Boundaries and connections Fredrik Barth The concept of boundaries is important and versatile, but often unclear and even quite mystifying in contemporary anthropological thought. In the following, I wish to raise the twin issues of how we think when we use a notion of boundaries, and to what extent it provides a figure of thought, a concept, that is used generally by people to perform mental operations and construct categories. Since I am associated with an influential formulation thirty years ago on the theme of 'ethnic groups and boundaries' (Barth 1969), I should signal clearly at the outset that what I take up here is quite a different set of issues from those that then concerned me — though I will, towards the end, have something to say on the question of ethnic boundaries and their variable salience. But my general purpose in this essay is to explore some cognitive aspects of the concept of boundaries: both its variable uses by actors in different social systems, and its uses and abuses for our own analytical purposes. This raises a number of themes. Let me start with the English word and concept, 'boundary'. We all have a sense, I think, that it embraces three levels of abstraction: 1 literally, boundaries divide territories 'on the ground'; 2 more abstractly, they set limits that mark social groups off from each other; 3 and finally, they provide a template for that which separates distinct categories of the mind. Boundaries and distinctions My first step in the present analysis is to unhitch the idea of boundary from the idea of categorical distinction. I claim that the two are not necessarily connected: making a distinction does not necessarily entail drawing a boundary. Let me be concrete and evoke a distinction that figures large in the experience of Scots and Norwegians: sea, as opposed to land. Our environment is composed on the one hand of firm ground, and on the other, the vast, restless, open sea - they are as different as day and night. But that difference does dot depend for us on drawing a boundary between them. In fact, for the very figure of speech that we use to epitomize a categorical distinction - as different as day and night - we use the patently continuous variation of light through a twenty-four hour cycle as our image. Likewise land and sea: should we, for example, draw a boundary line at high tide or at low tide to make the distinction? This matters only if we are making a representation of the distribution of sea and land in the form of a map - not to the distinction that we make in out minds. And we handle without difficulty that much broader intermediate zone: from the outermost rock, perhaps with a lighthouse, dominated by the surrounding sea to the innermost cove or harbour, dominated by the land. The gradient of sea-ness and land-ness does not in the least confuse our categorical, mental distinction between the two realms. Yet we imagine that we must be drawing boundaries. We are likewise inclined to use the image of boundaries to refer to the divisions that separate distinct social groups, especially when these are associated with territories. But let me use ethnographical materials that I am intimately familiar with to explore some other cultural imagery that serves to distinguish and differentiate territories and groups. I begin with the Baktaman (Barth 1975; 1987) who, when I first knew them in 1968, were a recently contacted population of 185 people in the rainforests of interior New Guinea. How did they construe their identity, their shared sameness and their distinction from others? Certainly with reference to place - Baktaman means literally 'People of the Bak river valley' - but without drawing territorial boundaries. Contacted by an Australian patrol only four years earlier, they had till then lived in a state of insecurity and often warfare with all surrounding people: so their world was dominated by a gradient of danger towards all neighbouring groups, not by delineated boundaries. Nobody claimed the zone of what ir» our imagery we might call no-man's-land that lay between these groups. The Baktaman made their shifting taro gardens compactly in a core area, Seeking security in numbers. The choice of locality for a new garden was influenced by push and pull factors: fully regenerated soils were preferred, but they tended to be further from the settlements, which was less safe and more inconvenient. Also on the smaller scale within gardens, boundaries seemed absent: areas of forest were cleared collectively by people who chose to team up, until the cleared area looked like it would be 'enough'. Within that cleared area, individual fields were not demarcated: planting of taro was started by each person in a place favoured by that person, and spread progressively until the cleared area was filled up, or until someone started harvesting, when all planting was discontinued. Gardens were differentiated from forest by distinct category words and distinct taboos - but not >y drawing boundaries, fences were sometimes built along the perimeter - but«tbese were barriers td discourage wild pigs from destroying the planted crops, not boundary markers. Place was immensely important in numerous ways, and was associated with social identities, but territory and social group simply did not seem to be conceptualized or distinguished by boundaries-Compare this to the Basseri, a group of Persian nomads with whom I once worked (Barth 1961). Do territorial boundaries play a salient role among them? Basseri are part of a larger society where plenty of attention is given to boundaries and boundary markers on land. But for the understanding and analysis of their concepts and cognition, the crucial question is how Basseri themselves know and experience their world. And all the evidence indicates that their salient nomadic experience makes territory the scene of movement, not a field for the demarcation of plots. Migrating caravans, and grazing herds, pass over the land. In the afternoon, tents were pitched in a camp, and sometimes corrals of thorn were made to keep predators out and animals in at night - but fhey did not appeal- to embody the idea of a boundary. How about people's attachment to place? When we sometimes happened to stop and camp at old campsites from previous years, they seemed to evoke something that I could best interpret as nostalgia, as people pc>ked around looking for the rock beds on which they had once placed their waterskins; but no local rights of precedence seemed to hold over from their previous occupation. From a sedentary person's point of view, indeed, their world seemed scattered and disordered, precisely because it appeared unbounded. But it was not disordered: groups hold elaborate and clearly defined grazing rights. However, these grazing rights were conceptualized not as bounded territories, but as migration schedules, called il-rah i.e. tribal roads. Each such 'road' was composed of rights of pasture and of passage during particular time periods. I have compared these rights to a train schedule: a train does not have rights to railway lines and stations, but the 'right' to be at certain points at certain times (Barth I960; also 1961: 5). In other words, land and place, and exclusive claims to particular lands and places, can be conceptualized in several ways by means of quite dissimilar cultural images; and such specificity and diversity are lost if one uniformly presupposes a concept of boundaries. On the sources and uses of concepts The recognition of such cultural diversity in images and concepts leads to a second assertion: a boundary is a particular conceptual construct: that people sometimes impress on the world. So we can ask: when is that done; and what does it presuppose, in the way of images and experiences, for people to choose to draw boundaries? To ask this is to raise empirical questions about people's cognition; and to answer it we need some kind of explicit or implicit theory of human cognition. The issue is important: for a cultural analysis of the concept of boundaries, we would need somehow to demonstrate that the particular conceptual construct of a boundary is indeed being employed by a group of people; and it would add to our insight to be able to identify the wellsprings of the boundary concept: the sources of its meaning and salience to them- It is not enough that I have seen people act in certain situations as z/they had constructed what I would call a boundary. As I have shown above, people cafi employ quite a different form of concept - but one that in that particular situation produces apparently similar results. As an anthropologist, I should not be content to fashion a series of logical constructs that will produce a simulacrum of the pattern observed in people's actions: I vvish to lay bare the concepts that people are actually using, and the connections that people themselves make, when they perform such actions. 'Boundary' to a English speaker, as I have already suggested, is really a quite complex cultural model. It signifies a syndrome of ideas, ranging from an imagined line drawn on the ground, through various abstract separations and distinctions in realms of political and social organization, to a schema for conceptualizing the very idea of distinction. It thus carries massive cultural entailments- if people make use of other concepts or models, differently constructed, then they may open quite different possibilities of thought and action fof the people who use them, and will presumably configure experience quite differently for their users. As anthropologists, we need to emulate the steps of people's thought and reason — with some degree of faithfulness, so as to discover how their ideas are indeed interconnected and their categories constructed. The need for cognitive theory To do so we need a set of propositions about cognition, about 'how natives think' - needless to say, with the reflexivity to include ourselves and our own thought processes- But after the general rejection of Levy-Bruhl's (1910) theory of the primitive mentality, there has been little critical discussion of such issues in our discipline until very recently (cf. Shore 1996). This has allowed mainstream anthropologists to leave their quite extensive assumptions implicit, indiscriminately projecting common Western folk models of thought, supplemented with some norms of formal reasoning developed in Western philosophy, on all human thought processes. There is no reason to consider this adequate. Maurice Bloch has indeed warned us clearly and bluntly, that anthropologists have so far worked from 'a false theory of cognition' - one that represents thought as language-like and linear, and does not match the way human beings conceptualize anything fundamental and familiar in any society or culture (Bloch 1992: 127). We need to do better, by being more theoretically sophisticated and explicit, so we can test the various ideas available in the cognitive sciences and related fields on our ethnographic materials, and use those that are most fruitful, and faithful to our observations. It must be sensible to proceed tentatively, testing the different frameworks that are available on the materials that we have or can produce — so I have no wish to foreclose what should be a thorough and continuing theoretical search. But for present purposes I have found in the work of George Lakoff the tools I need to pursue my ethnographic intuitions about the ways Baktaman think about place and identity, how social groups are experienced by Basseri nomads, and how our own concept of boundaries may be generated. As my next step, it may therefore be useful to characterize briefly what I find most stimulating in his approach. To many anthropological readers this may be old hat; but it does not seem to have been a much worn hat. It strikes me that this argument must affect our more traditional views of structure and cognition, and should transform our way of doing anthropology. The image of group boundaries I believe that for us, in our European tradition, the particular image of boundaries derives its power ultimately from humankind's ability as a tool-user. Using a tool involves an extension of one's self. When you hold a knife or a spoon in your hand and use it as an implement, the experienced limit of your body is no longer the skin of your hand, but the cutting edge of your knife or the cup of your spoon. Likewise, when we learn to use skis or a bicycle, they become extensions of ourselves. Have you not perhaps had that odd shadow of a feeling, as you park your bike, of leaving behind a part of yourself? Take this one step further: when you extend that merging of self and object through time, and make it endure, you create the relationship of possession: of individual property. But to be able to appropriate land to yourself as property, you need to separate a piece of it, detach it as a physical object and thus as an object that you can appropriate and claim as against the world. It is this that requires boundaries: by imposing the conceptual construct of a boundary line around the land you disengage it from its surroundings, and can appropriate it to yourself. I know of course that there is a vast philosophical and legal literature on the subject of private property and land. But I am bypassing all that, searching for preconceptual and proto-conceptual wellsprings. I claim that the very powerful and deep-rooted tie of identity between, for example, a Norwegian farmer and his land - land as a piece of ground demarcated and defined by boundary markers as against the lands of his neighbours, creating boundaries that enclose a household and its means of livelihood - this image evokes the prototype of the meaning of boundaries for us, as it does for many agrarian peoples in large parts of the Old World. This elementary figure of a farmer and his family, on the land they possess, bounded and separated from adjoining territories, can readily be projected as a figure of homeland-and-country, with national boundaries demarcating it, and defining the European concept of nation. Metaphorically, we then can project its image to other, non-territorial groups as having 'social' boundaries. In British social anthropology, the same image is enshrined in our model of corporate groups, defined by their exclusive and excluding rights to their respective shared estates. Finally, in a further imaginative leap, we can use 'boundaries' as a metaphor for how abstract categories, natural classes and kinds, are separated and marked off from each other. Other images of social groups But other lives create other images and schemas. If we return to the Basseri nomads, territorial boundaries lose their saliency, and no analogue to boundary-drawing is necessary to envisage social groups. The camp community as a group becomes a directly experienced unit, shifting between its two modalities: as a migrating caravan and as a cluster of pitched tents. Physically it is manifest and distinct: not as something contained within an imagined circle drawn on the ground around the group, but directly manifest as against an ever-changing background of new places, no sedentary spectators left behind with every move — and Basseri camps move, on an average, about once every third day throughout the year. Thereby the group itself becomes its own symbol and template, not through the metaphor of a territorial boundary. As the herd of a nomad is not defined by the limits of its pastures, but moves as a body through the open landscape, so also the Basseri camp community is defined and perceivable directly from its movement against a stationary background. There is a particular constellation of geographical, ecological and political circumstances that produces this figure - I am not claiming that nomadism as such will always produce such an image schema for its practitioners. But in the Basseri setting it is salient and striking; and therefore, to identify the prototype, the experienced source for the figure of a distinct entity among the Basseri, we would point to this direct experience of a social group, not boundary lines drawn on the ground. Furthermore, this social group is created not by shared possessions and a boundary, but by social bonds, that underwrite the positive decision made in each tent to move together and camp together with each move. One contrary decision, and the camp fissions and the parts go their separate ways. Camps are held together only by these positive bonds, not by shared rights, or by the labour and investment in immovables that tie a farmer to his land and wedges and hedges him in among his neighbours. Likewise, larger groups of these nomads are created by the social and political bonds of submitting to the same chief, bonding to his imperium of authority and protection: a group-ness that arises from the realm of domination and power, not from being within a physically bounded region. Linguistically, derivatives of the Persian word jam, 'coming together', serve to conceptualize social groups. I find it very interesting and challenging that we should be able to uncover such different and contrasting experienced sources for the defining templates of united social collectivities. Among the Baktaman, I find it difficult to identify any compelling experienced template for either territorial or social boundaries. I found striking their continual struggles to create an experienced group identity among members of an initiation cohort age set (Barth 1975). They try to do it by separating first-degree novices, suddenly one night, from mothers and siblings, physically aggregating them in a minute leaf hut out in the jungle; and then over the years empharically performing the same operations simultaneously and collectively on all, thus creating a long sequence of shared critical experiences. Once, uniquely during third-degree initiation, they ritually encircle the group of third-degree novices with a black scring/rattan — but the rattan is not a token or image of a physical boundary: it is identical to the rattan that holds together the vertical poles that form the walls of a house. Often, I was struck by how groups of men on longer treks showed a marked resistance, almost like fear, of separating and not returning to the village together at the end of their trek as they had left together. I sensed repeatedly that the achievement of conceptualizing and creating a discrete collectivity of people, i.e. a social group, was something elusive, obscure and opaque to Baktaman — also in the case of cult groups and descent groups, where the difficulties are compounded by the secrecy that always accompanied sacred or important acts and concepts. I could go on describing these absences: the absence of co-residence as a feature of family and household, for husband and wife and adolescents always resided separately. The absence of household commonality: joint producers would receive shares of cooked food from one hearth but carry it away with them and consume it, or exchange it, in same-status gatherings, except in the case of mothers and small children. The absence of truly exclusive private property rights. But I should rather make these observations meaningful by turning them into positive descriprions of what is there and provides templares for Baktaman people: married couples and rheir domestic cooperation are constituted by the positive bonds of teciprocity, prototypically in the exchange of cooked taro for wild game - though many food exchanges are non-exclusive, since prestations are dispersed by the effects of taboos that forbid persons of particular status from eating particular foods. In fact, groups among the Baktaman all seem to be formed by bonds, not boundaries or possessions — whereas social categories among them are defined by taboos, i.e. shared abstentions, and again not a boundary image. This being the case, it seems misleading to impott the particular cluster of images and assumptions that we associate with a concept of boundaries to describe the outcomes of group and identity formation among Baktaman, if no such boundary imagery has been involved in the processes that generate these groups and identities. Methodology My present concern, then, is to identify procedures whereby we, as critically and precisely as possible, can discover the cognitive operations and imagery that people use to conceptualize their acts, social groups and environment. As anthropologists doing fieldwork, we wish to accumulate evidence of these cognitive processes, so as to understand more richly and deeply what is going on between people. Tim Ingold (1993) has written tecently about fieldwotk and how the field anthropologist learns by 'becoming immersed in joint action' with people. Thus he writes, I experience the components of this environment as they do, not because I have learned to construct them in my mind accotding to the same categorical conventions, but because I have learned to attend to them in the same way… Such communion of experience, the awareness of living in a common world, establishes a foundational level of sociality... that constitutes the relational baseline on which all attempts at verbal communication must subsequently build. Ingold here points to something very important; but the opposition he sets up – between his learning to attend to components of their environment, and learning to construct their categories in his own mind – may not be as absolute as he suggests, and can be transcended. His immersion in joint action during fieldwork surely builds a growing community of experience with his companions which establishes not only that foundational level of sociality to which he refers, but also some of the preconceptual, experiential bases that his companions use to construct their cognitive categories and pathways of thought and reason. If so, the way for the anthropologist, as for the native person, goes through perception and embodied experience towards the construction of similar, increasingly shared conceptual categories. Perhaps this dual benefit – of enhanced sociality and enhanced preconceptual sharing – is the full harvest of a serious pracrice of our famed 'participant observation'. What ethnographic evidence might we muster to test the intuitions, or supposed understandings, that we win by such participation? For us, slavishly to emulate the linguistic focus of much of the work in cognitive theory, seems to me too narrow for anthropological practice, besides relying too heavily on an exceptional command of local language-in-use which we rarely have. After all, exotic linguistic materials from other lifeworids represent only the tip of the iceberg of our anthropological data, one which leaves all our observed and enacted experience unmentioned. Ingold is right, in my judgement, in his emphasis that we learn above all by becoming immersed in joint action with people. How to deliver evidence for our success in that process is a familiar and eternally troubling issue in anthropology; but we should not try to solve it by shifting our attention to other data that ate easier to document, yet less pertinent to our insights. Indeed, too great an emphasis on documentation and testing may miss the point. We cannot expect out readers of ethnographies from unfamiliar places to have that sense; but we can use his theoretical model as a guide to identify analogous elements in other conceptual constructions: the better to know what to look for; to sharpen our perceptiveness for the image schemas that are available in a lifeworld; and to test our interpretation for coherence with the other practices and codifications found in the population. It is not merely that many human conceptual categories are fuzzy, and not satisfactorily described by the rules of formalist logic. Studies of the ethnographic diversity of these cognitive processes promise to help us chart the forms, and the limits, of cultural relativity. But if we accept such a dynamic model of cognition, then we are faced with a new and unfamiliar problem: that of explaining the relative stability of the categories and knowledge that we ofren observe within particular communities at particular historical moments. Anthropologisrs, on the contrary, have been used to assuming stasis, sharing and isomorphy as givens, and as necessary features of cognitive models. Stasis, because meaning supposedly arose from the correspondence of arbitrary sign vehicles with structures in the real world. Sharing, because these arbitrary conventions were the key to interpersonal communicability and comprehension. And isomorphy, in the sense that cultural representations were, within a narrow sec of logical transformations, identical with that which they depicted (more on this last point later). If, on the contrary, thought is inherently imaginative and dynamic, we must explain its degree of stability by other means. I would suggest that father than look to the logical necessities of the structuralists, we should turn to the motivating force of experience and local circumstances, and especially to the particular patterns of social relations, to provide the framework for our insights. A degree of cognitive stability in a population may be generated in several ways. We may find, sometimes, that people have no known alternative and are unable to think of a different concept. Often, howevet, we will find that there are obvious alternatives, but that there is systematic indoctrination whereby authoritative teachings are drummed in and elevated to dogma. Third, where tasks are distributed in a group and smooth social cooperation is vital to their performance (as in the navigational teams analysed so elegantly in Hutchins 1995), the need for swift and unambiguous communicability favours shared convention. More generally, where people are locked into a social organization of vested interests and mutual controls, there will be positive encouragement for cognitive assent and agreement with the others who share those interests, and sanctions will be brought to bear against its breach. Thus we need not assume rampant flux in all matters, even if stability is not a definitional necessity for concepts and categories. And giving more scope to imagination and variation in cognition allows us to acknowledge and describe many more features of what we observe among people. Specifically, it allows us to trace more, complex relationships between cognition, cultural representations and outcomes, and to analyse processes of change. When people draw boundaries Returning to our specific theme of boundaries, I propose to show some of this complexity by developing a further thesis, namely: impressing boundaties on the world creates affordances as well as limitations. Since modern bureaucratic states are particularly strong in their dependence on boundary concepts, they can provide apt empirical materials and food for thought. On the basic level of the concept, boundaries are assumed to separate what they distinguish. One kinesthetic image schema that we use for this idea of the separating effect of boundaries, is of movement outward like a fluid that spreads — up to a limiting boundary, where it stops (cf. Lakoff 1987: 284, 431, 456, for various metaphorical and abstract extensions of this image schema). So we think of boundaries as the edges of container schemas, as barriers. Yet we know of course that such a schema does nor always fit a complex world. Human activities perversely create such leakages through conceptual boundaries by reconnecting what has been separated. They arise above all from two sources: inventive behavioural responses to the imposition of boundaries, and the effects of social positioning. For a simple illustration of the former, reflect for a moment on the scene of two English neighbours, conversing over the garden fence. The territorial boundary of their properties separates them but it gives shape to their interaction in a way that I suspect positively enables it, since it frames and defines the nature of the opportunity. Thanks to that boundary, the conversation can proceed in a more carefree and relaxed way, and be elaborated and pursued with less risk of other entanglements – a consideration that may loom large in shaping the role performances of neighbours. But if we also introduce social positioning into our analysis of boundaries, a veritable Pandora's box of social and cognitive inventiveness is opened. Often one will find that while some do the boundary imposing, others look for loopholes. Thus separation may be what the national regime sees when imposing boundaries, but that is not how it looks to a potential smuggler: his or her pragmatic view of the outcome of national boundaries focuses on the disparities of price, value and availability that result on the two sides of the boundary, and the opportunities thereby generated. For many formerly pastoral nomads who are more strategically located than the Basseri, for example tribes in the Syrian desert, national boundaries have created a smugglers' niche of such magnitude that camels and sheep have paled in comparison (see for example Lancaster 1981). In fact, throughout history, political boundaries have been rich in affordances, offering opportunities for army careers, cusroms-dury collecting agencies, defence construction contracrs and all manner of work and enterprise. They have provided a facility of retreat and escape for bandits and freedom fighters eluding the control of states on both sides; and they are a constant field of opportunities for mediators, traders and middlepersons of all kinds. These affordances are in turn reassimilated into cultural models of frontiers, based on people's observed experience on what indeed goes on in such places; but these features seem to be treated by them as cognitively secondary, and do not feed back transformatively on the basic schematism of boundaries as separations, rather than conjunctions. And other implications proliferate. Most dramatic in the contemporary world is the vast growth in numbers of the social category of international refugees and asylum seekers, predicated on the crossing of the boundaries that others have drawn. Given the foundational premise for modern states – that political sovereignty is identified with bounded territories – crossing that imagined line on the ground attains a magical or miraculous significance, setting the escaping persons free from the oppression behind them. In a world where national boundaries provide the dominant image of the state as a polity, we need to remind ourselves rhat it does not have to be that way. Geertz (1980: esp. 63ff.) in his analysis of Southeast Asia's traditional Negara form of statehood in Bali, gives perhaps the best known anthropological description of a polity based not on land and territorial boundaries, but on personal ties of lordship and submission between leaders and subjects. Among the Basseri nomads discussed above, likewise, the persona! bond of submission to a chief was the basis for the formation of tribal polities. There were, of course, disgruntled subjects, oppression and fugitives also in such polities, but escape depended on submitting to another chief and being received by him, not on crossing a territorial boundary of jurisdiction. One consequence of this was the effective incorporation of refugees into a new body politic as they joined their new chief: their status was defined by the positive social bonds they formed with him. Not so with modern international refugees. Modern nation states, in their bureaucratic institutional structure, seem deeply wedded to boundaries and objectivist definitions, also in their treatment of refugees and asylum seekers. This emergent category is rigorously defined by a limited set of necessary and sufficient features, providing the criteria for entitlement to rights and benefits; they form a social category constructed on the metaphor of boundaries and without any reference to bonds and social relations. We are all too familiar with the properties of such a system from our public welfare bureaucracies: administrations that operate strictly in terms of bounded categories of beneficiaries with qualifying attributes, generating a social organization of clients by entitlement vs exclusion. It is a form of organization that enables the mass dispensing of individual benefits – a difficult effect to achieve; but it is obtained, as we know, at the cost of defining the members of a society without reference to their sociality, i.e. their social relations, and pursuing a chanty that ignores all needs thai are not defined as rights and do not fit the categories of entitlement. Ethnicity, on the other hand, is a prototypical realm for the non-bureaucratic drawing of boundaries to define social groups. People's own experience of a cultural contrast to members of other groups is schematized by drawing an ethnic boundary, imposing a false conceptual order on a field of much more broadly distributed cultural variation. The boundary schema, on the contrary, constructs an assumption of shared homogeneity within the group and cultural difference between groups, with great potential consequences for the social life of larger communities and regions. Such social boundaries often create affordances which give rise to a division of labour between ethnic groups; but the cognitive image of a social boundary is also deployed to limit the kinds of social relations that are allowed to form between members of the groups that are thus distinguished. In the former Yugoslavia, we have seen how objectively rather small cultural differences can become emblematic of membership in deeply opposed categories, and lead to cycles of enmity, violence and 'ethnic cleansing'. Nor are the 'affordances' of purely territorial boundaries necessarily benign. For example, there can be little doubt that the simple existence of a boundary between the Gaza Strip and the place of employment for many of its residents in Israel creates for the Israeli government the political option of closing those boundaries, as a means of asserting collective pressure on the Palestinian leadership and population - an option that is not practicable against Arabs who reside within Israel. Cognition, social structure and change These are examples of the connections across boundaries that were announced in my title. The crucial point is that most of them are built by processes other than those that define the boundary. To draw a boundary is a cognitive act that lays down some premises; but it does not determine all the social forms that eventuate. The affordances of a boundary set the scene for social activities, and in that sense, yes, boundaries also connect. But the connections that emerge are the work of people who respond selectively and pragmatically to the affordances, spinning connections in forms that will be shaped by social and material processes, not by cognitive fiat as the drawing of the boundary was. The presence of the boundary sets the other processes in motion — with emergent results. Since these results are brought about by people and will be experienced by people, they can also be cognitively re-appropriated by people - but they are not themselves the pristine product of the cognitive process: they become re-attached to people's concept of boundary as that 'rich mental imagery', in Lakoffs words, of which people are capable, but which they obtain from experience, not by cognitive derivation. We need to distinguish between the cognitive premises that construct the boundary - by what I might call acts of imposition - and the sociology of people living and acting around that boundary and thereby shaping an outcome. Those contingencies produce the effects from which people in turn reconceptualize boundaries (endowing the concept of boundary with what we used to call 'connotations') that derive from what actually happens along that particular boundary as a result of the connections that people spin by their actions and by the consequences of those actions. My argument, I believe, is in line with my quote from Maurice Bloch: that we must make use of the best available theory of cognition in our anthropological analysis. But it adds a further caveat: that we must also take care to use it only to explicate such materials as it is appropriate for. There is a cerrain glib practice in some contemporary anthropology that foreshortens and trivializes our account by depicting it all as if human lives were made up only of words, culture and cognitive representations, and not of the processes of social interaction, politics and economics — as if our cognitive maps singlehandedly created the terrains they depict. In this questionable practice, a couple of concepts and a couple of distinctions are thrown together, which in the mind of the investigator seem to depict a pattern that they have glimpsed in the material. This structure in turn provides them with the bases for a thumbnail sketch of some supposedly empirical siruation (about which their audience is relatively ignorant, and so must depend on the reporting anthropologist); but the sketch is so simple and partial that it begs all the imporrant questions and easily disguises its own false premises and assumptions. I wish to insist on the need for us to embed our models of cognitive processes firmly among the other processes that also unfold in the world where people live. Every event has its precipitate in experience; each move made by someone will affect rhe nature of the connections that arise along the boundary, and thereby affect the 'rich image' of the boundary that is retrieved. Through such interaction, social processes determine cognitive models, as much as cognitive models determine social processes — but with a degree of complexity in the connections so the two never become mirror images of each other. Let me try to grope my way one small step further. An important feature of Lakoffs perspective is that the elements he identifies as materials for cognitive models (basic-level structures and kinesthetic image schemas) do not take their meaning from their correspondence with objects in the world: they are precisely what he calls them: schemas, mentally derived from the routines of recurrenr experience. To use rhem, people musr fit them to the particular contexts of events and lives, whereby they reason and try to comprehend what happens to them, and in their world. This shapes what they do - but their acts and comprehensions are also tesred against the acts of others, and the resistance of the environment. There is a constant creativity in this; and people may use multiple images and perform a multiplicity of operations as they grope for an undersranding of the world, fallibly exchanging, adjusting and reconstructing their models as they harvest the experiences that ensue. This highlights an aspect of cognitive functioning that is very poorly retrieved in anthropology's conventional, structural models. We prefer the simplicity of isomorphy between cultural models and social formations in each ethnographic case. As a result, our anthropological accounts become far too simplistic. For example, in the previous section I mentioned the former Yugoslavia and evoked the familiar painful syndrome of ethnic boundary-drawing, stereotyping, and escalating violence and 'ethnic cleansing' in Bosnia. The account has a certain validity as a way to depict a pattern; but to understand how and why it comes about, we need to go far deeper into the processes. People are not that simple, and collective behaviour not rhat determined - for reasons that are only mystified if such simple models are used to ftame our analyses. Firstly, deteriorating inter-ethnic relations are certainly not a necessary outcome of making ethnic distinctions: ethnic pluralism variably conflicted has been a persisting feature in many regions, and escalating inter-ethnic violence occurs only somerimes, and then as rhe outcome of particular historical circumstances and processes. What we have seen in Bosnia has been produced by polirical entrepreneurs responding ro the affordances of ethnic boundaries at a particular moment of historical opportunity (cf. Barth 1994; similar situations are vividly documented from the former Soviet Union by Tishkov [1997}). In the former Yugoslavia, under new political parameters, new leadership positions could be constructed and factional followings could be mobilized by appeal to inter-ethnic steteotypes and intra-ethnic interests; and the recent hisrory of Bosnia is the runaway effect of such activities on the part of politicians. What made these effects so pervasive and fateful? Our colleague Tone Bringa (1995; and in her film We Are All Neighbours), who worked in Bosnia both before and throughout the conflict, shows how ethnic confrontation asserts itself at the local level. She depicts a tissue of relations of many kinds that linked the members of an ethnically and religiously composire community. People's interrelations were modelled on many quite different concepts and schemas, pre-eminently cast as 'neighbourliness', but also as exchange, love, affinity, kinship, work and friendship - as well as congregation and ethnicity. People's ethnic identity was given primacy for some few purposes and in some limited contexts only; and the idea of neighbourliness was dominant in defining interactional practice in most everyday activities berween persons of differenr ethnic identity. But in a politically deteriorating situation arising entirely outside the community which she describes, Bringa has shown how the boundary-based componenrs of people's identities, particularly ethnicity, increased in salience within the community as the tumours, and then the experiences, of inter-ethnic violence increased, progressively displacing in people's awareness and interpretations all those othet connecting experiences and bonds that had been part of the world that they had known. In the end, the very people who formerly entertained multi-stranded social ties came to embrace alienating and hostile positions, and thus became entrapped in the dichotomizing and boundary-based Bosnian impasse. In this they were responding to a sutge of experience, pressures and rhetoric by increasingly using one rather than the other of their many cognitive schemas. The lesson is that the elements in flux are affected by a multitude of individual experiences, and to look fot a simple match, a single 'structure' that embraces a shared cognitive model and an aggregate configuration of society, is a futile exercise. Lakoffs perspective would have us expect just that: people ate not merely playing out a structute, they are each a locus of reason and construction, using complex embodied imagery that they are trying to fit to what they perceive and experience. Yes, boundaries and mistrust are cognitive models that facilitate stereotyping, large-scale collective action, and counterposed positions and judgements; but this does not mean that the same people do not also harbour other cognitive resources that may open other paths for reason, action and relationships. This is also a lesson we can draw from the recent history of Israeli-Palestinian negotiations. By the early 1990s, you will remember, formal negotiations had been going on for years between official delegations restating their positions and demands across the boundary that defined and separated them. Unbeknown to practically evetyone, some brave individuals had also for years been secretly in contact across that chasm, exploring possible bases for reconciliation. Yet neither mode of negotiation seemed to be bringing any results. In this situation, the Oslo channel was opened: a sectet venue between the two sides. The Oslo channel was new and different only in that it enabled the same persons to operate in both modes, and indeed put some pressure on them to do so, in a very small group and in intimate settings. This served to give delegates from both sides of an ossified boundary the aftordance to teason and comprehend with a range of other images and figures, and thereby made them capable of deploying more and alternative cognitive models and ways of reasoning and interacting. The secret was to break loose from the constraints, the doomed ttajectoty, of talks choreographed entirely by boundary images, where negotiators were locked in as spokespersons and representatives of preset, counterposed categoties. Thus the Oslo channel enabled the authorized negotiators to retrieve parts of their broader humanity and activate a range of other images and identities - especially by facilitating some individual bonding across the boundary. The consequence was a deep restructuring of what had until then been vety recalcitrant political facts. As we have seen since, the complex political issues and real differences of interest between patties in the Israel/Palestine conflict are of course not resolved simply by the wave of a cognitive wand; but matters did move out of a deadlock, with the effect of creating a new political situation, once a felicitous context for meeting was created. The general point is that there is not a finite and one-to-one relationship between cognitive models and social action, though the two are fundamentally connected. An analytical concept of boundaries The image of a boundary, in the abstract sense of a separation that surrounds a social group and divides it ftom other groups and from its surrounding environment, has proved analytically powerful for many purposes in social science. In the study of the firm, it facilitates an account of the organization's exchanges with its environment, its inflows and outflows. It helps us envision the processes of recruitment and shedding of members in corporate groups, the viability of households, and the management of information by political, commercial and other institutional bodies. In the study of ethnicity, it has helped us understand the signalling effects of cultural idioms whereby membership in contrasted ethnic groups are made visible and thereby socially effective. In these and many other ways, it has proved useful and insight-giving, and will continue to be a powerful concept in our analytical arsenal. But in the uses I have noted in the above patagraph, 'boundary' has consistently been our concept, made to serve our own analytical purposes. An experiential anthropology, on the other hand, seeks to “grasp the native's point of view, his relation to life, to realize his vision of his world” (Malinowski 1922: 25). In such a perspective it becomes an empitical question what concepts and mental operations are used by a gtoup of people to construct their world — in this case, whethet a concept of boundaries is deployed by them to think about territories, social groups or categorical distinctions. Both discoveting or falsifying this, and providing documentation for its presence or absence, raise a very different set of issues from those that arise when judging the fruitfulness of our own analytical use of a concept. To ask these experiential questions - the answers to which seem to me to form an essential patt of any cultural analysis – we need a workable general theory of cognition. It further seems to me important that we clearly acknowledge this need, so we can work systematically to build the theoretical foundation for our pursuit of these anthropological questions. But whatever theory we use should obviously also be reflexively applicable to ourselves, including our own anthropological reasoning, if it is to have validity for other fields of human thought. The evidence for the fundamental role in human thought of fuzzy categories, the preconceptual experiential sources for conceptual structures, and the role of analogy, metaphor, metonymy and mental imagery, then raises some difficult conundrums. It may hold a promise of access to other human conceptual worlds; but it seems to place us in a hall of mirrors when trying to represent categories and concepts different from our own by means of our own language and concepts. Yet it is difficult to evade the issue. We certainly wish to make our ethnographies as true to life as possible, and that means persisting in the ambitious task set by Malinowski. But what is more, because the configuration of people's experience affects their lives and their acts, such an analysis becomes a necessary component in any understanding of social and cultural forms. If so, it cannot be sidestepped, though it is not alone sufficient for all our analytical purposes. In a hall of mirrors, one needs to move with considerable circumspection. A failure to do so may explain some of the confusions we labour under, in casu our anthropological muddles about boundaries. By taking careful steps, as I have tried to do in this essay, we may be able progressively to work our way out of some of these confusions. Note 1 Levi-Strauss, of course, represents a salutary exception; but mainstream anthropology generally chose to adopt aspects of his structuralism without serious consideration of its accompanying theory of human cognition. References Barth, F. (I960) 'The land use pattern of migratory tribes in South Persia', Norsk Geografisk Ttdsskrift, vol. XVII, part I. —— 1961) Nomads of South Persia, Oslo: Oslo University Press. ——(1975) Ritual and Knowledge among the Baktaman of New Guinea, New Haven CT: Yale University Press. ——(1987) Cosmologies in the Making, Cambridge: Cambridge University Press. ——(1994) 'Enduring and emerging issues in the analysis of ethnicity', in H. Vermeulen and C. Govers (eds) The Anthropology of Ethnicity: Beyond Ethnic Groups and Boundaries, Amsterdam: Het Spinhuis, 11—32. Barth, F. (ed.) (1969) Ethnic Groups and Boundaries, London: Allen and Unwin. Bloch, M. (1992) 'What goes without saying: the conceptualization of Zafimaniry society', in A. Kuper (ed.) Conceptualizing Society, London: Rourledge. Bringa, T. (1995) Being Muslim the Bosnian Way, Princeton NJ: Princeton University Press. Geertz, C. (1980) Negara, Princeton NJ: Princeton University Press. Hutchins, E. (1995) Cognition in the Wild, Boston MA: MIT Press. Ingold, T. (1993) 'The art of translation in a continuous world', in G. Palsson (ed.) Beyond Boundaries, Oxford: Berg. Lakoff, G. (1987) Women, Fire, and Dangerous Things, Chicago IL: Chicago University Press. Lancaster, W. (1981) The Rwala Bedouin Today, Cambridge: Cambridge University Press. Levy-Bruhl, L. (1910) Les Fonctiom Mentales dans les Socittes Primitifs (English trans. 1926 How Natives Think) London: Allen and Unwin. Malinowski, B. (1922) Argonauts of the Western Pacific, New York: Dutton. Shore, B. (1996) Culture in Mind: Cognition, Culture, and the Problem of Meaning, New York and Oxford: Oxford University Press. Szymborska, W. (1995) View with a Grain of Sand, trans. S. Baranczak and C Cavanagh, New York: Harcourt, Brace. Tishkov, V. (1997) Ethnicity, Nationalism and Conflict in and after the Soviet Union, Oslo: PRIO.