©1989, 1996 Finn Sivert Nielsen

A general problem in anthropology, which is rarely treated systematically, is that of the "quality" or "weight" of meaning. In its simplest form the issue has always been recognized. Even the most rudimentary collection of data presupposes an evaluation of some information as "better", "carrying more weight", than other. Later, in analysis and generalization, such evaluations gain progressively in importance, indeed, in a certain sense a theory is nothing but a system of scales against which the qualitative importance of various types of information may be assessed.

This applies to all science. But social scientists, and anthropologists in particular, face an additional complication: our data talk back. This becomes a theoretical problem when (as is often the case) the relative analytical importance attributed to acts, statements or objects by the analyst does not coincide with the relative moral and esthetical importance attributed to them by his or her informants (see Larsen 1979). We claim, in such cases, to "know better" than they do (at least in the sense that we assert our right to study them). Any anthropological theory must either justify this claim or legitimize it as doxa – and go on to ignore it. Such legitimation, I suggest, is frequently what is sought when we are warned that qualitative evaluations of meaning controvert such pillars of anthropological orthodoxy as cultural relativism and participant observation – or indeed, instigate bigotry and racism. But we cannot think about people's lives without "judging" them, and our theories should reflect this fact, not deny it. Postmodernist anthropology, despite its concern for "reflexivity", has done little to address this issue, since its proponents often adhere to a strict relationism, and label any attempt to assign existential weight and priority to meaningful expressions as "essentialism". In contrast, medical anthropologists, and anthropologists working in the boundary zone towards cognitive science and psychology, show signs of a different appreciation. As Bateson suggests (Ref.), such empirical interests stimulate attempts to "diagnose" actors, institutions or social systems in terms intrinsic to the human condition: as "healthy" or "diseased".

It is instructive to note a dual emphasis in both Bateson's pioneering efforts and the work of modern psychologically oriented anthropologists such as Gananath Obeyesekere (1990): both imply a link between "health" and "beauty". In Bateson's work this linkage is established by the common cybernetic terminology utilized in analyses e.g. of alcoholism and art (1971, 1967). In Obeyesekere's case, the anthropologist's evaluations of his informants' state of mental health (in neo-Freudian terms) allow him to establish a continuum from "symptom" to "symbol", from cultural expressions that are crude, unformed images of common psychological dilemmas with no therapeutic value, to expressions that are esthetically balanced and psychologically sound responses to the same problems (pp. 3-14).

We are still some way, however, from an explicit discussion of the theoretical "weight" of our own qualitative judgements. Fieldwork is an analytically and emotionally taxing experience. It is notoriously difficult to carry through according to preconceived programs and plans, and much of the best data we acquire (e.g. on conflicts) is a result of coincidence. ("I just happened to be there at the right time".) In the face of these obvious sources of bias, it is surprising that we continue so confidently to separate "good ideas" from prejudice, "significant cases" from trash, the "typical" from the "atypical", without defining the analytical basis for such judgements.

This article addresses these issues. It does so however, within a wider theoretical framework, which has relevance for a series of questions ranging from action theory to social evolution. My contention is that the problem of "meaning quality", which is most forcibly encountered in the fieldwork context, has general theoretical repercussions, and that our failure to make our guiding assumptions during fieldwork explicit is symptomatic of a more general theoretical problem that we cannot afford to ignore.

In their programmatic work of 1966, Peter Berger and Thomas Luckmann argue that "social action" consists of three separable aspects, which they term externalization, objectivization and internalization. The meaning of these terms is roughly as follows: Externalization is an actor's projection of goals and intentions into the world. What we have learned to believe in and identify with is acted out, thereby changing or preserving the objects by which we are surrounded. Objectivization occurs when the act is seen as separate and independent of the actor with whom it originated. Having become objectivized, the act loses its ties to the intentions of the subject and attains a vitality of its own. The act seems to become a subject with whims and a will over which the actor has no control. In internalization the act "returns to its origin". The subject approaches it as a given and unknown quality which he/she adapts to or learns to understand.

What is implied is an idea of social action as a three-stage, cyclical process exiting from and returning to an individual subject. The simplicity of this concept should not mislead us. We may start by noting that the model differs radically from the traditional concept of "interaction", which has been dominant in anthropology at least since Marcel Mauss's Essay sur le don (1925). This concept presupposes that all action involves some kind of exchange. Something is given and taken, there is bargaining going on, resulting in a compromise which is (or is not) acceptable to more than one party: the point of all this being that in the course of a repetitive process of transaction a binding moral relationship between the parties is established. Implicit in this concept, though rarely stated, is the suggestion that empirical situations are analytically relevant, if only if two or more actors are present. The lone individual that stands outside, and independently evaluates phenomena (as the anthropologist must do during fieldwork), is therefore not encompassed by it at all – his acts are not regarded as social acts. This weakness is not shared by Berger and Luckmann's concept of "social action". The construct demands an external world into which the act can be projected and from which it can return, but that world need not be populated. It may consist of inanimate objects (art, texts) or of people regarded (in the mass or individually) as inanimate objects – which is how anthropologists see the "informants" that supply our "data".

I shall speak of "social action" as a subjective or intentional concept, while "interaction" is relational or dialectic. Let me expand somewhat on this difference. From a strictly dialectic point of view the act has neither origin nor destination. Both its subject and its object are epiphenomena of the relationship between them. The act itself is an instance in this relationship, a single pulse in the continuing process of interchange – which maintains the illusion that subject and object are real by "keeping them apart". A classical example of this way of thinking is George Herbert Mead's (Ref.) theory of the Self as an aggregated product of its relationships to others. Mead imagines that an infant knows no difference between itself and others. Nevertheless, it is immediately drawn into interaction with them: it rests close to mother's breast, then closeness is interrupted. Separation teaches otherness. In the baby's consciousness it leaves a residue, a reflection of the Other as Absence, which is in fact a context-specific and incomplete experience of Selfhood. After many separations some Othernesses are perceived to be connected. Gradually the image of the outer world is transformed into a mosaic of such compounded Others, contrasted to a similar patchwork of partial Selves within consciousness. At a certain point the child discovers in nearly all Otherness a single, Generalized Other – and in the same breath a more or less completely integrated Self arises. Contrast and relationship by stages generate the Other and the Self – that which is "not Other". Dialectically speaking, the "Self-and-Other" are a single synthesis arising out of the multiple contradictions in which interaction involves the child.

More generally, the dialectic viewpoint teaches us that all concepts involving purpose, meaning, identity – all objects, roles, symbols, persons, societies etc. – have a purely temporary and illusive reality which is ultimately derived from the position they occupy in a larger whole – a context (see Chapter 2). Their seeming content is actually a product of their form – of the contrasts, boundaries, relationships, oppositions, differences (the terminology varies), which tie them into and delimit them from their context. All meaning is situational, relational, positional. Gregory Bateson's formulation is classic:

"I suggest to you, now, that the word `idea,' in its most elementary sense, is synonymous with `difference'... In fact, what we mean by information – the elementary unit of information – is a difference which makes a difference." (1970a p.453)

Fredrik Barth, in his seminal discussion of ethnic identity, is no less unequivocal, if somewhat less emphatic:

"When defined as an ascriptive and exclusive group, the nature of continuity in ethnic units is clear: it depends on the maintenance of a boundary. The cultural features that signal the boundary may change, and the cultural characteristics of the members may likewise be transformed, indeed, even the organizational form of the group may change – yet the fact of continuing dichotomization between members and outsiders allows us to specify the nature of continuity, and investigate the changing cultural form and content." (1969, p.14)

To cite a final authority: Claude Lévi-Strauss, in his critique of Marcel Mauss's reflections on the "power in the gift", stresses that

"... the primary, fundamental phenomenon is exchange itself, which gets split up into discrete operations in social life; the mistake was to take the discrete operations for the basic phenomenon." (1950, p.47)

Difference, boundary, exchange – these are dialectic terms, and the theories they generate effectively deny substantive content to subjects, symbols, groups and things. In contrast, the intentional point of view sees the Self as the source of acts, and hence of meaning. The subject is a locus of intent, who purposefully directs his activity at specific goals or objects which he influences or creates, and which, in turn, resist him in terms of their own intentionalities. The relationship is now reduced to a medium through which intentions are realized, and emphasis is transferred to the unique identity of the actor and the qualitative differences between the objects/objectifications of his/her action/perception. Thus, both subject and object are seen as entities with content, an inherent meaning and purpose of their own.

It seems obvious that these two points of view are complementary. Both capture a part of reality, while another part eludes them. But the dialectic view has such immense prestige in social science that we are often hard-pressed to find examples of analyses which do not take it for granted – which is perhaps why social science is often accused of neglecting "reality" as the man on the street experiences it. Thus, it is commendable that the "sociologists of knowledge" have sought an alternative model of action. It is still however, only a model, and its theoretical implications are very incompletely explored. As a result, most of its concepts are not very different from those of traditional – dialectical – social science; and its potential is rarely realized in analytical practice.

What is the status of our own concepts – the "facts" and "theories" on which research is based? Are they intentional or dialectic? Do they have meaning in themselves, or is their significance derived exclusively from their relationships to each other? Do they have content, or are they pure form? In The Structure of Scientific Revolutions (1962) Thomas Kuhn argues convincingly that the latter is the case. Briefly, his idea is that any corpus of scientific knowledge rests on a set of fundamental paradigms which determine not only how "facts" are interpreted, but what is to be defined as a "fact" in the first place. A scientist cannot see what his paradigm does not point out. Conversely, when paradigms change, sudden, radical inversions of the meaning of "the facts" occur. The significance of any scientific statement is thus determined by the nature of the context in which it is embedded. It is pure form.

Kuhn's perspective represented a timely contrast to the positivist idea of science as an accumulation of objective facts. Anthropologists in particular should be thankful for his insistence that we apply the same standards to our own world-view that we so happily imply in our study of other cultures. Still, something is lost, I think, if his standpoint is carried too far. It seems absurd to suppose a total relativity of ideas. David Herbst (1988) quotes an example that illustrates this. If a man is married, he is a husband. If he is divorced – this paradigm shifts. But where did the husband go? To Kuhn, the leap into a new context obliterates the husband. But to the man himself that need not be true. The idea that he is a husband (potentially, if nothing else) stays with him all his life.

Though contexts shift, some seed of meaning remains constant in an idea. Perhaps not in all ideas, and perhaps not in spite of all contextual leaps. Take the case of the divorced husband. We might object that the marriage context is enveloped in the wider context of his life. Marriage is only one element in this larger whole. Whatever is experienced in its restricted sphere is therefore also experienced in the context embracing it – and when the restricted context ceases, the experience retains its wider sense. Other examples may be treated similarly. The Bible has been a staple of Western culture throughout two millennia of violent contextual leaps, and although its meaning has altered, some kernel of constant significance doubtlessly persists – but this is true only in a Judeo-Christian setting; outside that context no continuity of interpretation is assured. Or take a third example, a 1985 exhibit of traditional Maori art in San Francisco (Te Maori). Its beauty was not lost on the (non-Maori) visitors, neither was a deep-seated impression of a sensitivity and ethic definitely neither American nor Western. But then, it has been pointed out that Western culture has specialized in abstract modes of understanding which easily subsume any local form of expression into their over-arching system. So we retreat to a last instance. A colleague once brought along two cassettes with Western music to a village in India. The peasants listened politely to a selection of rock, but were captivated by Mozart's Requiem. Again, this "innate attraction" can be explained relationally. The peasants' sympathy is probably due to chance similarities between the stylistic patterns of Indian and Western classical music – perhaps the result would have been reversed in a village in Africa where there would be a rhythmic affinity between traditional music and modern rock.

In all these cases we can juggle with contexts to explain the continuity of meaning – and this fact must not be dismissed lightly. Anthropology has proven beyond all reasonable doubt that there is no pure communication across contextual barriers – and some kind of contextual similarity is perhaps needed for communication to occur at all. I am very definitely opposed to the kind of ideational absolutism that insists e.g. that the ideas of Nietzsche or Marx necessarily (by virtue of their content) must lead to the excesses of Hitler or Stalin. Nevertheless the examples also indicate that communication across contextual barriers does occur, and that some ideas and expressions do "carry", perhaps to the very limits of the human community in time and space – a "context" which for social scientists is synomymous with no context at all. Some ideas even seem to have greater impact out of their "mother context" than in it – and not all of these are as warped out of all recognition as Marx and Nietzsche.

Thus, I shall assume that at least some ideas have an inherent meaning, a mana, a "life of their own". I shall refer to such ideas as "Figures", a term derived from Gestalt psychology. My use of this concept demands a certain clarification. To start with a purely subjective impression – I think a "figure" is often grasped intuitively. Take the following story: In the 1920s John V. Atanasoff became preoccupied with the idea of building an electronic computer. He worked on the project for years, pondering alternative solutions and deciding on some general principles – but by the winter of 1937 success seemed further off than ever:

"The evening had not begun with particular promise. It had, in fact, been so frustrating that he left his laboratory, got into his car and began driving eastward from the college at Ames at high speed, concentrating on his driving to take his mind off his troubles. After several hours he ended up some 200 miles away in the state of Illinois, where he stopped at a brightly lit road-house for a drink. `It was extremely cold and I took off my overcoat,' he recalled... `I had a very heavy coat, and hung it up, and sat down and ordered a drink, and as the delivery of the drink was made, I realized I was no longer so nervous and my thoughts turned again to computing machines. Now, I don't know why my mind worked then when it had not worked previously, but things seemed to be good and cool and quiet... I would suspect that I drank two drinks perhaps, and then I realized that thoughts were coming good and I had some positive results.'" (Mackintosh 1988, p.73)

Despite this understatement (he more or less invented modern computing that evening) we recognize the urgency of Atanasoff's description: suddenly, at times even in sleep, you simply "know" the solution. This story alerts us to an essential quality of "figures". They have an unmistakable immediacy, a force, a weight of implications and repercussions which you at first may only glimpse. You may be unable to prove your "knowledge" – it may take months or years or a whole life to work out how it is that you "know", or what specifically you have "learned". But the forceful reality of the figure itself cannot be denied.

Part of the reason for this is implicit in Atanasoff's story, which he told in 1967, thirty years after the event. In spite of the lapse of time, note his attention to detail. You hear and feel his physical surroundings – the cold, the heavy coat, the light and quiet after the long, dark ride. What I think he is saying is that he did not only come across a good thought in that road-house in 1937 – he encountered something more: a figure is not just an idea, it is an experience involving body as well as mind. This lends it a strange intimacy. It seems like living thing, a person. It may be sensual, capricious, amusing, captivating or deadly. You love and hate it. It haunts you when you want to be alone, then suddenly leaves you in the lurch when you need it most. It may be associated with pain, or happiness – or both.

"Like many intelligent men, Stone took a rather suspicious attitude toward his own brain, which he saw as a precise and skilled but temperamental machine. He was never surprised when the machine failed to perform, though he feared those moments, and hated them." (Crichton 1969, p.208)

During a course for graduate students of anthropology at the University of Oslo in 1989, two of my students spoke of this experience in the following way:

"Life seems so clear-cut during fieldwork. You're so different from your informants that you can't avoid simplifying and stylizing interaction. In normal social situations it's the ambiguities and nuances that create misunderstandings. Half of the time the fieldworker doesn't even notice these things, and so he's left with a crystal-clear, almost archetypical image of the foreign reality."

"When you leave home you think the excitement will come from your informants, and then it's a disappointment to discover that their lives are as dull as your own. But in the end you realize that what's really interesting is in your own mind – the sudden, intuitive grasp of interconnections, the flashes of understanding."

These statements have two important things in common. First, both involve the kind of "knowing" I described above – the "flashes of understanding" I refer to as figures. Secondly, they are both, it seems to me, themselves examples of such figures. They compress months of fieldwork into a compact crystal ball which our understanding can ponder indefinitely.

Some characteristics of these figures are worth noting. First, the fundamental insight in both cases is obviously true. The ideas are nothing but common sense. Given the fieldwork experience, or any other similar experience with the same intensity, the same pretensions, and the same bleak prospects of anything approaching complete success – both insights are self-evident. But the qualities which make them obvious also obscure them. After going through "all that" how can we admit to ourselves that this was all it amounted to? Being able to do so anyway demands an intuitive and rather painful honesty. Secondly, the ramifications of each statement are virtually limitless. These are non-trivial insights, fundamental ideas about the nature of anthropology and human understanding, which if explored deeply, may generate moral, intellectual and political dilemmas it would take lifetimes to work out. Finally, and as a result of the two preceding points, in real life both ideas are more likely to be abandoned than followed up. They are simply too intractable, too demanding, too much in opposition to the mainstream to be explored for any length of time without posing such a wide range of problems that professional life might become very uncomfortable.

It is my contention that all intellectual activity is based on insights such as these – and that the health of the intellectual enterprise depends on our exploring their consequences in depth in spite of all that entails. As Paul Feyerabend remarks:

"The ideas which scientists use to present the known and to advance into the unknown are only rarely in agreement with the strict injunctions of logic or pure mathematics and the attempt to make them conform would rob science of the elasticity without which progress cannot be achieved." (1975, p. 303)

One of the most striking and potentially hazardous qualities of figures is the aura of unfailing certainty with which they are surrounded. The "flashes of understanding" give an immediate, almost overpowering sense of truth. When we "see" in this way, we do not doubt – we know. And although experience shows that we are often right – still, we are sometimes wrong. What are the reasons for such mistakes? In one of George Simenon's novels, superintendent Maigret discusses this problem in connection with the solution of a murder case. You must isolate what you know from everything else – he insists. But when describing or even merely taking note of your certainties, you immediately start weaving a cocoon of rationalization and explanation around them. Perhaps you observed a suspect at point A and at point C. You automatically conclude that in the mean time he must have visited point B. But you do not know. The case cannot be solved as long as you cling to such assumptions.

The first step in solving a murder case – or in analysis – is thus to separate the true figures from the "ground" to which they cling like magnets to iron filings. As Maigret implies, the function of this ground is to tie various insights or figures together into a coherent whole. But we do this so automatically and in response to such an urgent need that we cannot trust our spontaneous grounding. It links some figures by suppressing others, or by deforming their true content. The real peril of intuition lies in this automatic grounding rather than in the figures themselves. It is what Hitler did to Nietzsche – silenced all but a part of his thought. In scientific analysis as in murder cases we therefore "clear the ground" as far as possible. We let the figures stand out as "naked facts" and let each "fact" speak for itself. Some authors even do this explicitly, as Bateson, who repeatedly starts an article or book with a long list of disconnected observations, which he afterwards tries, often without success, to connect (e.g. as in Mind and Nature (1979)).

It is important to note that "de-grounding" is not what is often referred to as "decontextualization". When we decontextualize a concept the point is to keep its meaning within the limits prescribed by a formal definition, to remove the metonymic associations that overflow onto it from its context. Figures cannot be treated in this way, for their meaning is intrinsic, and therefore not known ahead of time: therefore it cannot be defined. Each figure has its own content, its own story to tell, and if it is not allowed to tell the whole tale before we ground it, its potential will be violated. "De-grounding" is more similar to what a good author does with his persons – who are literary figures, rather than intellectual:

"If I create a river with two banks and place a fisherman on one of them and if I supply this fisherman with a furious temper and a rather ugly criminal record, then I can just start writing and give expression to that which necessarily must happen." (Eco 1983, p.29; my translation from Swedish)

The figure speaks on its own accord if we only let it. But what Umberto Eco neglects to mention is that two figures, once they have "told their tale", rarely agree about the facts. When we let a figure have its say we discover that it resists grounding. It has too much content, too separate and complex a "personality". Each figure contradicts the others we seek to combine it with – and its resistance can only be overcome by taking all these contradictions into account. So "clearing the ground" is only a first step. It leaves us with something very like one of Bateson's "lists", crisscrossed by paradoxes and ambiguities which must be addressed in depth when "re-grounding" is attempted. In analysis these contradictions are the core around which all further argument revolves.

The process of "re-grounding" – of building an argument – has as its ultimate goal to move each figure from a "list" to a precise location on the ground of a single canvas, thus creating a continuous and logically consistent "picture", in which none of the figures are silenced. It is vital to realize however, that no matter how meticulously this is done, no part of the ground will ever acquire the depth and vitality of the figures it connects. The figures have content – they are "true in themselves". The ground is derived from them – a complex, but purely formal network of argument and interpretation connecting a few fixed points.

Gestalt

Kuhn remarks on the similarity between scientific inquiry and Gestalt experiments (1962, p.114). A classic example of such an experiment is the drawing of two facial profiles turned towards each other – which may also be seen as the silhouette of a vase (Figure 6). The two views are mutually exclusive, and the transition between them abrupt and total. In one case the vase is figure, the faces continuous ground, in the other case the situation is reversed. What you see depends on a paradigmatic choice of what the figure is. Paradigmatic change in science leads, according to Kuhn, to a similar abrupt and total conceptual revolution.

"Re-grounding" a list of figures is a tortuous exercise of this kind. We tentatively collect the figures into small groups, only to discover that each group, though internally consistent, contradicts the others – threatens to define them as its ground – thereby silencing their voices, reducing their content to form. We flit to and fro between various parts of the total image, defining each part in turn as figure, all else as ground. Only gradually do we work out the contradictions and find a formula for the whole – a ground which ties the entire ensemble into a single, over-arching "meta-figure".

An amusing example of this process ocurred during the graduate course referred to above, when my students were told to reflect on the nature of anthropological writing. They were given an obviously mediocre essay and asked to suggest how it might be improved. Afterwards they were supplied with two quotations and requested to prepare an outline for the new essay, which rearranged the points raised in the original and included the two quotes as well. I had chosen the quotes on the basis of a vague intuition that they had some kind of relevance to the subject being discussed, but the connection was neither direct nor obvious. The result was unexpected. The initial critique of the essay did not change its content in any fundamental way. But when the quotes were inserted, each work-group arrived at a radically different concept for a new article. In each case the location of the quotes was different, and this seemed to be the crucial factor determining the differences between the solutions.

It strikes me that the students were trying to fit these quotes into the original essay as figures onto a ground. They let the figures have their say, instead of grounding them at once. But since the exercise did not last long, each group discovered only a small part of the figures' "story" and thus developed highly inconsistent "re-groundings" – which in turn gave entirely different meanings to the essay as a whole. In a more thorough piece of work all of the versions might have been combined, and the process of "re-grounding" would have become exceedingly complex.

"Re-grounding", as I have described it here, is reminiscent of Mead's model of how the Self is aggregated out of multiple contrasts to Others. But what is rarely remarked on in that example, is that the ideal unity is never quite achieved. Not all Others are integrated into the Generalized Other – and the child therefore always excludes from its Self some part of the separateness it experiences. The same is true of the intellectual endeavors we have been discussing. Even the most thorough "re-grounding" is never completely consistent, nor can it avoid reducing essential figures to the truncated status of ground. This is why periodical "shake-ups" are needed, to "de-ground" the entire structure again and start from scratch. This, I believe, is what Gestalt therapy tries to attain in a therapeutic context. In science it is what happens in Kuhn's paradigmatic revolutions. In both cases, the "figure-ground" relationship is changed. The scientist suddenly sees that part of what he thought of as ground is in fact a hidden figure. The patient in therapy sees that part of what he has denied or suppressed is in fact himself. But the goal of Gestalt therapy is not ultimately to assert a new figure-ground relationship, but to deny that relationship altogether. In the experiment with the faces and the vase the point is not to invert the picture, but to realize that what you see is neither vase nor faces – only a crumpled boundary line out of which many alternative interpretations may grow. Such relativization may be a precondition for all true innovation and learning.

Modalities of Intention

What emerges from this argument are three fundamentally different attitudes to scientific research – and in a wider sense, to human activity as a such. Since these attitudes derive from an intentional (rather than a dialectic) viewpoint, I shall refer to them as modalities of intention: The Figure mode sees acts, objects and persons as having inherent and imperative meaning in themselves. This meaning is projected into a world which includes other figures – onto a "list". In the Ground mode, the figure surrounds itself with a formal matrix, through which its imperative meaning may be expressed and connected to that of other figures. In the Gestalt mode, the relationship between figure and ground is relativized, letting hidden figures emerge from the ground, and permitting learning and change to take place.

Note the similarity between this model and that of Berger and Luckmann. As in their model of "social action", the three modalities represent stages in a single, cyclical process – exiting from a subject (figure, externalization), acquiring autonomous dominion in the outside world (ground, objectivization), and returning to the subject as given fact (Gestalt, internalization). But these concepts have greater depth and more fundamental implications:

Our argument so far has shown that the three modalities are not merely a convenient formal categorization of different aspects of the act. Rather, the cycle is a continuous process of re-adjustment and compromise – a process which can never be brought to a close because it cannot succeed. This last statement is not merely a polite reservation, it reflects the fundamental nature of the concepts under discussion. To grasp this, we must again contrast these intentional concepts with conventional, dialectic ones. A simple example might be the idea (mentioned above) that ethnic groups acquire their identity from the difference, the boundary between them (Barth 1969b). This boundary is not a water-tight barrier, but a "rule" regulating and restricting interaction between the groups in specific ways. To be real, the "rule" which is the boundary must be obeyed in practice – i.e. without interaction across it the difference between the groups disappears (becomes irrelevant). Thus the boundary defines a continuous, circular process – just as the three modalities of intention, it would seem. In fact, however, the differences are fundamental. The boundary concept "implies action" in the purely logical sense that it cannot be conceived of without it. Ethnic groups which do not interact have no ethnic identity, in the sense we are here discussing. They are separate, but not different. The modalities imply action in another sense altogether. They are not formal attributes of action – like boundaries – but practical ways in which action may be organized. We may – and must – choose between them in any given case, and our choices are never ideal, either-or alternatives – always compromises. We must keep on choosing for the simple reason that no single, ideal, logical choice can ever be made. Our scientific arguments are never consistent. The Self we build is never complete. We stalk success – but it eludes us, leaving us with partial answers that are dissolved by the new questions they generate. The boundary is either "on" or "off" – either there or not. The modalities are always there, constantly modifying each other, because they are different ways of coping with the same continuous reality, the underlying content of life – which cannot be turned "off". They are not concepts of formal logic but of morality and intention. In this perspective an ethnic group is a figure – it has identity as a matter of course, whether or not it interacts with other groups. The problem is not how it got it, but what to do about it. This is a practical and personal problem – because it has no solution. In a perfect world there would be no need for the modalities. They would neutralize each other and cease to exist. The boundary hand becomes more clear-cut and real the more idealized the world is: it is a perpetuum mobile driven by logic. The cycle of modalities is driven by moral and practical necessity.

By now it will have emerged why I consider it important to label these concepts intentional rather than dialectic. They are modes of action and organization, and the choice between them can mean life or death, truth or falsehood, beauty or distortion, health or illness, reality or fantasy. We shall shortly see how they may be expanded into modes of social organization, but to bring out some of their vitality I shall first give the reader a glimpse of the underlying moral attitudes they imply:

"What exhausts a motive is going beyond what the motive prompts us to do. Therefore the proportion of energy put at God's service will increase in the soul if great care is taken never to go beyond what one feels almost irresistibly driven to by obedience." (Weil 1970, p.150)

The figure mode projects the content of an idea into the world. An individual might call this self-expression or self-realization. The corresponding moral imperative must be that the total content of the Self (its figure) be expressed – in a balanced statement in which every part of the Self finds its place. Self-expression thus implies a form of truthfulness, an obedience to the inner laws of the self, a subjection to the limits and demands which they pose.

"But we must note the first strange fact about this terrible May evening. There was not a living soul to be seen – neither by the booth itself, nor on the tree-lined path along Little Bronnaya Street. After scorching the city all day long, the sun had finally gone off to sulk in a dry haze beyond the Sadovoe Ring. You'd think people would be gasping for a breath of fresh air – but no-one came to walk in the shade under the linden trees or sit on the benches. The park was empty." (Bulgakov 1929-40, p.423-24; my translation from the Russian.)

The ground mode separates figure from ground and at the same time subjects the ground to the figure's inner laws. This is a mode of authority. It is "objective" in the sense that the individual now stands as a thing, an object among others, and must determine the limits beyond which they cannot pass and the logic which they must follow. Likewise, he must respect their limits. The moral demand placed on the actor is that of consistency, the ability to accept separation and abide by it.

"An ex-Marine Texan, removed to Oakland, California, lamented to me in the mid-1960s about the anarchy of northern cities compared with the southern towns of his boyhood. When I mentioned that the South still had high homicide rates, and that Texan boom towns used to have very high rates, he said, 'Yes but it was a gutsy kind of killing.'" (Wilkinson 1984, p.99.)

The Gestalt mode annuls the distinction between figure and ground, relativizing all given categories, leaving only a matrix of distinctions out of which the new can grow. This is a mode of receptivity, perception, subtlety – in which the potential is sought rather than the real. It implies the freedom of a child or novice – moveable, but independent. For it is also a mode of interchange and communication. In order to receive, one must give – one must trust in order to doubt.

From the ethical and literary we return to the mundane realm of social science. In Legitimationsprobleme im Spätkapitalismus (1973) Jürgen Habermas enters one of the classical debates on the border-line between Marxism and anthropology – and extricates himself from it with only partial success. The problem is that of stages of social evolution. His initial list comprises a total of seven (or perhaps nine?) stages starting with the "pre-highcultural" and ending with the "postmodern". This list he boils down to four stages – whereof the fourth exists only in the hypothetical future; so we are left with three stages which he proceeds to discuss: "pre-highcultural" (or classless), "traditional" (pre-capitalist class society) and "capitalist" (pp.34-41). Each stage is dominated by a specific type of institutional order. Classless societies are based on kinship, pre-capitalist class societies on centralized political power, and capitalist societies on self-regulating, non-political economic mechanisms. I shall refer to the three stages as kinship, political and economic societies.

Once again the simplicity of the scheme is deceptive. This is brought out if we contemplate the political stage. It comprises such a wide range of social types that we would hardly think it possible to group them together – from small-scale, unstable African and Polynesian chiefdoms, through the "hydraulic" states of ancient Egypt and Peru, on through the slave-states of the classical Mediterranean, through the entire history of China and India, the Ottoman Empire, and European feudalism, along with Western absolutism of the 17th, 18th and 19th centuries. All of this is presumably one "evolutionary stage".

The idea is very general – and this is what makes it interesting. Since the birth of Marxism, various schools have quarrelled about the status of the examples listed above. Some have seen European feudalism as the prototype of all the others and forced glaring empirical differences into the background. Others have made increasingly subtle differentiations in European history, while lumping the entire rest of world history into a so-called "Asiatic mode of production". The historian Perry Anderson has cut through all this very nicely, by insisting that the historical differences be taken into account and defining "feudalism" as a narrowly European concept (1974 a and b).

What Habermas is doing is not to confound the issue again, but to address if from another angle. If we accept the historical differences once and for all, we may take another step and define a general category for all pre-capitalist class societies. Stating the criteria which separate these cultures – as a group – from classless societies on the one hand, capitalist systems on the other, gives us a clearer view of all three stages of evolution. This, I believe, is Habermas' intention – or at least it is the spirit in which I shall discuss him.

What is the virtue of separating human history into three such general epochs? We already have a voluminous literature discussing the evolution of social systems along a bi-polar continuum, from "primitive" to "modern" forms. Do we gain anything by inserting a midway point and labelling the sections of the line "stages"? I think we do – and to explain why, I must refer back to the preceding section of this essay. Clearly, "stages" are idealized abstractions, just as the primitive-modern bipolarity. But they are abstractions of a different order and influence our thinking in different ways. The bi-polar continuum is nothing but a distinction, a purely formal boundary – like the ethnic boundary I discussed above. It states an opposition. The primitive is the opposite of the modern – and since we are modern, the primitive is what we are not. If we are creatures of culture, the primitive belongs to nature. If we are impersonal, he is personal. If we have states, classes, institutional diversity, contractual relationships, nuclear families – then the primitive does not. Obviously, some of these contrasts are legitimate, but the point is that they are derived from logic, from the form in which the problem is stated, not from its content. Habermas himself falls victim to this malady in his definition of "pre-highcultural" society. His twenty-line description contains a total of seven major negations and only a couple of positive statements.

Introducing a third point on the line changes nothing in itself. But if the consequences of this move are thoroughly explored – if each "stage", each "figure" is allowed to speak – we are forced to think differently about the whole problem. If we have three stages there must clearly be some kind of continuity between them. On the other hand they must also differ – and not only in a formal sense. They must be thought of as concrete, organizational solutions to central problems of social organization – and the only way we can understand these solutions is by examining real, historical societies which used them. So we are back to the old discussion of content versus form – and this time we shall discover that it has far-reaching implications for general anthropological theory.

As I have implied, I find Habermas' argument spotty and at times misleading. His general point however, is clear enough. What defines each stage is not a specific type of human activity, but a certain, characteristic kind of institutional diversification. In kinship societies all social functions are interrelated within a single institution; in political societies a separate institutional sphere controlling access to power is established; economic societies separate a third institutional complex, specializing in production and exchange. When this idea is integrated into a wider theoretical framework its potential becomes clear. We might, for example, connect it to the model proposed by "substantivist" economic anthropologists such as Karl Polanyi (1957) and Marshall Sahlins (1965). These authors also speak of three stages of social evolution, in which different modes of economic circulation predominate – reciprocity, redistribution and market exchange. Generalized reciprocity is non-monetary exchange, in which prestations are not expected to balance. One may perhaps entertain a vague idea that balance is maintained "in the long run", but if not, that is considered favorable for the donor, not the receiver. To be generous is to acquire prestige – and this is sufficient motivation to keep circulation going. Redistribution is initially based on the same rationale, but the giver has now attained a permanent position to which the produce of the population is collected, and from which it is later "given" out. The center controls both collection and redistribution – and to secure this privileged position it must have means of coercion at its exclusive disposal. In market exchange prestige and coercion persist as motivating factors, but are overridden by the profit-motive. This reflects the indirect and mediated character of market exchange. One does not barter product for product, but reduces each object to a common value standard, expressed in money. The value of money is not dictated by central authority – it is an epiphenomenon of the exchange system itself, in a sense it is a measure of the state of that system as a whole.

The underlying rationale of these three stages is consistent with those proposed by Habermas. The prestige-motive can dominate only in a small group, where all members are dependent on each other, where there is no central authority, and where the collective can sanction miscreants effectively by simply ignoring (i.e. excluding) them. Trust, loyalty and solidarity are obvious necessities in such groups (which does not, of course, imply that they are all-powerful and omnipresent) – and the "mono-institution" of kinship is uniquely equipped to maintain these qualities. It is a comprehensive and dominating matrix, which organizes all types of activity – religious, economic, political, educational etc. Furthermore, within such a "mono-institution" all important human relations acquire a similar comprehensive, multiplex quality. Anything you do has obvious repercussions for everyone around you, and for every field of life. The kinship system is a densely woven net of interconnections, and within this net it is clear that prestige – i.e. the approval and support of the collective – must be the dominant motivation for economic action.

Similarly, the coercive logic of redistribution must dominate political society. It is true that limited forms of centralization are achieved in purely reciprocal economies, when some individual is in a position to continue giving more than anyone else over a long period. But true redistribution demands that the position of the center be secured independently of individuals, i.e. it must be underpinned by armed force. In its developed forms redistribution is associated with complex bureaucracies, professional armies, police-forces and judicial systems. But even the simplest political society must break the institutional monopoly of kinship at this one point. It must insulate an institution regulating access to power from the sphere dominated by the collective sanctions of kinship. Underlying redistribution is thus a strong institutional duality: On the one hand these societies are segmented into ranked status-groups (castes, estates) – whose internal relationships may be similar to those of kinship society. But on the other hand the status-groups are linked in an over-arching hierarchy, and their external relationships are thus of a fundamentally different character, narrowly defined by the imperative need for controlling access to power.

Money, as we have seen, is the motivating force of market exchange. But it also has a wider function. It dictates prices, and thereby regulates the dynamics of the entire exchange system. This eliminates the need for political centralization (other than as a means of curtailing the power of money) – the system is controlled by its own internal logic. The dualism of political systems is in a sense even more marked here. On the one hand an impersonal and abstract general order is imposed – so distant that we speak of it in mechanistic and mathematical metaphors rather than as a product of human action. On the other hand economic motivation is no longer dependent on loyalty to a group – the individual seeks his personal gratification. But this is actually a step beyond dualism. For what is opposed is the self-regulating aggregate of exchange on the one hand, the productive power of individuals on the other: economic societies juxtapose exchange and production. These are not separate entitities, but form a single, purely economic, institution, which has been split off as a whole from the previously dominating fabric of kinship and politics.

Several points can be made on this basis. First, the three consecutive stages strike a nice balance between continuous and discontinuous evolution. On the one hand we see continuity. Reciprocity in its pure form permits only situational hierarchy. More developed kinship systems become more specific in their prescriptions and limit the flow of certain goods and services to a restricted range of kin relations, thus supplying a system of circulation which may be utilized to form more durable (but still individually centered) hierarchies. An example is the Trobriand big-man, who manipulates rules of marriage and descent to build up a firmer, but still not permanently institutionalized, position. We may imagine such rules crystallizing gradually until a firm centralized bureaucracy is formed. This increases in strength and complexity from simple to more developed forms of redistribution, until at last the controlling power of formal bureaucracy is transferred to the impersonal medium of money. We thus find a continuous evolutionary movement along a simple bi-polar axis.

But although this description has its own merits, it loses sight of the main issue raised above. The stages are qualitatively different, and utilize unique methods of social organization: Reciprocity without kinship is unthinkable; true redistribution is impossible without an army of some form; and a market economy presupposes generalized money. The description by stages directs our attention at these concrete, organizational methods which make evolution possible. It emphasizes the content of change, rather than its form.

But kinship, armies and money are only rather obvious examples of the methods of organization prevalent at each stage. To understand the underlying organizational principles at work, we must now dig deeper into the moral climate and rationality which each stage represents.

I have emphasized the need for trust in kinship societies, but this is only a symptom of a deeper need. One way of approaching this is by stating an obvious fact – societies of this kind lack generally available methods of recording knowledge. This is clearly true in the sense that they lack writing, but perhaps – as Maurice Bloch (1975b) points out – it is still more important that they have few permanent edifices of architecture or engineering which might embody their past in an "objective" form. All the complexities of social organization must therefore be memorized. The mono-institutional framework of kinship greatly simplifies this problem – you need to remember only one system, not many. This solution is symptomatic: In such societies the entire mechanism of social reproduction and differentiation is traditional – one must be able to teach and learn every part of it. Society is therefore dependent on control and organization of everything that can be learned, of skills, knowledge, classifications, symbols, myths. These are societies based on meaning. No action is possible if you don't know what it means – where it (and you) fit into the matrix. If the mnemonic myths break down, society falls apart.

This brings us very close to our definition of the Gestalt mode of social action – in which, as we remember, receptivity, trust and the ability to learn were prime characteristics. An example may help clarify this point. The Tsembaga Maring – a group of swidden horticulturalists on New Guinea who have been described by Roy Rappaport (1968) – have a world-view which differs rather fundamentally from our own. They live from gardening and pigs – and because of the scarcity of animal protein in the area the pigs acquire a very special place in the delicate balance they maintain between their own group, the surrounding ecosystem, and the tribes around them. The pigs are the focal point of a complex ritual cycle stretching over a generation or more, which regulates the equilibrium between all these factors: When enough pigs can be slaughtered the ancestral spirits are appeased, there can be alliances and war – but the preconditions for this are complex and involve almost every aspect of Tsembaga life. All-important among these are the division of labor (predominantly between men and women), the relationship to the land (mainly to the "high" and the "low" ground), and the relationship to the spirits (which are red, hot and dry, or black, wet and cold). Tsembaga ideas of this balance are not formulated in terms of proteins, population size, carrying capacity, labor input or political relations, but in the language of spirits. The spirits regulate the ritual cycle. They cause happiness or misfortune. They govern human relations and the land. This may seem a rather risky way of keeping things going – but on closer inspection one realizes that the mythical world is in fact a language comprising all other aspects of life. A red spirit is a spirit of the high ground, of men, of war; a black spirit is of the low ground, it is female and governs gardens and fertility. The complexities are endless. Even on simple agricultural tasks the combination of sex, ecology and color puts its explicit mark. Now the point is not that the Tsembaga "mean" ecology or war when they "say" spirits, but rather that the spirits are their way of thinking about everything from pigs, war and rituals, to gardening tools and love. They are concepts which bridge all important realms of Tsembaga life and are as such aesthetically, morally, humorously and practically satisfactory. This is Gestalt-type language. It is eminently flexible, permitting suppressed figures to surface out of the ground at any time. It is exceptionally learnable, since its terms are simple – though its application is not. It presupposes trust, because full participation in society is impossible without agreeing on its basic terms. Moreover, it is this organizational principle – a structure of pure meaning – which governs society, and does so without recourse to central power or money.

Political societies on the other hand, are organized in a hierarchy of status-groups, bound together internally by solidarity, externally by the dominance of centralized power. What is perhaps most striking about such societies is a consistent dualism, an insistence on keeping their internal and external aspects separate. This is the function of the theatrical rituals of feudal power which Michel Foucault (1975) has described so well – in which the ruler demonstrates his power on the market place by public torture of criminals. In less powerful redistributive systems the same trait is evident in the conspicuous consumption and demonstrative generosity of petty chiefs and big-men. The function of this theatrical ethos is to separate the stage from the public – the inner world of a status-group from its outer relations to the hierarchy of power – personal life from performance. Berger et al (1973) describe the ethics of feudal Europe in these terms. Belonging to a status-group is like wearing a mask. This mask is the honor of your group – to which you must conform in all your dealings with the outside world or discredit the basis on which your security rests. The knight errant wears such a mask – he rescues maidens, goes on quests, fights dragons, and participates in tournaments – all this sounds like stereotype or caricature to us (as it did already to Cervantes), but our words are inadequate for describing what his mask means to the knight. His stereotype is not a symptom of bigotry, but of his ability to maintain separation. It is a measure of his consistency.

So again we are back to terminology we know. Consistency, authority and separation were the moral imperatives of the ground mode of social action. In political societies, where this mode is the dominant organizing principle, they pervade all life. An interesting case in point is religion. It seems striking that the great world religions – which all originated and thrived in societies of the political stage – tend towards a clearly dualistic world-view. There is the Manichaean tradition – running from the ancient Middle East through Judaism, Christianity and Islam – with its oppositions of Good and Evil. There is the subtler Chinese tradition, with its philosophical grounding in the interchange of Yin and Yang and its practical division into complementary life-ways – the worldly way of Confucius and the spiritual way of Lao-Tse (evidently, Chinese officials would be Confucians while in service and turn to Taoism on retirement). And there is the still more complex Hindu world, with its thronged Pantheon – but each god has dual aspects, and in everyday morality all social relations are regulated by the opposition of pure and impure. These descriptions are clearly rudimentary, but it seems possible that the great religions are ground-oriented in this very general sense. Perhaps this also explains the problems they face surviving in anything like their original form in the modern world, where consistency and separation are not valued to such a degree.

In economic societies the great dualities no longer survive. The centralized locus of power has transcended humanity and been incarnated in an abstract principle. God has risen on high – and all we can see of Him are the rational, mechanistic laws He has left us with. If there is a God today, He is worshipped in the temple of Science. "Life is hard. Then you die." "Birth, copulation and death – that's all." The catechism of modernity, according to American folklore and T.S. Eliot respectively. The governing power of the economy has split the world into an impersonal and a personal sphere, but this is no longer dualism in the sense described above. One of the most perceptive analyses of modern life is given by Michel Foucault in Surveiller et punir (1975). The book describes the transition from the theatrical modes of punishment referred to above, to modern methods where discipline is the key-word. Central to the new society Foucault describes is the idea of the "institution" – the prison, school, barrack, hospital, asylum. Each of these, with different methods, seeks to attain the same goal. Human thought and action are portioned out into segments – one is taught how to control these segments and assemble them in a rational and productive way. The examination is a prototype of this rationale. You learn to "master" a field of knowledge and your mastery is tested under laboratory conditions. The test is confined in time and/or space, it focuses on specific questions, and – above all – a specific technique is required to succeed. The test is typical in its dual orientation. On the one hand you are trained thoroughly, and (ideally) in exact conformity to your personal capacities and requirements. On the other hand it is demanded that you express your knowledge in accordance with general, impersonal laws. It is further supposed that these laws have an objective status, so your success or failure can be measured. Law on the one hand. Personal application on the other.

There is an almost eerie correspondence between Foucault's description of economic dominance and the morality of the figure mode described above. In both, the central point is a paradox – self-expression, but within limits which are self-assumed – "... a regulated economic life with the economic impulse functioning within bounds", as Max Weber put it (1923, p.445). Self-discipline and obedience to the "inner voice". Responsibility and conscience. These are modern concepts, and it is not strange that the language of psychoanalysis was developed in our time rather than in an age of honor.

This last point directs our attention to one of the most important paradoxes of this whole presentation. We have treated the modalities of social action first as general categories, then as moral imperatives. Later we have shown that historical forms of social organization may be elucidated by closely related terms. But social organization is related to power and control – categories which are in principle amoral. Furthermore they are limited in their relevance to specific times and places. Morality, on the other hand, must have value-content, and must – at least in some sense – claim universal relevance. How can one then, as I have above, speak of the moral and amoral, on the one hand, and the general and the specific, on the other – in the same conceptual language?

This is only part of a more general problem. Each historical stage we have described is an aggregate of all three modalities of action; hence the isomorphy I have posited between modality and stage must not be understood literally, but should rather be taken to imply e.g. that one of the modalities is dominant. We might operationalize this in terms of ideological hegemony, causal primacy, existential imperativity, symbolic centrality etc. – in each case greater intrinsic meaningful "weight" is ascribed to certain phenomena than to others, an absolutism is defended, an essentialization proclaimed, a relational symmetry denied. But this must not be taken to imply an exclusive, binary scheme. Meaningful "weight" is unevenly distributed throughout reality, and "dominance" is a matter of statistical frequencies and relative intensities, rather than of bounded locations and plateaux. In each historical stage, all modalities are active, indeed, the "dominant" modality may be occur only sporadically. Entire institutional complexes may be "sub-dominated" by non-dominant modalities, throughout most of their "normal" cycle of existence, and only intercept the dominant structures in deviant cases.

The three modalities of action are thus "elementary particles" similar to the binary oppositions proposed by Lévi-Strauss, out of which socio-cultural aggregates (stages, institutions, roles, symbols) may be assembled. All action (and all structures built out of action) may be thought of as a particular "blend" or compromise between the three modal forms. Fundamentally, therfore, the modalities must be value-neutral and descriptive. When I associate them with morality, it is a practical morality I have in mind. Given an ideal – how can we attain it? Given for example, the problem of how to live life, of how to organize society, or of how to do fieldwork – one is faced with a series of 'tactical', i.e. short-term and highly unpredictable choices. The cycle of modalities seeks to illustrate three basic types of tactics one might utilize when making "crucial" (i.e. 'qualitative') choices. These are general – but not abstract – terms, because choice between them is real: obedience, consistency and receptivity are not formal mirror images, but different modes of life