Definitions of power are legion. To the extent that there is any commonly accepted formulation, power is understood as concerned with the bringing about of consequences. But attempts to specify the concept more rigorously have been fraught with disagreements. There are three main sources of these disagreements: different disciplines within the social sciences emphasize different bases of power (for example, wealth, status, knowledge, charisma, force and authority); different forms of power (such as influence, coercion and control); and different uses of power (such as individual or community ends, political ends and economic ends). Consequently, they emphasize different aspects of the concept, according to their theoretical and practical interests. Definitions of power have also been deeply implicated in debates in social and political theory on the essentially conflicting or consensual nature of social and political order. Further complications are introduced by the essentially messy nature of the term. It is not clear if power is a zero-sum concept (Mills 1956; Parsons 1960); if it refers to a property of an agent (or system), or to a relationship between agents (or systems) (Arendt 1970; Lukes 1974; Parsons 1963); if it can be a potential or a resource (Barry 1976; Wrong 1979); if it is reflexive or irreflexive, transitive or intransitive (Cartwright 1959); nor is it clear if power can only describe a property of, or relationship between, individual agents, or if it can be used to describe systems, structures or institutions (Lukes 1978; Parsons 1963; Poulantzas 1979); furthermore, it is not clear whether power necessarily rests on coercion (Cartwright 1959) or if it can equally rest on shared values and beliefs (Beetham 1991; Giddens 1977; Parsons 1963). Nor is it at all clear that such disputes can be rationally resolved, since it has been argued that power is a theory-dependent term and that there are few, if any, convincing metatheoretical grounds for resolving disputes between competing theoretical paradigms (Gray 1983; Lukes 1974; Morriss 1987).
In the 1950s discussions of power were dominated by the conflicting perspectives offered by power-elite theories (Mills 1956), which stressed power as a form of domination (Weber 1978 [1922]) exercised by one group over another in the presence of fundamental conflicts of interests; and structural-functionalism (Parsons) which saw power as the ‘generalized capacity of a social system to get things done in the interests of collective goals’ (Parsons 1960). Parsons thus emphasized power as a systems property, as a capacity to achieve ends; whereas Mills viewed power as a relationship in which one side prevailed over the other. Mills’s views were also attacked by pluralists, who argued that he assumed that some group necessarily dominates a community; rather, they argued, power is exercised by voluntary groups representing coalitions of interests which are often united for a single issue and which vary considerably in their permanence (Dahl 1957; 1961; Polsby 1963). Against class and elite theorists the pluralists posed a view of US society as ‘fragmented into congeries of small special-interest groups with incompletely overlapping memberships, widely differing power bases, and a multitude of techniques for exercising influence on decisions salient to them’ (Polsby 1963). Their perspective was rooted in a commitment to the study of observable decision making, in that it rejected talk of power in relation to non-decisions (Merelman 1968; Wolfinger 1971), the mobilization of bias, or to such disputable entities as ‘real interests’. It was precisely this focus on observable decision making which was criticized by neo-elite and conflict theorists (Bachrach and Baratz 1970; Connolly 1974; Lukes 1974), who accused the pluralists of failing to recognize that conflict is frequently managed in such a way that public decision-making processes mask the real struggles and exercises of power; both the selection and formulation of issues for public debate and the mobilization of bias within the community should be recognized as involving power. Lukes (1974) further extended the analysis of covert exercises of power to include cases where A affects B contrary to B’s real interests—where B’s interests may not be obtainable in the form of held preferences, but where they can be stated in terms of the preferences B would hold in a situation where B exercises autonomous judgement. Radical theorists of power have also engaged with structural-Marxist accounts of class power over questions of whether it makes sense to talk of power without reference to agency (Lukes 1974; Poulantzas 1973; 1979). Although these debates have rather dominated discussions of power in social and political theory, we should not ignore the work on power in exchange and rational-choice theory (Barry 1976), nor the further criticisms of stratification theories of power which have been developed from positions as diverse as Luhmann’s neofunctionalism (Luhmann 1979), and Foucault’s rather elusive post-structuralism (Foucault 1980; but see Foucault 1982).
Definitional problems seem to be endemic to discussions of power. One major problem is that all accounts of power have to take a stand on whether power is exercised over B whether or not the respect in which B suffers is intended by A. Similar problems concern whether power is properly restricted to a particular sort of effect which A has on B, or whether it applies in any case in which A has some effect on B. These two elements, intentionality and the significance of effects, allow us to identify four basic views on power and to reveal some of the principal tensions in the concept (White 1972).
The first view makes no distinction between As intended and unintended effects on B; nor does it restrict the term power to a particular set of effects which A has on B. Power thus covers phenomena as diverse as force, influence, education, socialization and ideology. Failing to distinguish a set of significant effects means that power does not identify a specific range of especially important ways in which A is causally responsible for changes in B’s environment, experience or behaviour. This view is pessimistically neutral in that it characteristically assumes that power is an ineradicable feature of all social relations, while it makes no presumption that being affected by others in one way or in one area of life is any more significant than being affected in any other. One plausible version of this view is to see power as the medium through which the social world is produced and reproduced, and where power is not simply a repressive force, but is also productive (Foucault 1980). Note that with this conception there is no requirement that A could have behaved otherwise. Although this is an odd perspective, it is not incoherent, as it simply uses power to refer to causality in social and interpersonal relations.
The second view isolates a set of significant effects. Thus, A exercises power over B when A affects B in a manner contrary to B’s preferences, interests, needs, and so on. However, there is no requirement that A affect B intentionally, nor that A could have foreseen the effect on B (and thus be said to have obliquely intended it). Poulantzas’s Marxism provides one such view by seeing power in terms of ‘the capacity of a social class to realize its specific objective interests’ (Poulantzas 1973). Any intentional connotations are eradicated by his insistence that this capacity is determined by structural factors. The capacity of a class to realize its interests ‘depends on the struggle of another class, depends thereby on the structures of a social formation, in so far as they delimit the field of class practices’ (Poulantzas 1973). As agency slips out of the picture, so too does any idea of A intentionally affecting B. Although idiosyncratic, this view does tackle the problem of whether we can talk meaningfully of collectivities exercising power. If we want to recognize the impact which the unintended consequences of one social group’s activities have over another, or if we want to recognize that some group systematically prospers while others do not, without attributing to the first group the intention of doing the others down, then we shall be pushed towards a view of power which is not restricted solely to those effects A intended or could have foreseen. The pressures against this restriction are evident in Lukes’s and Connolly’s work. Both accept unintended consequences so as to ‘capture conceptually some of the most subtle and oppressive ways in which the actions of some can contribute to the limits and troubles faced by others’ (Connolly 1974). Both writers, however, also recognize that attributions of power are also often attributions of responsibility, and that to allow unintended effects might involve abandoning the possibility of attributing to A responsibility for B’s disbenefits. Consequently both equivocate over how far unintended effects can be admitted, and they place weight on notions of A’s negligence with regard to B’s interests, and on counterfactual conditionals to the effect that A could have done otherwise. Stressing ‘significant effects’ also raises problems, as the criteria for identifying such effects are hotly disputed. Thus radical theorists criticize pluralists for specifying effects in terms of overridden policy preferences, on the grounds that power is also used to shape or suppress the formation of preferences and the articulation of interests. Again, two pressures operate, in that it seems sociologically naïve to suppose that preferences are always autonomous, yet it is very difficult to identify appropriate criteria for distinguishing autonomous and heteronomous preferences. Taking expressed preferences allows us to work with clearly observable phenomena, since B can share the investigator’s ascription of power to A—it thus has the advantage of methodological simplicity and congruence with the dependent actor’s interpretation. However, taking ‘repressed’ preferences or real interests can be justified, since it provides a more theoretically persuasive account of the complexities of social life and of the multiple ways in which potential conflicts are prevented from erupting into crisis. Yet this more complex theoretical account is under pressure to identify a set of real interests, and the temptation is to identify them in terms of autonomous/rational preferences; the problem with this is that it often carries the underlying implication that power would not exist in a society in which all agents pursued their real interests. Power is thus used to describe our deviation from utopia (Gray 1983).
The second view is primarily concerned with identifying the victims of power—not the agents. The focus is on A’s power over B. The third view, which attributes power only when A intends to affect B, but which does not place any restrictions on the manner in which A affects B, switches the focus from A’s power over B to A’s power to achieve certain ends (Russell 1938; Wrong 1979). Power is concerned with the agent’s ability to bring about desired consequences—‘even’ but not necessarily ‘against the resistance of others’ (Weber 1978 [1922]). This view has a long pedigree (Hobbes 1651) and it satisfies some important theoretical interests. In so far as we are interested in using A’s power as a predictor of A’s behaviour, it is clearly in our interests to see A’s power in terms of A’s ability to secure high net profit from an action—the greater the anticipated profit, the more likely A is to act (Barry 1976). Another reason for focusing on A’s intention is the difficulty in identifying a range of significant effects which is not obviously stipulative. Concentrating on A’s intended outcomes allows us to acknowledge that there are a number of ways in which A can secure B’s compliance and thereby attain A’s ends. Thus force, persuasion, manipulation, influence, threats, throffers, offers, and even strategic positioning in decision procedures may all play a role in A’s ordering of A’s social world in a way that maximally secures A’s ends. But seeing power solely in terms of A’s intentions often degenerates into an analysis where all action is understood in power terms, with behaviour being tactical or strategic to the agent’s ends. On this view agents become, literally, game-players or actors, and we are left with a highly reductive account of social structures and institutions (Rogers 1980).
Finally, the fourth perspective analyses power in terms of both intentional action and significant effects. It concentrates on cases where A gets B to do something A wants which B would not otherwise do. Two sets of difficulties arise here. The first concerns the extensiveness of the concept of power and its relationship with its companion terms, authority, influence, manipulation, coercion and force. On some accounts power is a covering term for all these phenomena (Russell 1938); on others it refers to a distinct field of events (Bachrach and Baratz 1974). Getting B to do something that B would not otherwise do may involve mobilizing commitments or activating obligations, and it is common to refer to such compliance as secured through authority. We may also be able to get B to do something by changing B’s interpretation of a situation and of the possibilities open to B—using means ranging from influence and persuasion to manipulation. Or we may achieve our will through physical restraint, or force. Finally, we may use threats and throffers in order to secure B’s compliance—that is, we may coerce B (Nozick 1969). In each case A gets B to do something A wants which B would not otherwise do, although each uses different resources (agreements, information, strength, or the control of resources which B either wants or wants to avoid), and each evidences a different mode of compliance (consent, belief, physical compliance or rational choice). Although exchange and rational choice theorists have attempted to focus the analysis of power on the final group of cases, to claim that the others are not cases of power is clearly stipulative. Yet it is these other cases which introduce some of the pressures to move away from a focus solely on intended effects and significant affecting. Where A’s effect on B is intended, instrumental to A’s ends and contrary to B’s preferences, and where B complies to avoid threatened costs, we have a case which firmly ties together A’s intention and the set of effects identified as significant (B’s recognized costs are intended by A and functional to A’s objectives). But the other cases all invite extensions, either in the direction of covering cases in which A secures A’s will, disregarding the nature of the effects on B, or towards cases where B’s options or activities are curtailed by others, either unintentionally, or unconditionally. Also, this view of power risks focusing on A’s exercise of power over B, to the detriment of the alternative and less tautological view (Barry 1976) that power is a possession, that it may exist without being exercised, and that a crucial dimension of power is where A does not secure B’s compliance, but is in a position to do so should he choose. Wealth, status, and so on, are not forms of power, but they are resources which can be used by A to secure B’s compliance. An adequate understanding of power in a given society will include an account of any systematic inequalities and monopolies of such resources, whether they are being used to capacity or not. The pressure, once again, is against exclusive concentration on A’s actual exercise and towards a recognition of A’s potential. But once we make this step we are also likely to include cases of anticipatory surrender, and acknowledging these cases places further pressure on us to move beyond easily attributable, or even oblique, intention on A’s part. These pressures are resisted mainly by those who seek to construct a clear and rigorous, if stipulative, theoretical model of power. But there is also some equivocation from those who seek to match ascriptions of power with ascriptions of moral responsibility. Part of the radical edge of Lukes’s (1974) case stems precisely from the use of ascriptions of power as a basis for a moral critique. But much is problematic in this move. A may act intentionally without being sufficiently sane to be held morally responsible; A may intentionally affect B to B’s disbenefit without violating moral norms (as in a chess game, competitions, some exchange relations with asymmetrical results, and so on); it is also important to recognize that B’s compliance must maintain proportionality with A’s threat in order for B to be absolved of moral responsibility (Reeve 1982).
The theoretical and practical pressures which exist at the boundaries of these four possible interpretations of power account for much of the concept’s messiness. Each has its attractions. The fourth view is most promising for model or theory building, the third for the prediction and explanation of action, the second for the study of powerlessness and dependency, and the first for the neutral analysis of the strategic but non-intentional logic of social dynamics. Although metatheoretical grounds for arbitration between competing conceptions of power seem largely absent, we can make a few comments on this issue. Although restrictivist definitions of power may serve specific model and theory-building interests, they inevitably provide a much simplified analysis of social order and interaction. However, more encompassing definitions risk collapsing into confusion. Thus, while there are good theoretical grounds for moving beyond stated preferences to some notion of autonomous preferences—so as, for example, to give a fuller account of B’s dependence—we should be cautious about claiming that A is as morally responsible for B’s situation as when A intentionally disbenefits B. Indeed, depending on how we construe the relevant counterfactuals, we might deny that agents are liable for many of the effects of their actions. Thus, we might see social life as inevitably conflict ridden, and while we might recognize that some groups systematically lose out it might not be true that A (a member of the elite) intends to disadvantage any individual in particular, or that A could avoid harming B without allowing B to harm A (as in Hobbes’s state of nature). Also, although we are free to use several different definitions of power (such as the three dimensions identified by Lukes 1974), we should recognize that each definition satisfies different interests, produces different results and allows different conclusions, and we need to take great care to avoid confusing the results. Finally, we should recognize that although definitions of power are theorydependent, they can be criticized in terms of the coherence of the theory, its use of empirical data, and the plausibility of its commitments to positions in the philosophies of mind and action.
Mark Philp
University of Oxford
References
Arendt, H. (1970) On Violence, London.
Bachrach, P. and Baratz, M.S. (1970) Power and Poverty, Theory and Practice, Oxford.
Barry, B. (1976) ‘Power: an economic analysis’, in B.Barry (ed.) Power and Political Theory London.
Beetham, D. (1991) The Legitimacy of Power, New Brunswick, NJ.
Cartwright, D. (1959) ‘A field theoretical conception of power’, in D.Cartwright (ed.) Studies in Social Power, Ann Arbor, MI.
Connolly, W. (1974) The Terms of Political Discourse, Lexington, MA.
Dahl, R. (1957) ‘The concept of power’, Behavioral Science 2.
——(1961) Who Governs? Democracy and Power in an American City, New Haven, CT.
Foucault, M. (1980) Power/Knowledge, Brighton.
——(1982) ‘The subject and power’, in H.Dreyfus and P Rabinow (eds) Michel Foucault: Beyond Structuralism and Hermeneutics, Brighton.
Giddens, A. (1977) ‘“Power” in the writings of Talcott Parsons’, in Studies in Social and Political Theory, London.
Gray, J.N. (1983) ‘Political power, social theory, and essential contestability’, in D.Miller and L.Siedentop (eds) The Mature of Political Theory, Oxford.
Luhmann, N. (1979) Trust and Power, London.
Lukes, S. (1974) Power, A Radical View, London.
——(1978) ‘Power and authority’, in T.Bottomore and R. Nisbet (eds) A History of Sociological Analysis, London.
Merelman, R.M. (1968) ‘On the neo-elitist critique of community power’, American Political Science Review 62.
Mills, C.W. (1956) The Power Elite, London.
Morriss, P. (1987) Power: A Philosophical Analysis, Manchester.
Nozick, R. (1969) ‘Coercion’, in S.Morgenbesser et al. (eds) Philosophy, Science and Method, New York.
Parsons, T. (1960) ‘The distribution of power in American society’, in Structure and Process in Modern Societies, Glencoe, IL.
——(1963) ‘On the concept of political power’, Proceedings of the American Philosophical Society 107.
Polsby, N. (1963) Community Power and Political Theory, New Haven, CT.
Poulantzas, N. (1973) Political Power and Social Classes, London.
——(1979) State, Power, Socialism, London.
Reeve, A. (1982) ‘Power without responsibility’, Political Studies 3.
Rogers, M.F. (1980) ‘Goffman on power hierarchy and status’, in J.Ditton (ed.) The View from Goffman, London.
Russell, B. (1938) Power, London.
Weber, M. (1978 [1922]) Economy and Society, eds. G.Roth and G.Wittich, Berkeley, CA. (Original edn, Wirtschaft und Gesellschaft, Tübingen.)
White, D.M. (1972) ‘The problem of power’, British Journal of Political Science 2.
Wolfinger, R.E. (1971) ‘Nondecisions and the study of local polities’, American Political Science Review 65.
Wrong, D. (1979) Power: Its Forms, Bases and Uses, Oxford.
