard for a thought-experiment consisting in abstracting and generalizing a set of ideas about subjects and objects, bodies and souls, humans and animals, and then sketching what could be called the “virtual ontology” underlying these abstracted generalizations. (Viveiros de Castro 2012, 64)

Treating the indigenous ideas as a systematic ontology is Viveiros de Castro’s way of taking them seriously. It should be noted that doing so is not very novel. Radin’s (1927) Primitive Man as Philosopher inaugurated such an approach in anthropology. Hallowell’s discussion of the Ojibwa conceptualization of “person”—on which both Viveiros de Castro and Ingold are reflecting—follows Radin, calling the project “ethno-metaphysics” (Hallowell 1960, 20). Radin, Hallowell, and Viveiros de Castro are deploying a method indistinguishable from traditional history of philosophy. The ancient Greeks, not to mention the ancient Arabs, Indians, Chinese, Tibetans, and so on, worked in conceptual fields very different from ours in the 21st century. The goal of the historian of philosophy is to articulate a coherent and consistent account of the underlying ideas, and to treat these ideas as viable alternatives to those current today.

The substance of Viveiros de Castro’s account of shamanistic metamorphosis depends on the notion of perspectivism. This is the idea that the difference between humans and animals is a difference in perspective. To predator animals, humans appear as prey; to humans, animals appear as prey. Each appears as human to itself. Humans and animals thus share a spiritual form and are different in their bodies. It is impossible to do justice to Viveiros de Castro’s subtle, wide-ranging, and insightful development of this idea. The point for now is that the possibility and significance of metamorphosis is understood in the light of perspectivism. He sums it up this way:

We must now face the question of metamorphosis. My point here will hardly surprise you, I am afraid: I take metamorphosis as just a synonym for “perspective,” or rather, for the exchangeability of perspectives characteristic of Amerindian ontologies. (Viveiros de Castro 2012, 145)

When undergoing metamorphosis, the shaman has the power to change perspectives. He sees our world from “the other side,” and from the human perspective, he has become an animal.

Viveiros de Castro’s analysis of shamanistic metamorphosis is “ontological” in precisely the same sense as a treatise on Plato’s metaphysics is ontological. That is, it is concerned with Amerindian ontological categories and concepts, rather than their epistemologies or value theories. It proceeds entirely at the level of ideas. Nothing in the procedure or the substance of the interpretation requires a commitment to anti-representationalism. Indeed, because it treats content as independent of anything material, it is representationalist through and through.

Much in Ingold’s discussion overlaps with Viveiros de Castro’s. He too is concerned to extract consistent conceptualizations out of the ethnographic material. He too develops a conceptualization of philosophical notions like perception, life, and the self. He too takes these as serious intellectual possibilities and argues that they are superior in certain ways to the western philosophical framework. However, Ingold does not end the inquiry with the “ethno-metaphysics.” He also articulates the practice through which the world might come to be populated with powerful individuals who can take animal form.

Ojibwa lifecycle rituals include a puberty ritual wherein a boy would fast alone in the forest. The goal was to see a vision of a “grandfather,” a guardian who would help him later in life. Recounting Hallowell’s ethnography, Ingold writes:

In one account, for example, a boy encountered a human-like figure in his dream, who then turned into a golden eagle. This person was the “master” of the eagles. The boy, too, was transformed into an eagle in his dream—thus winged and feathered, he flew to the south with his new protector, before returning to the point whence he originally departed. (Ingold 2000, 93)

Dreams had epistemological significance for the Ojibwa. Hallowell discusses this extensively in his original presentation:

Although there is no lack of discrimination between the experiences of the self when awake and when dreaming, both sets of experience are equally self-related. Dream experiences function integrally with other recalled memory images in so far as these, too, enter the field of self-awareness. (Hallowell 1960, 40)

Dream experiences are genuine, knowledge-producing experiences. The vision of a golden eagle, then, is not a hallucination. It was among the boy’s genuine, veridical experiences, albiet one that happened under the special circumstance of a vision quest.

The Ojibwa thus had a practice wherein young men were put in position to have the perception of metamorphosis. The experience emerged from a particular ecological relationship of the boy to his environment. Experiences of this kind are formative. This experience would then influence his later behavior and “contribute to the shaping of a person’s own sense of self, and of their attitudes and orientations towards the world” (Ingold 2000, 99). A golden eagle grandfather would require particular ritual and taboo observances, and the boy would be attuned to dreams of eagles or glimpses of eagles. Just as teaching the boy to hunt deer involves the practice of putting him in a position to see deer in the forest, turning the boy into a man involves the practice of putting him in a position to see a powerful grandfather.

Through these practices and others, the Ojibwa created an environment populated by powerful spirit beings. Having had the experience of metamorphosis in dreams, and knowing that bears are powerful beings, an unusual encounter with a bear is an encounter with a bear-walking shaman. The cognitive difference between the Ojibwa and myself is made quite explicit by Ingold’s account. It is not a matter of differing beliefs, in the traditional, representationalist sense, though it does involve differing ontological (and epistemological) commitments. The crucial point, however, is that on the version of the ontological turn exemplified by Ingold’s analysis, I cannot come to live in their world simply by doing philosophy. Pace Holbraad and Pedersen, it is therefore not a matter of saying that the Ojibwa define bears or shamans in a particular way. Rather, I would have to have an ecological relationship to the environment constituted by a specific set of practices. And to have that, I would have to be raised as an Ojibwa boy. That’s about as ontological as it gets.