The Macro/Micro Problem and the Problem of Structure and Agency

Barry Barnes. Handbook of Social Theory. Editor: George Ritzer & Barry Smart. Sage Publications, 2001.

The General Form of the Macro/Micro Problem

All fields of empirical enquiry face a macro/micro problem in some form. Typically, such a field will concern itself with the study of things or processes of some particular kind, and their complexity and variability will lead to the thought that they are composite, and to conjectures about the nature and properties of their components. Or else it may be obvious from the start that this is the case, that the organisms being dissected, or the chemical substances being experimented on, or the nebulae being photographed, or the communities being studied, are made of smaller things. Similarly, although in practice less significantly, the things that a field studies may be themselves recognized as parts of a larger whole. Either way, the question arises as to how the properties and propensities of ‘macro’ things are related to those of ‘micro’ things, and how the enquiries of fields studying the one should be related to enquiries in fields that study the other. Of course, we cannot simply presume that there are general answers to these questions, but it is worthwhile, none the less, to look at the very general theoretical responses that have been made to them. In practice, they have tended to fall into one of three kinds. Nearly all accounts of the macro/micro relation involve reductionism, or dualism or else some form of pragmatism or constructivism.

Suppose we have two accounts, both of which refer to the same observable state(s)-of-affairs. One speaks of a single object perhaps, with such-and-such properties and propensities; the other speaks of many things, in proximity to and interacting with each other. Reductionism typically asserts that one of the accounts should be replaceable ‘in principle’ by a sufficiently elaborated version of the other. It may assert, for example, that an account of a biological organism as an integrated system reduces ‘in principle’ to an account of a cluster of molecules. This kind of example, wherein the thing described in the macro-account is held to be reducible to those encountered in the micro-one, is the most relevant to the present discussion. But there are also ‘reductions’ that move in the other direction, and seek to assimilate the micro to the macro. It may be suggested, for example, that particles are but the local properties of a field; or that colonies of coral, or of ants, are really extended single organisms; or that there are no such things as individuals, but only societies.

A macro-object, on a standard reductionist account, can be no better than some composite entity, attention to which is drawn purely for convenience. The world could be described, ‘in principle’, without reference to the ‘macro-object’ at all. Indeed such a description, sidestepping it in favour of its ‘fundamental’ constituents, would not merely substitute for the original description without loss or remainder, it would improve on that description in accuracy and detail and hence be preferable to it and more trustworthy than it. On this account, it is only because the ‘in principle’ reduction is in practice too difficult or time-consuming that the macro-level description is persisted in, if indeed it is persisted in at all; for this, of course is only the best case scenario. In the worst case, a reductionist analysis may claim to have exposed the fantastical character of the macro-account, and the non-existence of the macro-object. And between the best and the worst case are positions that may render the macro-object as a reification or hypostatization, or as a distorted and misleading representation of what is really the case.

Renditions of this last kind are particularly common in the social sciences, but in the natural sciences a more accommodating reductionism exists as a widely diffused ideology, employed to justify the disciplinary hierarchy. The familiar, longstanding ranking, wherein, for example, physics stands above chemistry, which in turn stands above biology, is often made out as a hierarchy of dependence wherein those who study small things are needed by, but themselves have no need of, those who study the larger things that the small things make up. Whether or not this familiar picture could be elaborated into an empirically adequate account, encompassing, for example, celestial mechanics and molecular genetics, is moot. And how far the picture will continue widely to be credited, as greater and greater prestige accrues to work in the biological sciences is, similarly, an open question. But for all that, the hierarchy continues to command significant credibility.

At the other extreme from reductionism are radical dualist positions that insist on the independence and autonomy of both micro- and macro-accounts, and on the irreducibility of the distinct and separate phenomena they describe. In practice, the task that dualism faces is that of acknowledging the composite character of macro-objects, and the fact that micro-objects and/or micro-processes are encountered within them, whilst at the same time insisting on their ‘irreducibility’. Reference to the existence of ‘emergent properties’ is probably the most widespread method of accomplishing this task. These properties are said to come into existence only as and when macro-objects do, and to be unpredictable from, and irreducible to, the properties, proclivities and interrelations of the smaller objects or entities ‘within’ them. Distinctive emergent properties of this kind may be attributed to macro-objects as diverse as chemical compounds, biological organisms, human brains, crowds of people and social systems. But at the same time it is interesting to note a certain deference to reductionism in the very language here, which invites us to imagine the macro, as it were, arising out of the micro. Why, for example, do we not speak instead of the ‘emergent’ properties of isolated, individual microentities, created by our acting upon ‘fundamental’ macro-objects?

How should we compare and evaluate reductionism and dualism? If our criterion is that of conformity to the contingent features of empirical enquiries themselves, it is arguable that both reductionism and dualism are inadequate. Even in the more recondite areas of the physical sciences there are ‘composite’ objects (benzene rings, electron beams, etc.) that must not be ‘reduced’, where the very equations acknowledged to apply to them scream holism, as it were. Yet it would be ludicrous to hold that those who have studied, say, benzene rings, have needed to take no interest in work on carbon atoms. Thus, reductionism faces fundamental difficulties even with (accounts of) atoms and molecules in the favourable context of physics and chemistry; but it does none the less point up the interconnectedness of fields of empirical enquiry, and the salience of claims made at one ‘level’ of description for those ‘above’ and ‘below’ it. Dualism, in contrast, whilst it rightly emphasizes the problems of reductionism, has difficulty in explaining why claims or findings made in one domain may be consequential in another supposedly autonomous one.

Let us assume that there are things like carbon and benzene in the world, and, more, that there really are atoms of carbon and molecules of benzene. We may still ask what in the world insists that atoms are ‘fundamental’ and that molecules are not; and what in the world tells us that anything we might conceivably ever learn about the one can, or cannot, be reduced ‘in principle’ to knowledge of the other. Questions of this kind, asked in a spirit of empirical curiosity, quickly prompt the thought that nothing in the world so insists, and that reductionism and dualism are alternative metaphysical accounts, or ontologies. And this is indeed what they are often understood to be. But this opens the possibility of walking away from both ontologies, and adopting instead a pragmatic view, wherein the relation of macro- and micro-things, being a matter of indifference to the world, is decided by people for their own convenience. On this view, the ontologies of reductionism and dualism take on the character of dogmas, or postulates accepted by convention. Of course, a pragmatic approach of this kind entails a changed understanding of the nature of macro- and micro-things as well as of their relationship. Just as the world itself no longer tells us about the reality of their relation, so it no longer tells us about their reality per se. Macro- and micro-things both become the products of the classifying activities we decide to employ in our dealings with the world. This entails monism, but not reduction. Both kinds of thing have the same standing. As to their relationship, that is a wholly contingent matter.

If we wish, we may address this contingent matter, and ask what as a matter of history has most often inclined people working at different levels of enquiry to take account of each others’ findings? Common sense would suggest that a need for consistency in practical inference is perhaps the dominant consideration. The thought here is that a certain kind of consistency has to exist between micro- and macro-descriptions, because these descriptions do not relate to distinct and separate states-of-affairs (Barnes, 1995: 85-8). The cat on the mat may be described as an organism, a system of cells, a molecular system, or even as a cat, but these descriptions must not place conflicting demands on the cat itself that is all these things. Thus, whereas no practical difficulty need arise if, for example, the cat as specific organs were observed to be far more stable than the cat as specific molecules; if the cat were apparently far more massive as organs than as molecules then serious questions would arise, however much dualists insisted on the independence of organic and molecular phenomena.

As an illustration of more obvious relevance to sociology and social theory, it is worth noting that suicides as suicides, and the same suicides as a suicide rate, are not independent states-of-affairs, and that to treat them as independent micro- and macro-phenomena risks creating just this kind of practical difficulty. And, finally, to illustrate the real historical importance of the point in the natural sciences, it is worth mentioning that the planet earth described as a physical system is no other than the earth described as a life-supporting system. The particular fascination of this lies in the fact that, according to the accepted science of the late-nineteenth century, the age of the earth as a life-supporting system was orders of magnitude greater than its age as a physical system of sufficient temperature to be such. What we have here is the only major historical episode wherein knowledge of physical and chemical materials and processes has come into conflict with biological and geological accounts of living things and their history. As it happens, the outcome of that clash was clearcut: biology won (Burchfield, 1975).

The Macro/Micro Problem in Sociology and Social Theory

For those working in a specific field of empirical enquiry, weakly linked to other specialized fields by modest interdependence and a modicum of mutual respect as is typical, the macro/micro problem might be thought of little moment. The patient observer of the three-toed sloth need not consider whether said sloth is a real essence, or a reified process, or a complex aggregate of molecules; or whether the resulting observation-reports are simple truths, or convenient simplifications, or illegitimate constructs. It suffices for her to speak of the sloth as sloth, even if her reports describe its responses to adrenaline, say, and are consciously intended to be of interest to chemists as well as biologists. The relative merits of different descriptions here, and the relationships between them, are likely to appear as ‘merely philosophical’ issues. Yet controversy about these ‘philosophical’ issues has frequently erupted in the natural sciences. And it has engendered debate in sociology and social theory as long as these fields have existed. Indeed, controversy over the macro/micro problem remains remarkably intense in these last fields, wherein merely pragmatic considerations are often overshadowed by ‘fundamental’ arguments about ontology. Although it is far from being a universal obsession, many macro-sociologists and social theorists, in contrast to observers of the three-toed sloth, are anxious to establish the reality of the macro-objects they describe, and eager to broadcast the defects of reductionism.

It is not hard to see why macro-sociologists are sensitive on this matter. On their right, they face the individualistic reductionism of what in terms of external recognition is the most successful of all the social science fields, economics. On their left, they face deconstruction by micro-sociologists and their allies and affines. And neither of these sources of difficulty seems sufficiently well disposed towards them to allow their references to institutions, or classes, or cultures, or social systems, any standing as ‘convenient simplifications’. In macro-economics, firms, markets and currency flows represent an alternative ‘convenient simplification’, whilst for many micro-sociologists no simplification is ever convenient, and interest in macro-objects tends to be confined to identifying the precise way in which they are harmful.

It is not only problems with the neighbours, however, that give macro-sociologists a sense of vulnerability. In most fields, however much macro-descriptions may over-simplify by ignoring the composite character of macro-objects, they do at least remain less problematic epistemologically than reductionist micro-accounts; for macro-objects are easier to see, as it were, than micro-objects. But in sociology and macro-social theory this compensating virtue of macro-descriptions does not exist. Here, macro-objects are the harder to see, and indeed they often have the standing of invisible theoretical entities and not of objects that may be seen at all. Whilst individuals, and situations, and encounters, are by no means unproblematically ‘there’, they give rise to fewer practical-epistemological problems, as it were, than institutions and social systems do.

None the less, sociologists and social theorists remain reluctant to renounce macro-entities. Social theorists, in particular, often regard their references to invisible entities and mechanisms as essential to their central task of critical evaluation, and associate micro-description with the uncritical acceptance of appearances. Whether they are right to do so is, however, another matter. If critical potency and profundity are indeed to be taken as the primary criteria of good work here, then nowhere is honour more deserved than in the various traditions of micro-sociology, where careful descriptive studies remain able to inspire deep-seated insecurity in apologists for existing institutions and hierarchies. In contrast, the most eminent practitioners of what currently passes for critical social theory belong amongst those apologists. In the macro-theory of Jiirgen Habermas, for example, we find an account of how modern capitalist societies embody a balance between the spheres of ‘system’ and ‘lifeworld’ that is pretty well just right. And this anodyne vision of how we currently stand is, if anything, even more clearly apparent in the recent writing of Anthony Giddens. Both these ‘critical’ social thinkers now deploy theory to identify and applaud in capitalist societies a politics that represents the apotheosis of the one great resilient and enduring moral ideal of our century, that of ‘bourgeois equality’.

Be that as it may, it remains the case that many macro-sociologists and social theorists currently regard the macro/micro problem as that of maintaining the defence against reductionism. And they are surely right in one respect: there is a lot of reductionism around to defend against, particularly in the guise of rational choice theory and related forms of individualism. Enthusiasts for this theory have long entertained the ambition of rendering the whole of our social life as so many calculated individual actions, and whatever macro-order there might be as the unlooked-for by-product of the relevant calculations. Conversely, critics have often regarded it as an imperialistic and undiscriminating intellectual movement lacking any genuine empirical curiosity—one that has sought to make sense of all it has encountered within a preordained framework, rather as Marxism used to do. There is certainly some justice in this description. But imperialistic tendencies and the urge to a thoroughgoing reduction have also had the beneficial effect of bringing the theory into prolonged and fruitful confrontation with its most serious difficulties and most recalcitrant counter-examples. In my judgement, it is greatly to the credit of rational choice theorists that they have themselves focused attention on these difficulties and opened the way to the conclusion that they are insuperable, and fatal to their position. The powers of human beings to engender shared understandings across cultures, and coordinated action for the indivisible good of collectives, is simply unintelligible on the assumption that they are independent individuals.

Of course the irreducibility of macro-social phenomena to the actions and calculations of independent individuals does not preclude the possibility of alternative reductions. All the micro-objects of sociology are possible resources here: we may explore whether macro-entities are perhaps ‘really’ sets of encounters, or chains of interactions, or discursive exchanges or sequences of practices. And it is hard to see how we might identify any macro-entity or property, by inspection as it were, as ‘emergent’ and ‘irreducible in principle’, and thereby confound reductionist aspirations altogether in relation to it. None the less, whilst it is perfectly possible in a formal sense, it would be seriously misleading to render the relationship of micro- and macro-sociology as if it turned on the philosophical issue of reductionism. Unlike the argument with rational choice, the ‘internal’ arguments between macro- and micro-sociologists and social theorists have not been dominated by metaphysical and ontological issues. Indeed pragmatist and constructivist perspectives have been and remain well represented in this context, and their characteristic methods of understanding objects as secondary to processes—not as entities self-evidently there in the world, but as the products of human activity ongoing in the world—are much in evidence.

Pragmatist and constructivist orientations to macro- and micro-objects, and hence to the macro/micro problem, have interesting implications for all fields of enquiry, but they have a special significance in social theory and sociology. For these fields have a special responsibility to provide an understanding of activities, and if objects are constituted through activities then both those objects and the relations between them become foci of sociological curiosity. The macro/micro problem is then not a problem within theory, of the relations of those levels of theory associated with different real-world objects; it is a problem for theory, an observable product of human theorizing activity which has to be made intelligible in and through that very activity. It must now be asked how and why human beings choose to set different orderings on the world, and why they relate those orderings to each other as they do, for example, as micro- and macro-orderings.

Interesting illustrations of how a perspective of this kind may address the macro/micro distinction can be found in the work of Michel Calln and Bruno Latour, who have repeatedly confronted this issue from the perspective of their ‘actor-network’ theory (Calln, 1986; Calln and Latour, 1981). In their studies, ‘the observer follows the actors in order to identify the manner in which these define and associate the different elements by which they build and explain their world, whether it be social or natural’ (Calln, 1986: 197). In the course of this defining and associating by actors both micro-and macro-objects are constituted; their properties are attributed; and the relations between them are established. No intrinsic differences are presumed between macro- and micro-objects. Objects of different ‘sizes’ are alike in being the products of the ‘associations and definitions’ that actor network theorists seek to describe. And they are alike in their standing: neither kind of object is reducible to the other; neither has ontological priority over the other. Although given macro-objects may be the products of the growth of smaller objects as a matter of history, all objects are the products of the same kinds of processes. The macro/micro distinction dissolves into an uncompromising monism.

The monism of Callon and Latour is indeed all-pervading. Not only do they treat micro- and macro-objects as equivalent, they try to treat humans and non-humans as equivalent as well and to mark no distinction between natural and social objects. For them all objects are alike in being the products of associations, and in their work associations created by scientists and engineers figure especially prominently. They have sought to emphasize how social and natural orders are simultaneously constituted in and as these associations, along with all the range of objects, large and small, they are acknowledged to contain. A major part of the message of this work is indeed that the (scientific) study of the natural world constitutively involves sociological enquiry, that social life similarly involves the (scientific) study of natural phenomena and that the macro-objects engendered in the course of all this are hybrid ‘quasi-objects, neither ‘natural’ nor ‘social’.

Actor-network theory is now very widely used in studies that seek to document how macro-objects get constituted. It has, of course, been subjected to extensive technical criticism, like any theory, but it also faces formal difficulties because of its status as a theory of theories. The objects and entities ‘created’ by use of the theory, and in that sense internal to it, can be addressed in just the way that the theory itself addresses objects and entities external to it. Calln (1986: 200) gives a clear response to the problem of justification implicit here: ‘our narrative is no more, but no less valid, than any other’. What he does not clarify, however, is what pragmatic or practical considerations make this particular narrative preferable to ‘no less valid’ alternatives.

Actor-network theory is sometimes defended as the most rigorous extant expression of the principle that all accounts and all actions, including those of the sociologists themselves, must be treated equally in sociological study. But there are difficulties here. Objects are said to be the creations of the actors being followed, and the properties of the objects to be those that actors assign to them. But at the same time, actors are amongst the created objects and their agency is just such an assigned property; indeed agency may be assigned to non-humans, whereupon they become actors, presumably with powers of assignation themselves, whatever that may imply. The evident circularity here – wherein actors are the creations of actors, that are the creations of actors, that …—gives no concern to users of the theory. They simply decide for themselves which actors they will follow, and proceed. But in making prior decisions of this kind actor-network theorists come very close to deciding in advance just how the historical episodes they are about to address will be made visible.

It is important to notice how much of what Calln and Latour describe as they ‘follow the actors’ is preordained, as it were, rather than ‘found’. In practice, prior external decisions fix not only which actors will be followed, but also what theoretical schemes will be used to make sense of what those actors do. These schemes seem predominantly to be drawn from economic and political theory; and on the face of it they offer unedifying visions of human beings as Hobbesian warriors, or ruthless entrepreneurs, or Machiavellian calculators. The stories constructed from these theories, however, never aspire to a systematic explanation of what the actors are doing. Instead, they render events ex post facto, as displays of human activity and potency, and this is perhaps what makes them so attractive.

It is intriguing to notice here how their inclination to render humans as active is related to the conception of social order favoured by Calln and Latour. For them, the social order is the entirety of manifest social activity, which means that more or less by definition anything that an agent does reconstitutes and in a sense transforms the social order. Notice the intensified sense of agency conveyed by this vision of things, in comparison with that encouraged by the alternative vision wherein a stable social structure underlying manifest activity is the central macro-social object. It is a nice illustration of how prior decisions on frame can have profound remote consequences as theories unfold.

In the narratives of actor-network theory human beings construct and manipulate their world, overcome resistance, create and then realize interests, all with amazing facility. It is indeed a valid cause of dissatisfaction with these narratives that justice is not done to the manifest difficulty of carrying out such projects. These are projects involving learning about and acting upon real-world objects, or so those engaged in them might want to say, and they are as likely to end in frustration and failure as in success. But actor-network studies seem attracted to success and pleased to fashion it into mock-heroic history (Latour, 1988). For all that at one level actor-network theory modestly ‘follows the actors’ and marks no distinction of its own between humans and things, at another level it is a profoundly intrusive monism engaged in the celebration of human agency. For many readers it provides a narrative of contingency revelatory of how things could have happened otherwise (Michael, 1996).

‘Structure’ and ‘Agency’ in Sociology and Social Theory

Sociology and social theory have always been interested in the nature of voluntary action and in how, if at all, to take account of the free agency of human beings whilst constructing theoretical accounts of their actions. Currently, however, these issues are being debated as never before. Perhaps a spreading sense of empowerment has something to do with this; especially through the way that empowerment engenders negative attitudes to the burgeoning ranks of bureaucrats and technical professionals whose knowledge is structured in terms of the institution of causal connection. But whatever the reason (or the cause) may be, it is clear that there is now an unprecedented level of interest in the nature of human agency, and that macro/micro debates have largely become debates about the relationship of agency and structure.

To grasp what is at issue in the structure/agency debates it is necessary to glance back to the sociology that established itself in the English-speaking world half a century or so ago, and sought acknowledgement as an authentic science. Human societies, so it seemed to its practitioners, were ordered and patterned. The order and pattern was evident in the voluntary actions from which they were made. The task of a science faced with such a pattern was to explain it or account for it. Accordingly, the task of sociology was to establish why voluntary actions were patterned as they were. There was a need for voluntary actions to be explained, and, by analogy with other sciences, explanation meant linking the actions to causes, or influences, independent of them. Among the cited influences were class interests, external coercive powers, social pressures and, in the structural functional sociology that came to dominate the field, rules and social norms. Curiosity was satisfied by appeal to these kinds of things as externalities. What is making people act thus and so? They are conforming to norms. Why is there an overall pattern in their actions? Because there is an overall pattern in the norms. What is the pattern? It is that of the social system or structure of the society in question; and by reference to that system or structure, wherein rules and norms are ordered around statuses to form social institutions, actions may be understood and explained. Thus emerged the vision of social structure famously associated with the work of Talcott Parsons and long the theoretical mainstay of sociological work in the English-speaking world.

It is worth noting the oddity of this version of structure. Cut down a tree and the eye remarks the ringed structure of the trunk, the array of concentric circles visible in the cellulose fibres that make it up. There is a standard macro/micro problem here. Are the rings really there, or can accounts of them be ‘reduced’ to accounts of cellulose fibres? Will the growth of the fibres explain or fail to explain the existence of the rings? What we don’t tend to ask is whether the ringed structure explains the growth of the fibres, as if the structure is there in advance, as it were, and the fibres grow as it requires or specifies. Yet this is how social structure has sometimes been understood; as real, separate, prior, macro and explanatory. Actions manifest pattern; the pattern is described; the pattern is taken as a separate real macro-entity; the macro-entity explains the actions. This, at least, is how critics of this form of macro-sociology have seen it, and why they have dismissed it as illegitimate reification.

As it happens, this peculiarity of social structure and of ‘social-structural explanation’, did not prevent their constituting a perfectly acceptable macro-sociology for a considerable period, and thereby constituting the backdrop to the debates here discussed. And indeed it is tempting to suggest that a secular pattern of social change involving an ever-increasing concern with the dignity and standing of ‘the individual’ is what brought about its demise, and not its inherent inadequacies. In any event, distaste for the passive role allegedly being accorded to the individual ‘human subject’, did eventually become irresistibly strong, and the offending ‘structural functional’ theories were set aside. Indeed the reaction against the institution of causal connection and the urge to stress internal autonomy and activity came to encompass far more than human beings. Entire monistic cosmologies of agency were constructed, and accepted as appropriate descriptions of objects of all kinds, animate and inanimate. The move to celebrate human agency now to be discussed was actually part of a larger, possibly still more fascinating, shift of perspective.

One famous formulation of what was wrong with the older causal theories was that of the ethnomethodologist Harold Garfinkel. He denounced them for making out individual human beings as ‘dopes’, that is as ‘passive producers of actions to the specification of whatever structural or cultural models of their conduct were available in the cultural setting, rather than as knowledgeable human beings aware of and able to take account of the existence of those models’ (1967: 68). Consider now a macro-sociologist, disinclined to doubt the large-scale orderliness of human societies or that it has something to do with norms or rules, but impressed notwithstanding by the ethnomethodological argument and by the lovely studies of the artful and reflexive character of human actions that Garfinkel’s work inspired. For such a sociologist at least three problems are likely to arise. First, how might human beings be understood, other than as dopes of some kind? Secondly, what new account of the relationship of human beings and social norms or rules might such an alternative understanding lead to? Thirdly, will the evident macro-orderliness of societies be intelligible on the basis of any such new account? The case studies of ethnomethodologists offered direct assistance on the first two problems. They documented human beings taking account of rather than merely ‘following’ rules, and actively construing and interpreting them in ways that suited their practical purposes. In effect, rules were more the outcomes than the determinants of actions, and in that guise they were known of and utilized by human beings actively pursuing their practical purposes.

Much the most widely known attempt to reconceptualize macro-sociology on this kind of basis is Anthony Giddens’ structuration theory. Giddens refers to language to convey his thoughts about rules. Human beings draw upon the resources of language in all kinds of ways, for all kinds of purposes. The language does not tell them what they may or may not say. They may use it how they will. And in using it as they will, however that may be, they reconstitute the language as a set of resources available for use. Similarly, a set of rules may persist as something that is drawn-upon in and by a collective. In an idiosyncratic departure from normal usage that has sometimes caused confusion, Giddens calls such a set of rules a social structure. A social structure both facilitates and constrains action that draws upon it; it facilitates action by virtue of what it makes available and constrains action by virtue of what it lacks and cannot make available. It does not, however, stand as a determinant or even as a cause of that action, since it does not ‘make’ the active agents who engender it do one thing rather than another. Indeed it is those agents, deciding to do whatever they decide to do by drawing upon structure, and thereby manifesting their agency, who reconstitute structure through their actions and secure its continued existence. Thus, Giddens is able to give both the macro (structure) and the micro (agency) an essential role, whilst rejecting any causal connection between the two, and in particular any suggestion that the operation of human reason may be externally determined.

There is no point in this brief space in attempting either to review or to add to the innumerable commentaries on structuration theory, but some remarks on its macro and micro dimensions may be worthwhile. It is important to note, first of all, that ‘structure’ is not the only macro-entity in structuration theory. Just as the members of a language-sharing community routinely continue to say some things and not others, to use language in some patterns and not others, so it is with those who draw upon the repertoire of rules that is social structure. The particular patterns thereby engendered and reproduced are social systems, and the persistence of social systems needs to be accounted for as well as that of structures. What is necessary here is an account or explanation of the particular ways in which agents choose to draw upon the resources of structure. Giddens’ response is to suggest that individual agents have a need for ontological security, essential to tension management and anxiety reduction within themselves, and itself dependent on their implicit faith in ‘the conventions (codes of signification and forms of normative regulation) via which, in the duality of structure, the reproduction of social life is effected’ (1979: 218-19). In order to sustain this sense of ontological security agents tend to act to reproduce the specific social system they inhabit as it already exists, and this is how the system persistence observable almost everywhere is to be explained.

It needs to be asked whether this does not return the analysis very close to the point whence it started. Accounts of anxiety-reduction and the maintenance of ontological security are drawn from a broadly Freudian tradition of psychology, and what they refer to are causes of, or causal influences upon, how individual persons act. So it could be said that implicit here is an individualistic micro-causal account of how social systems are reproduced. Indeed, it could be said that the need for ontological security makes persons act in accordance with the system-status-quo, that it disposes them to conform to system requirements and existing ‘forms of normative regulation’. This makes the account look very like the outmoded functionalism of Talcott Parsons, wherein the reproduction of the status quo was similarly linked to ‘forms of normative regulation’. And the resemblance becomes still more striking when it is recalled that what Giddens calls ‘system’ is very like what earlier theorists called ‘social structure’. Indeed, the enduring value of earlier theories, which rendered social structures in terms of statuses, institutions and other macro-objects, is readily appreciated once this line of thought is pursued. That human beings everywhere orient to each other in terms of social statuses, for example, must be accounted for in any sociology; but it cannot be accounted for simply by reference to the individual agents and rules that dominate a ‘structurationist’ perspective. There is currently a manifest need to renew and reinvigorate the traditional macro-sociology of statuses and institutions.

Now to the micro level, where Giddens’ account of the notion of agency may be taken as typifying that now favoured by large numbers of theorists. Clearly, it is necessary for his theory that individuals are not the dopes of rules, that they ‘have agency’ in the specific, narrow sense of being active in relation to rules and norms and not predictably compliant with them. But his actual vision of agency far transcends this necessity: it is a larger vision of agency altogether, of the kind we often entertain in everyday life when we speak of someone as a free agent. Agency in this large sense is the independent power of the individual human being to intervene in the ongoing flow of events and make a difference to them, her power to ‘act otherwise’ as Giddens sometimes says. This assertion, in effect of the freedom of the individual and her ability to change things, is a metaphysical postulate, not an empirically plausible claim. It expresses a prior belief in ‘the freedom of the acting subject’ and in the standing of that subject as an undetermined source of power. ‘It is analytical to the concept of agency’, Giddens says, ‘that a person (i.e. an agent) “could have acted otherwise”’ (1976: 75; emphasis added). And it is similarly held to be true of action, even routine action, that in being action and not mere behaviour it ‘could have been otherwise’ by virtue of the agency of its performer.

This characterization of human beings as active agents who can ‘make things happen’ is unquestionably the source of much of the very widespread appeal of structuration theory. But it is important to ask what considerations might dispose us to accept it. There is indeed reason to believe that persons are active in relation to norms and rules. But whilst we might consequently wish to speak of agency in relation to rules and norms, why would we want to go further and speak of agency per sel Giddens himself offers us nothing here. Indeed, his efforts are largely confined to explaining why agency is so rarely manifest as a transformative power; why, even in relation to rules, it is the routine and familiar modes of use that are overwhelmingly in evidence and relatively little that is new is ‘made to happen’. On ‘agency’ in its unrestricted sense, the agency of his metaphysics, he is unenlightening.

It is interesting to notice that the ‘agency’ of this metaphysics is very close to the ‘agency’ of everyday discourse—to the metaphysics of everyday life as it were. We routinely say of other people, ‘she was a free agent’, or ‘she didn’t have to do it; she could have acted otherwise’, or ‘it was her choice to do it’; and we do not speak in this way of objects. It is, of course, the everyday use of such notions in the context of references to voluntary actions that makes theoretical discussions of ‘agency’ and ‘choice’ so readily intelligible. Thus, having asked what the metaphysics of agency does for social theorists, and found it hard to arrive at a satisfactory answer, it is worth asking what that metaphysics does for us as ordinary members.

‘Agency’ and the Institution of Responsible Action

Custom has it that writers of pieces such as this may conclude with some of their own ideas and reflections, and I shall grasp the opportunity custom presents to offer some thoughts of my own on ‘agency’. The discussion will, however, differ radically both in form and focus from what has gone before. The initial form will not be provided by macro/micro, or structure/agency, or individual/society, or any other pair of distinctions, but by a monistic vision of interacting human beings presumed to be intrinsically sociable and interdependent. And the initial focus will be on ‘agency’ as used in the context of ordinary life, wherein it is deployed along with a number of other notions, sometimes called ‘voluntaristic’ notions, as a constituent of the institution of responsible action. Naturally, these changes in the initial frame of reference produce changed conclusions. The conception of agency that then emerges (which I discuss in detail in Barnes, 2000) contrasts sharply with those that dominate the current literature (Emirbayer and Mische, 1998).

First, form. Imagine that the social world is formed of sociable, interacting, interdependent human beings. This is more than mere assumption. There are good empirically grounded arguments for it, although these will be passed over here save only for one that is particularly relevant in this context. Think of the relationship of people with rules and norms again. Older macro-social theories rendered people as the ‘dopes’ of rules, or so it is now claimed. More recent theories have them ‘drawing-upon’ rules. But what and where are these rules? Just what is it that we are said to be dopes of, or to draw upon? This is an unresolved issue of enormous sociological importance, that continues to engender controversy between rule realists, rule individualists, rule collectivists and so forth.

The crucial thing in any attempt to grasp what rules are is to keep the imagination, and ideally the senses, close to the actual business of their invocation and use. Look at people driving cars perhaps, or exchanging courtesies, or eating a meal together, or participating in a march, or even individually, adjusting dress, or measuring a carpet or drawing a circle. All that will be found in this way is instances cited as examples of rule use, and clustered into sets by being so cited. It is through examples that rules are learned and as examples that they are encountered from that point on. This does not imply, necessarily, that ‘there are no such things as rules’, although it does point up their character as reifications. What is implied, however, if rules are only apprehended through instances, is that their application must consist in the making of analogies with existing instances. The importance of this point is that persons are liable to differ on how to extend analogies, and if they were to proceed as so many independent, isolated individuals, even if they were to start with identical examples or instances, they would lose coordination and be liable to disagree. Because examples differ in detail from each other, and what constitutes proper analogy between them is always contestable, uniformity in the following of a rule shared by many individuals cannot be put down to their common awareness of some essential feature or property of ‘the rule itself. If a continuing sense that there is ‘a rule there’ is to be generated, and a right way of extending the rule to future cases and situations is to be identified, all the various attempts at correct rule use across the collective will have to be considered by its members; a sense of which of these attempts are right and which wrong will have to be engendered; and a shared sense of how to move from agreed ‘right’ instances to new applications will continually have to be sustained.

In all manifestations of social life, what would otherwise be diverse individual inclinations in the specification and application of rules must be ordered into a tolerably coherent collective practice. Without this there can be no sense that one way of continuing to follow a rule is better than another, and hence there can be no rule. But independent individuals have no incentive to do the work of evaluation and standardization constantly necessary to sustain a sense of ‘what the rules are’. Only non-independent human beings will do this work, human beings who are indeed active and independent in relation to rules but not in relation to each other. Sociable human beings, capable of affecting each other implicitly, causally and continuously in their communicative interaction, may coordinate their understandings and their actual implementations of rules in this way, whilst independent individuals may not. And this coordination is, of course, just what is invariably found in practice.

There are many forms and varieties of the argument given above, but all point in the same direction. They evoke a vision of people disposed all the time to retain coordination in their practical sense of what rules amount to, what their implications are, what their normativity consists in, and of their managing to do so through interacting together and influencing and being influenced by each other. It is important to be clear as well that coordination of understandings is involved here, and not just coordination of physical actions. Shared understandings are the products of the effort to coordinate, not its basis, which means that ‘rational’ verbal exchanges, reliant upon ‘shared meanings’, are only possible, if at all, when coordination already exists; they can in no way be understood as the means of securing it. This is a point of major importance. If it is correct then rationalist versions of communicative action, and notably that of Habermas (1984, 1987, 1990, 1992), are fundamentally flawed, and indeed the rationalist theory of individual human agency so persuasively advanced by Habermas is revealed as fundamentally defective as well. Proponents of this kind of theory should steel themselves to recognize that there must be something causal at work, all the time in the course of communicative interaction, that keeps people sufficiently well aligned with each other for mutual intelligibility to continue. People must be mutually susceptible, in a causal sense, in their interactions and communications. Susceptibility of this kind has, of course, long been recognized and documented in the classical interactionist studies of micro-sociology. In Erving Goffman, for example, it is visible as the concern of members to avoid loss of face. The ubiquitous need to keep face is what constitutes mutual susceptibility.

Let us put that on hold, now, and switch attention from the form to the focus of the discussion. In everyday life we speak of and to each other as responsible agents, deploying all the familiar concepts of everyday voluntaristic discourse. We describe each other as acting voluntarily, or else under constraint, as making choices, as seeing reason, or else of not seeing it or having lost it. It is common to codify this discourse of responsible agency using an individualistic idiom, wherein the state of individual responsibility involves the possession of two key powers or attributes: the capacity for rational conduct and freedom of will. Indeed, rational choice theory is very close to being an individualistic codification of this kind, wherein action is related to the internal individual states of rationality and power to choose. But there is a notorious problem associated with these internal states. Not merely are they invisible: there seems to be no fact of the matter to tell us when they are operative and when not, or indeed whether or not they ‘really’ exist. It is difficult, for example, to understand just how we distinguish in practice between chosen and caused behaviour, or voluntary actions that ‘could have been otherwise’ and behaviours that could not. These are crucial distinctions; and we make them all the time, as we must, usually with consummate ease. On what basis we make them, however, is highly problematic, and not even systematic studies by psychologists have thrown much light on the matter. None the less, we evidently do believe that a distinction is there to be made, and indeed we evidently do make it and imagine ourselves guided by empirical considerations when we do so.

Let me now bring the form and the focus of the discussion together. We are given to speaking of ourselves as responsible agents, and to deploying the notions of our everyday voluntaristic discourse in characterizing ourselves as such. A widespread individualistic codification of that discourse renders us as independent agents, possessed of rationality and free will. But rationality and free will, understood as the internal states of independent individuals, are elusive, and just what prompts our use of these notions in this sense is notoriously hard to discern. Nor ought this to surprise anyone who recognizes that social life is actually constituted as interacting, non-independent, mutually susceptible human beings. However, if this is indeed the case, then it must be asked how voluntaristic discourse actually maps onto such human beings, and what it is about them that prompts us to describe them in voluntaristic terms.

Human beings are properly described, let us agree, as sociable creatures whose interactions are characterized by mutual intelligibility and mutual susceptibility. Human beings none the less describe each other as responsible agents, possessed of rationality and free will. Both descriptions refer to the same human beings. The conjecture must accordingly be that responsible agency, as specified in everyday discourse, comprises mutual intelligibility and mutual susceptibility, the basic necessities for the maintenance of coordinated interaction. And we can conjecture further that, in a rough and ready way, the rationality of the responsible agent is her intelligibility (or accountability), and the free will of the responsible agent is her susceptibility. But of course it is not merely that we describe each other (in our normal ‘responsible’ state) with this voluntaristic discourse; it is that we communicate in terms of it and affect each other through these communications. To identify someone as a normal responsible agent is to recognize her as open to influence, as capable of being affected, by our communications; and to remind her in those communications that she is a responsible agent is part of the business of affecting her through them.

Voluntaristic discourse needs to be understood primarily as the medium through which our sociability is expressed, not the medium in which our independence is celebrated. And the sociability so expressed needs to be recognized as a distinctive strength of our species. Most of the order in our social and mental life is sustained by mutually susceptible responsible agents who press each other to do what is necessary to create it, continue it and change it. Communicating and evaluating, learning and suffering, in the course of our social interaction, our collective accomplishment far transcends anything that could be hoped for from independent individuals. Parts of this accomplishment are groups and collectives, offices and hierarchies, institutions and organizations, all recognizable in turn as features of the landscape in which we live our social lives, and hence as loci of responsibility and accountability in their own right. In this way the everyday voluntarism of the institution of responsible action now expresses itself in the complex and elaborate systems of practice we recognize as those societies in which we presently live. The simple institution allows the elaboration of the larger society: the larger society is the simple institution elaborated. And voluntaristic discourse is the medium through which all this is achieved and sustained, the characteristic form of communication in which the inherent sociability of human beings is so potently expressed.

In modern theorists such as Anthony Giddens (just as in Immanuel Kant), agency is the individual capacity to act otherwise. This alleged capacity is amongst the most opaque and obscure of all the inner states, or powers, or capacities, spoken of in voluntaristic discourse. What there is, we might say, is all there is; what is otherwise, if it were, would not be otherwise. But we have a strong sense that human action ‘could have been otherwise’ none the less. What prompts it? Consider whether it is not an awareness that the action ‘could have been otherwise’ if only communication, or persuasion, or some kind of symbolically mediated ‘pressure’, had been directed toward the agent involved. Consider, in other words, whether normal free action is not that which we regard as capable of modification through communication. This is what is being suggested here. And if correct, it transforms a vision of agency as a mysterious independent uncaused power into one where all actions can be understood, if that is what we wish, by routine recourse to the institution of causal connection. Free action need no longer be an uncaused intervention; it may be action with whatever causal antecedents, given only that we reckon it readily variable by use of the causal powers of communicative interaction. In that interaction we do things with voluntaristic notions; we do things to and with those we address as responsible agents.

To suggest this, of course, is no small matter, since it is to deny an individualistic conception of agency long accepted amongst us and with significant implications for how we live our lives. But the account being proposed, whilst it does indeed deny individual agency as generally understood, will provide more than sufficient compensation if it has correctly identified the interactive basis of the collective agency that engenders the cultural and institutional order intrinsic to human life. And of course if it has correctly done this, then it will also have identified the appropriate starting point for reflection on macro/micro problems in the context of sociology and social theory. There is no space here to explore the many macro-sociological implications of the inherent sociability and susceptibility of human beings, but what follows above all is that macro-entities should be identified by attending, not to similarities in the characteristics of individuals, or even to similarities in their situations, but to the connections and relations between those ‘individuals’. Consistent with this, I have suggested elsewhere that status groups and extended status orderings will be extremely important elements of any macro-theory designed to be consistent with a micro-sociology of interacting sociable agents. Indeed, a micro-sociology of face and a macro-sociology of status could well prove amenable to synthesis into a coherent overall perspective that could underpin and inform social theory as a whole.

There remains one final point to be dealt with if these brief suggestions are to carry plausibility. If we are agents in the sense outlined here, why do we not explicitly recognize ourselves as such? Why instead do we reify our collective agency into an individual power? Part of the answer may be that reifications of this kind are extremely common in our culture, and that the conversion of accountability and susceptibility into rationality and free agency exemplifies a common process. We are very prone to render the relational properties of things in context as the internal powers or properties of the things. We render statuses as states. The responsible agent possesses the rudimentary relational properties to count as being of that status, whereupon the status is misread as a part of her nature, as her internal state. Similar reifications include the value of money and the power of political leaders. It is important, however, to recognize the special importance of what is reified here. The status of the responsible agent is the most rudimentary of all social statuses, and the crucial default status of the institution of responsible action. As such, it is a vital focus for the attribution of responsibilities, for praise and blame according to how they are met and for demands for response when failure to fulfil them gives rise to damage to others. But responsibility must be localized, whereas causal connection delocalizes. The ideal carrier of responsibility is an uncaused cause, a clearly distinguished and demarcated target for demands and expectations (and, as optional extras, rights and powers). And even if in reality human beings are buffeted by causes like flotsam in foam, it may still be that these causes are largely discounted, or airbrushed from reality by various devices, for this reason. It may be indeed that such devices have always and everywhere been employed to identify ‘individuals’ and their powers, since all societies attach responsibilities to individual persons. But their deployment surely reached a zenith of skill and sophistication in our own recent history: the myth of individual agency as an independent power is, after all, the very essence of the myth of the Enlightenment.