You are on page 1of 18

European Journal of Political Theory http://ept.sagepub.

com/

Two dogmas of liberalism


Glen Newey European Journal of Political Theory 2010 9: 449 DOI: 10.1177/1474885110374007 The online version of this article can be found at: http://ept.sagepub.com/content/9/4/449

Published by:
http://www.sagepublications.com

Additional services and information for European Journal of Political Theory can be found at: Email Alerts: http://ept.sagepub.com/cgi/alerts Subscriptions: http://ept.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://ept.sagepub.com/content/9/4/449.refs.html

>> Version of Record - Oct 22, 2010 What is This?

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Article

EJPT
European Journal of Political Theory 9(4) 449465 ! The Author(s) 2010 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1474885110374007 ept.sagepub.com

Two dogmas of liberalism


Glen Newey
University of Keele

Abstract This article is on political normativity. It urges scepticism about attempts to reduce political normativity to morality. Modern liberalism leaves a question about how far morality can be accommodated by the form of normativity characteristic of politics. The article casts doubt on whether individual moral norms carry over to collective, for example, political, action, and whether the former trump other kinds of reasons in politics. It then sketches an alternative view of politics as an irreducibly collective enterprise. Reasons for acting politically, including the understandings on which perceptions of legitimacy rest, are largely artefacts of the political culture and thus only marginally subject to generic conditions of validity: this is true in particular of liberal acceptability-conditions. Thus legitimacy, though not a redundant notion, must be geared to local political norms. Keywords liberalism, morality, normativity, political design, politics

Introduction
This article examines political normativity and its limits. It does not deal at all with the scope, and only tangentially with the source, of that normativity. Its main focus is on the type of normativity which is appropriate within politics, and, by extension, which is the appropriate object of philosophical theorizing about politics. In fact, the focus is narrower still, since I deal only with certain aspects of the typology. I do not have much to say, for instance, about the notion of bindingness, except insofar as that raises the question about the possible role of morality in grounding politics.1 I argue, particularly with respect to Rawlss work, that certain features of modern liberalism leave a question about how far morality can be accommodated politically. I urge scepticism about attempts to reduce political normativity to morality, in particular via the philosophically cohesive but also morally very demanding moral theory of Kant, particularly the notion that morality trumps reasons of all other kinds. I shall question how far Kantianism can show that moral reasons have this trumping quality, as a demand of pure practical
Corresponding author: Glen Newey, University of Keele, Keele, Staffs, ST5 5BG, UK Email: g.f.newey@keele.ac.uk

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

450

European Journal of Political Theory 9(4)

rationality. I also cast doubt on an assumption often tacitly mobilized in support of Kantian reductivism, namely that the norms which justify an individual in acting in a certain way carry over to the sphere of collective action. This does not however demand a wholesale abandonment of Kantianism, in attempting to construct a serviceable notion of political normativity. Then the argument moves to the second dogma of liberalism, namely the attempt to reduce collective to individual rationality. The main point here is that, even were it the case that at the level of individual action there were an argument to show that morality overrides other kinds of reasons for action because it is a demand of pure practical rationality, this would not warrant the conclusion that political reasons for action can be inferred from this demand. Politics is a sphere of collective action, and the rational constraints which bear on collective action relate to what can coherently be authorized by those on whose behalf the collective is taken to act.2 Towards the end, I sketch an alternative view of politics. This will assume that politics is a collective enterprise. It also takes the presumption of authorization as a basis for political legitimacy. But it does not presuppose an originative account of that authorization or, consequently, political legitimacy, and does not attempt to establish a pre-existing order of reasons by reference to which legitimacy is to be understood. Reasons for acting politically, which include the understandings on which perceptions of legitimacy rest, are as much a part of the local political culture as are, say, political institutions. This dims the prospects for theories which seek to lay down foundations for politics using reasons with purportedly universal domain. There is however no special reason to greet the specicity of political reasons with dismay.

Liberalism and moralism


Rawls famously remarked at the start of Theory of Justice that justice was the rst virtue of societies. By this, it becomes clear, Rawls means that the overriding demand to be made of a social and political system is that it conforms to certain moral requirements, whose content is generated by a philosophical theory: the very one which Rawls unfolds after making that remark. In this section I wish to question whether it is reasonable to impose moralized demands on political arrangements at the most fundamental level, that of the basic structure. Rawls exemplies philosophical liberalisms project of political design,3 namely devising a blueprint for politics, including designing institutions, procedures, entitlements and so on. The philosophical task is taken to be the justication of a particular conception of political design. The project of justication is often seen as giving moral reasons why the favoured conception would be adopted, for example because it is the only one consistent with persons moral standing as free and equal agents. The assumptions behind liberal moralism are often not elucidated. But frequently the idea seems to be that morality trumps other practical reasons, so political design must t a moral template.4 It would be over-simplied to suggest that the source of this idea lies wholly within Kantian moral theory, but Kant, whose inuence Rawls

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

451

of course acknowledged,5 provided the most cohesive and systematic statement of the trumping view. Indeed it is quite hard to think of a philosophical argument for this conclusion which is both clearly non-Kantian and not clearly inadequate. The liberal moralist approach to political design could be expounded using the following argument. 1. Morality consists (at least) of obligations. 2. (Moral) obligations trump other kinds of reason for action. So 3. In any practical project, moral obligations trump other kinds of reason for action. 4. Political design is a practical project. So 5. In the project of political design, moral obligations trump other kinds of reason for action. This is obviously very schematic, and various qualications need to be entered. For one thing, although talk of obligations makes it more perspicuous, 1 and 2 could be replaced with the claim that moral reasons trump other kinds of reason for action, the rest of the argument then being couched in terms of moral reasons generally. Second, it makes an important dierence at what stage in the process morality is thought of as coming in. How far an option may be thought binding may depend, for instance, on what else has been rejected, how and on what grounds. Relatedly, how a norm binds may be seen as an artefact of the process, rather than one of its initial conditions. Liberalism lls in the moral reasons which apply to political design, such as those arising from persons status as free and equal. In what sense, however, can it be said that the obligations, or other trumping moral reasons, apply to the project? The most obvious reply is that the obligations always apply to everyone, or to all rational beings, etc., so they apply in the circumstances of political design. This, however, does not dispose of the query, which seeks an explanation of why the obligations, etc., are salient in the circumstances imagined. If they are salient always and everywhere, any specication of these circumstances seems redundant. On the other hand, if the circumstances matter, then one can ask what justies this starting-point, given the wider project. Most practitioners of political design opt for a measure of circumstantial specicity. Notably, Rawls begins A Theory of Justice with a description of the Original Position, while many 17th-century contract theorists began from an account of the state of nature in which certain norms, those of natural law, apply. That the set-up envisaged people as occupying some ur-situation, from

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

452

European Journal of Political Theory 9(4)

which they had to nd their way to politics, clearly inuenced the conclusions drawn from it, and that was part of the point. The initial set-up serves its justicatory purpose only if it is compelling in itself, since a dierent starting-point may yield dierent conclusions. In Rawls, the set-up is thought compelling because the Original Position encapsulates an idea of justice as fairness. Unlike classical stateof-nature theory, it starts in medias res, assuming that society already exists as a scheme for mutual benet; the question is how to share the goods accruing from cooperation. One can imagine social and political institutions dissolved, take some moral claims as given and make human beings as motivationally plastic as is needed to get them to act as the theory demands. But nothing compels one to begin just there. It is not obvious why the balance between the given and the counterfactually revisable should be struck this way. Someone may say that that morality obviously trumps other kinds of reason for action. This is not an argument, but a contention in need of validating premises. The contention can be made stipulatively true, by identifying morality with reasons which come out top in practical deliberation. That leaves the eld open regarding any independently specied class of reasons for action. The claim gains philosophical shape if it is held, for instance, that there is one kind of thing with unqualied or supreme value, such as subjective internal states of human beings; so the most important task is to realize this value, in which morality guides us. There is the question whether that axiological claim, or the ethical naturalism which seemingly underlies it, is right. But even if they are, it is not obvious how these thoughts guide political design; nor how they mould politics to a distinctively liberal shape.6 I shall not seek to refute the trumping thesis directly. My aim, more narrowly, is to suggest that modern liberalisms own commitments make it dicult to use the thesis in the project of political design. Some of the reasons for this, as I shall suggest, derive from specic commitments of modern liberalism. In addition, morality does not present itself in a form which is usable straight o. Of course, even (or especially) constructivism has to start with some raw materials. But it is not clearly more compelling to take morality as given, and resolve for politics, than to take politics as given, and resolve for morality. Consider, for instance, the notion of the public interest. Little, if anything, is simply given as constitutive of the public interest. A functional theory might say the public interest must at least include whatever the constitutive ends of political association are, since that is a condition of there being a public in the rst place. But whatever was disputable about the concept of the public interest may well remain so when the discussion turns to the ends of association. Of course, one may say that the public interest demands the upholding of certain moral norms. But why treat such a judgement as authoritative? Moral judgements can serve as inputs into deciding the public interest. But it does not follow that these judgements should commandeer the whole process, either by ltering out certain substantive judgements at the start, or determining how the process unfolds.

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

453

What, if anything, is the basis of the normative authority of political procedures?7 The question arises, for instance, with democratic theories which hold that political decisions have authority because they express the collective will of the citizens to bind themselves by them. Can the procedures, or their outcomes, contradict the basis of political authority, given that the people confers that authority, and that its status as an agent seems to require the procedures? Behind this lies a more general question: how does the normativity of political authority relate to that of the justication on which it is held to rest? There seem to be two possibilities. Either the normativity of political authority is sui generis, with respect, in particular, to that of morality, or it is not. If it is, then there is no reason why political authority, expressed in the bindingness of its procedures outcomes, should conform to the dictates of independent moral norms. Political authority and morality are disparate sources of normativity. Insofar as it makes sense to talk of the bindingness of decisions made by political authority, this will not be because it is a species of morality. Dierent judgements can then be made about what happens when there is apparent conict between the two sources of normativity. It can be argued that one source must trump8 the other in such apparent conicts; or that each has its own discrete sphere of operation; or that there is some more complex negotiation to be made between morality and political authority as sources of normativity when they seem to be in conict. On the other hand, if political normativity is a species of moral normativity, then whatever binding force it has will derive from morality. So, when the political procedures throw up results seemingly at odds with morality, there is a prima facie conict between two manifestations of moral normativity: one issuing from the justication of the procedures, and the other from whatever justies the moral objection to their outcome. But this is a conict within moral normativity, not between political normativity and it. Rawls holds that no political procedure could guarantee morally indefeasible outcomes, assuming an independent standard of moral assessment exists. Rawls argues, accordingly, that politics embodies imperfect rather than perfect procedural justice.9 However, there is no politically authoritative standard by which to ratify political outcomes. Any situation where the outcomes authority is at issue must also put in question the authority of objections to them. Liberals endorsement of reasonable disagreement about ideals of the good makes such objections likelier.10 Alternatively, politics embodies pure procedural justice:11 political processes generate the norms content and binding force. If so, a form of decisionism results: the content of norms is valid just because the procedures yield them as outcomes. Moral ideals can enter into formulating the procedures and in choosing outcomes. But any moral residues left by the procedures will not be correctible by an authoritative independent standard. Pure procedural justice allows for no such standards, so the notion that they could nd political outcomes wanting for instance, by aronting an individuals moral intuitions must be illusory. The upshot is procedural decisionism.

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

454

European Journal of Political Theory 9(4)

Collective action and individual agents


Politics involves collective agency. Here, I am using the term collective agency generically. For a particular action-token to instantiate collective agency, the action must be ascribable to each of n individuals, where n is greater than 1; where is ascribable means roughly that it can be attributed to the individuals concerned on a basis which makes true the usual act-descriptions, such as holding responsible, imputing relevant intentions, beliefs, etc. The primordial political question is What do we do? though the reference of we need not be well-dened. This account is intended to deal with corporate rather than merely cooperative agency. Uncontroversially, agents often cooperate, as when they join forces to push-start a car. More controversially, agency is ascribed to supra-individual bodies like governments, civil-society associations, etc., at which methodological individualists may balk. For irreducibly corporate agency the relevant action, while meeting the conditions for collective agency, must not be truth-functionally reducible to propositions about individual agents.12 The question is whether liberalisms standard approach, justifying political design by reecting on what individuals have good reason to do, makes sense of politics as collective agency. There is no obvious general way in which to construct collective agency from individual reasons for action. This applies even in rational choice theory, which has aimed to reduce collective to individual agency. For example, if maximizing payos is dened as rational in respect of an agent and a choice situation, then in the standard one-shot Prisoners Dilemma the sector <Defect, Defect> is a Nash equilibrium: Defect is strictly dominant for each player. Then, given that rationality is dened as maximizing payos relative to an agent and choice situation, <Defect, Defect> emerges as rational for each. In the PD matrix the Nash equilibrium is Pareto-suboptimal with respect to a non-Nash equilibrium (<Cooperate, Cooperate>). So, the latters aggregate utility exceeds that of <Defect, Defect>. It is unclear however why one cannot dene a corporate agent which will rationally pursue utility-maximization, on the denition of rationality already given. Of course, if rational collective action is identied with the aggregated rational actions of each individual, then the players each rationally defect. But what compels one to say that? To assume that that is what must be done seems simply to beg the question as to how individual and collective rationality relate to each other. The conclusion that <Defect, Defect> is a Nash equilibrium follows only given the revealed preferences of individual agents. Discussion sometimes confuses the quantitative measures arising from the actions available to the players with their associated utilities. If the payos are dened as utilities, it follows trivially, on the assumptions already made, that the rational course of action for each player will result in <Defect, Defect>. But it does not follow that in any description of the situation where the utilities are not specied, defection must be the rational action. For example, it doesnt follow that players conned as described in the standard exposition, even with the ordinal rankings of (say) penal taris, must rationally

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

455

choose Defect. That depends on the associated utilities. Access to the utilities is via revealed preferences, and the preferences are revealed by the agents actions. So the collectively rational action cannot be elaborated from a prior specication of individual utilities. It is not my aim to suggest a way out of such problems, such as some nonaggregative decision-procedure. Intuitively compelling constraints on rationality at the individual level may be be breached at the collective level if one follows a procedure which aims to construct decisions from given individual preferences. Of course, this does not show that one cannot treat a collectivity, once a decision has been arrived at, as an individual writ large, as regards the rational constraints to which such a body is subject. For instance, it is no doubt still an intuitively compelling constraint on the preference-orderings of a collective that they avoid violating acyclicality. But that applies once the ordering has been determined. It does not show how to arrive at the ordering without violating such constraints. There are various ways one might construct politics from individual actions. Perhaps the most obvious is simple analogy: whatever an individual agent has overriding all-in reason to do, in circumstances C, is what the collective agent has overriding, etc., reason to do in C. As it stands, this is far too simple. First, certain kinds of action-description cannot be appropriately applied to individual rather than corporate agents, at least in non-traditional societies, so some political actions are not covered by the supposed analogy. Moreover, the notion of given circumstances needs elaboration. The circumstantial description risks begging the question, by making an analogy look compelling which from a broader perspective is arbitrary. It is unclear why what individuals have overriding, etc., reason to do, such as acting on a moral obligation, should be dispositive for political agencies like the state. The method suers the failing of all analogical argument: it is inductive, and so begs the question against the sceptic. Such arguments infer from a set of analogical properties between A and A* that a further property of A must also belong to A*; but this inference is powerless against someone who denies that the analogy holds for the disputed property. Political design usually seeks to construct politics out of individual reasons for action. Since it usually draws no analogy between individual and collective reasons, it avoids the problems facing the argument from analogy. Again, however, its exponents need to show why a (usually hypothetical) description of individuals reasons for action supports inferences about collective action. The initial description may be arbitrary when used in its intended justicatory role. Such, for instance, is the nub of Rousseaus objection in the second Discourse to Hobbess description of the state of nature: it imagines as given traits which are not native to human beings, but artefacts of socialization.13 Finally the problem is that there are no compelling grounds for thinking that collective agency, now or at the notional point of its origin, can be constructed out of individual reasons for action. The fact, if it is one, that a corporate body like the state can be the subject of action and associated ascriptions such as belief, reasons, intention, and so on, shows only that some analogy can be drawn between

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

456

European Journal of Political Theory 9(4)

individual and corporate agents. That does not mean that action-ascriptions to the latter must be reducible to those of individuals. It may be hard to prise away in thought the state from the individuals whom it represents, particularly when questions of justication slip so easily into foundational ones. But one can think of the state as being simply there, a fact of social nature.

Kant again
Liberal political design owes its inspiration above all to Kant. For Kantians, the requirement to act morally follows from pure practical rationality. They have oered several explanations as to why Kant thought this; that someone whose maxim fails, say, the rst formulation of the categorical imperative, stands convicted of a contradiction in the will, so that the proposed course of action is one to which no rational agent could assent, or in the maxims conception. In Kants own writings, of course, the demands of practical rationality are thought of as applying paradigmatically, or even solely, to individual agents. But, to repeat, politics involves collective agency. The question is whether demands appropriate to individual action must apply at the collective level. Collective action requires authorization. In fact, authorization goes all the way down through agency. It was an insight of Hobbess mature political theory that agents, whether individual human beings or corporate bodies, can be seen as a compound of two elements, the author and the authorized.14 In the case of natural persons, these elements are combined in a single individual. The repertoire of ascribing blame, praise, responsibility, and so on, follows from seeing the individual as self-authorizing: as at once author and authorized. Freedom is in fact a condition of regarding the agent, qua natural person, as being such, i.e. someone whose acts are self-authorized. Only someone who is free and therefore self-authorizing will bear the ascriptions necessary for agency, as an originator of action. For Hobbes, individual agency merely instantiates a more general relation of authorization. A further instance of this relation, as Hobbes noted, is the authorisation of one person by another, i.e. articial personation.15 For one person to authorize another to act on her behalf, she must be thought free as regards authorization itself. No doubt in some cases, such as mental incapacity, it may be necessary to authorize someone on a persons behalf. But insofar as this action is undertaken for the incapacitated person, it is only through ction that it can be represented as authorized by her; just as with inanimate objects, such as bridges, it is only by ction that these things count as authorizing the acts done in their name. Corporate action requires authorization. A single ctional person represents the collective by the latters consent. They have to be thought free regarding the act of authorization itself, as a condition of their incorporation, though there need be no originating act of will which eects this incorporation.16 But individuals must identify themselves as a part of the corporate body. In the political case, there is no reason to think that all citizens will do this, so to the extent that such an act is

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

457

deemed necessary for political obligations, not everyone within a states jurisdiction will be so obligated.17 In Kantian terms, a citizen the determining ground of whose political will lies outside himself must act heteronomously (or not act at all). The very possibility of politics, as a eld of corporate action which embraces individual wills, demands that the individual be self-determining as regards the incorporation of his will into that of the collective. Since it involves authorization, corporate action is a form of prosthetic agency. The prosthesis consists in a relation between an author, and an agent acting on the authors behalf. While this relation often is one-to-one, in politics it is typically many-to-one. For instance, in democratic theory, the author is usually regarded as one collective, the demos, while the agent is another, the government, acting (within limits) on the authors say-so. Admittedly, each of these collectives may sometimes be viewed as a corporate individual, such as in civil litigation. But even here, it is not obvious that rational constraints on action duplicate those of the selfauthorizing agent. The state, and politics, confront us in their facticity, as extant features of the social landscape. In what sense does one freely assent to coercion by the state? The point is not that assent in this environment is elicited by the very power which it endorses. Nor is it that the facticity of the state alters individual actors feasible set of actions, and their rankings of them. It is that the citizen comes already incorporated, and in adopting the ction that this body resulted from an originative act of authorization, one moves beyond the rational constraints which bear upon individual action. One might think that since for Kantians the legislator is the paradigm of individual moral agency, there is no disanalogy between individual and corporate agency. But the universality of legislation need not apply exceptionlessly or derive from an abstract paradigm of individual rational agency. In the rst instance law presents itself to those on whom it is incident as a manifestation of will, of which citizens may or may not regard themselves as authors. To whatever constraints, such as those arising from pure practical rationality, such a will may be subject, its status as practically rational will depend only on what it can consistently conceive of and will to enact. What can be rationally authorized by citizens, as incorporated agents, can be inferred back from what the state can rationally do. An obvious example involves the use of force. On Webers well-known denition, states are agencies claiming a monopoly of legitimate force (Gewalt). Now, one can say that any situation involving the use of force by one agent on another is necessarily one in which consent is withheld, or at least not granted (as in cases where consent is not overridden, but bypassed). If an agent acts in a certain way only because force is applied, she must withhold her consent at that time. If force always overrides consent, it will override the consent of rational agents. A rational agent cannot consent to be forced: this forms the basis of Robert Paul Wols Kantian case for philosophical anarchism.18 But most Kantians are not anarchists. For them, presumably, the impossibility of enforced consent will not make

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

458

European Journal of Political Theory 9(4)

illegitimate all uses of force by the state, even though in the nature of force someone to whom it is applied cannot, at that time, consent to it. One cannot consent to the states achieving its ends by this means. One can argue, for instance, that the authorization model can be applied in this instance. If the state coerces me, I necessarily do not consent to this treatment. But this hardly precludes my authorizing coercive treatment of me in advance, as indeed voluntarist accounts of political obligation envisage. To do so requires a contradiction neither in conception, nor in the will: its very intelligibility shows that authorized coercion involves no contradiction in conception. Nor is there any volitional contradiction in directing that in certain situations, ones will may be overridden or ignored. Someone may say that there is a contradiction between my will now and my will then: I will now that I be coerced in conditions which include my refusal to act or refrain from acting in a certain way, but in those conditions my will must oppose this. But even here there is no contradiction either in the will present or in the will future. The validating ground of authorization lies in the freedom of self-determination. The will can be imagined as making pure practical rationality its determining ground. But it is far from clear that this constrains action in the same way as the individual moral legislator. One has to regard oneself as free in relation to acts of authorization as a condition of being incorporated with others. A practical constraint on action could arise if one thought that certain things contradicted the possibility-conditions of authorization. But this does not seem to preclude, for example, voluntary enslavement. The problem, again, is that no contradiction arises in binding ones future will by ones present will. That is what contracts do. Perhaps one could argue that a contradiction would arise with the possibility-conditions if such acts were adopted universally, since there must be at least one, an enslaver, whose will remains free, for there to be an enslaved. This assumes that any consistently willed authorization must be based on a universalizable maxim. But even here the maxim seems consistently universalizable, given that it can be made conditional on an act of will. As a further example, consider lying, which Kant saw as an important test of the working of the categorical imperative. Commentators oer dierent explanations of why the theory rejects maxims to lie. It is said, for instance, that there is a contradiction in the would-be liars will, because the project of lying wills the inculcation of falsehood while willing belief in the very institution of truth-telling which the lie subverts. Or it is said that the liars conception of her act is contradictory because she relies on the dupes capacity for rational consent to her purposes while using his agency in a matter to which he cannot consent. Again it seems clear that no contradiction arises if I authorize an agent to lie to me, for example, in certain circumstances. Someone may suggest that if I recognize that these obtain I will then be forewarned against the lie, and this is a kind of contradiction. But no contradiction arises if I instruct you to lie only when you know that I do not recognize this. In cases of prosthetic agency, then, it seems that the contradiction in the will or in conception which is held to be ruled out by practical rationality does not arise.

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

459

The explanation is fairly obvious if contradiction is located in the will: authorization is a mode of willing, and there is no contradiction between willing Z at one point and willing not-Z at another, or between willing the authorization and including, as part of what is authorized, that ones later will may be overridden in certain situations. Equally, however, no contradiction in conception arises either, just because one can conceive without contradiction of willing contents which Kant took as paradigmatic of practical impossibility. The conditions of authorized agency are those of free agency itself. As a result, the limits on how an individual can authorize others to act on his behalf are broadly those of contract, seen as an undertaking by free individuals to determine their future acts on the basis of mutual agreement. A knowledge condition will apply at this point, of course, as with ordinary commercial contracts. But it will not apply very robustly. First, one can ask what a free contracting party would expect to know as a condition of corporate agency. Then there are questions about knowledge itself: who knows what, and what distinguishes knowledge, for these purposes, from mere opinion. These questions are highly politicized. The prospects for an apolitical specication of antecedent epistemic conditions for valid contracting are, accordingly, dim. That provides reason for doubting that the foundational enterprise of contract theory can succeed. But this not mean that, in thinking about their political agency, citizens can dispense with the notion of freedom.

Politics and holism


In Two Dogmas of Empiricism, to which this paper owes its title, W. V. O. Quine advocates a form of epistemic holism.19 I suggest that a form of political holism can be asserted in place of the two liberal dogmas identied: morality as trumps and corporate-to-individual reductivism. That holism rejects strong foundationalism, as a set of (say, moral) norms, resting on intuitively compelling premises, which determine basic political and civil rights and obligations, the design of key institutions and procedures, and the distribution of basic goods. It also questions general claims to political authority constructed from individual reasons for action. Above all, it takes seriously the idea that politics can generate reasons for action with normative force. I take the following statements to be true, indeed truisms, of political life. Politics is a form of collective action. The basic political question is What do we do? for some imputed but not necessarily determinate we. The holism applies to the habits of thought and action which mark recognizably political life. There is good reason to avoid dening the political here, not least because disputes over the extension of politics are themselves irreducibly political:20 attempts to x that extension raise the very question they aim to settle.21 Accordingly, the holism aims not to delimit the political, for example via philosophical construction. Someone may say that if there is no antecedent denition of the political, there is no reason why politics should not be constructed philosophically, for example on moral foundations.22 That will lack, however, the authority to forestall further

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

460

European Journal of Political Theory 9(4)

disputes about whether, say, politics should rest on such foundations. The project of political design, except as a provisional contribution to the very process which it aims to conclude, is chimerical. The holism does not work at a high level of determinacy. The justicatory resources available to political actors need not enable them to specify outcomes very closely, and reasons given for outcomes can be challenged more the more outcomes are dened. But even at a general level resources may be limited. Take, for instance, Dworkins claim that liberty is subsumed in equality, since citizens only have a claim in liberty regarding some political outcome if the claim does not contravene equality.23 If Dworkin is right, many political objections to egalitarian policy are, straightforwardly, wrong. But it remains possible that a theory with this consequence is itself wrong, when set against the judgements with which it is inconsistent. The holism implies that the implications of theory can be used contrapositively against it. How far this is true in any particular case will vary, among other things, with the intuitive compellingness of the premises. Someone may ask whether, on this view, there are any normative considerations at all which bear on collective action. This question has two answers. First, constraints on collective action will be generated collectively. What considerations are normatively salient is decided by political processes. This again limits the contribution of philosophy. There is no obvious reason why considerations can only be salient in virtue of their binding quality. Any theory which gives bindingness special prominence will be tabling matters for consideration, like any other political claim. Of course, anyone can think of constraints which they dislike, such as gag rules. But nothing saves politics from going badly. As already argued, there is no credible prospect that appeals to the public interest will guide action very closely, without its being determined collectively what the public interest is. In any concrete political situation, such as in taking the decision whether to go to war, people will oer dierent assessments of where the public interest lies, and the issue will then be determined by political means, such as by majority vote. This is not, however, to say that the public interest must coincide with the upshot of the political process. Nonetheless, there is nothing which in general can determine the public interest more reliably than do political processes themselves, although specialist knowledge will clearly contribute to making specic decisions. The second, more general, response to the question about normative constraints interprets it as about the source of political normativity. This is not a question which gets asked much politically. I believe that the only ultimate source it can have is as an object of agreement which is thought of as free. But this needs to be correctly understood. That the source of normativity lies in free agreement follows from the conditions on collective action. It is necessary, for me to incorporate my will in a collective, that I regard its will as my own. That is, I have to understand the basis for the collectives will as one in relation to which I can think of myself as being autonomous, or self-determining. In regarding myself and my relation to the collective in this way, I think of myself as free in relation to it, as an author of its actions.24

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

461

The justicatory resources which this yields are rather limited. They fall well short of what theorists have hoped for from contract-based and other voluntarist theories. This is implicit in the critical argument made so far. A problematic move made by these theories is to posit agreement as necessary for incorporating ones will, but then to nesse the notion of agreement to encompass what individuals would agree to if, for example, they were fully rational. But this amounts to saying that they should agree, since the normative force of the theory lies in its telling people what their reasons for action are here and now. But then the notion of agreement is an idle wheel, since the theory mobilizes a strongly normative account of reasons for action.25 The conception of agreement in the present holistic account is post factum, or not essentially ante factum. It accepts that political reasons, such as reasons for agreement, are often artefacts of the political process itself. As such they are suspect to foundationalist liberals and exponents of Ideologiekritik alike. But it is not obvious why reasons for action have to be regimented into an abstract general theory. The reasons which people do accept are variously the product of domination, misinformation, etc., as well as the predictable eects of egoism or limited altruism. This follows the guiding idea that reasons commanding peoples agreement, as called for by standard notions of legitimacy, cannot operate at any very great distance from the reasons which they accept. It follows, given the heterogeneity of modern societies, that the reasons themselves will be diverse. If so, the prospects are dim for political theories which assign specic reasons for agreement to individuals, and base the justication for political authority on that. Of course, the phrase cannot operate at any very great distance is vague. It is meant to allow adjusting for cognitive error in framing the reasons one has for acting. It will immediately be said that it is often unclear whether some attitude expresses cognitive error or malfunction. That is true, and is a prime reason why the formulation has to remain vague. For example, should liberals see some womens acceptance of practices like cliterodectomy as invalid? There is no reason to think that questions like this can be decided pre-politically by a prescriptive account of justied belief. The holism allows a signicant role to political power. It is indeterminate what invalidates acceptance of a regime. For example, does it matter that power-relations inuence a persons normative beliefs? People often seem more attached to their beliefs than to a particular method of arriving at them. Attempts to free people by relieving them of dubiously acquired normative beliefs will have familiar perverse eects. It is not just that we lack a credible general account of the justiable use of power in shaping belief, or that it is hard to tell when power has played a causal role in generating unjustied beliefs. Even if we had these things to hand, it remains questionable whether beliefs acquired unjustiably must invalidate the acceptance of some norm by a person who holds them. At least two ripostes may be made at this point. First, what people will accept is strongly inuenced by the penalties for non-compliance. Unless might indeed reduces to right, surely an agency which uses coercion, such as the state, cannot

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

462

European Journal of Political Theory 9(4)

be legitimate if people accept it only because of that coercion. This may seem like cosa nostras oer you cant refuse. But there are important dierences. For instance, in the case of the Maa or protection rackets, the victim is essentially blackmailed: the threat promotes, in the list of ones preferences, an option which, but for the associated threat, would not come top. But an executively competent state puts everyone in the same boat as regards coercion, and this aects what it is rational for individuals to do. Then the choice for the individual is not between Maa-style coercion superadded to the extant coercion of the state, but between coercion of everyone, on the one hand, and on the other, no monopolistic coercive agency, where one is prey to the random coercion of private individuals. So the fact that acceptance is made under coercion need not make it illegitimate. It may be part of a coercive system which one prefers to any achievable alternative. It may be said that people would not or should not accept the system, even if the alternatives are presented in such a way. But that will then be an argument about the conditions under which coercion is legitimate in general. The argument is not ended by noting that acceptance would have been lacking without coercion. A second objection urges that the disposition to accept a justication may result, not from the immediate threat of coercion, but from power. Williams suggests26 that legitimacy can be vetted by means of a Critical Theory Test, which states that a justication fails if someones acceptance of it is itself produced by the coercive power supposedly justied by it. One trouble27 with this test is that it is hard to spell it out without resorting to counterfactuals whose truth-conditions are indeterminate. Suppose, for example, the counterfactual is understood as follows: in no possible world where the coercive power is absent would people accept the justication in question. But presumably there are possible worlds where people accept it, even though no coercive power compels them to do so. Moreover, it is very unclear what produced by coercive power means here. Education and wider acculturation help to shape which justications people will accept. How far are these processes coercive, and how far does this vitiate acceptance as a test of legitimacy? I do not have an answer to these questions. I do not think that there is an answer to them which enjoys the generality which political theorists, both liberal and others have aimed for. The holistic position I am sketching itself suggests that no clear-cut answer to them is to be had, at least if that means that it takes the facts of acceptance and subjects them to certain tests, such as counterfactual ones, in order to decide whether the acceptance has legitimacy-conferring force. At the same time, it suggests that the t between presented legitimations and the status quo need not be a particularly close one. The experience of normative incoherence is widespread and is not exclusively an artefact of modernity. The holism itself implies that imposed coherence will leave normative residues. The holism advocated here lies open to the charge of vulgar relativism, neatly encapsulated by Martin Hollis as liberalism for liberals, cannibalism for cannibals. Relativism comes in various avours and strengths. But suppose we characterize it along the following lines. According to political relativism, there are political systems, call them Lib and Can, such that for some normative judgment

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

463

N, Lib asserts N whereas Can denies (or at least fails to assert) N; and there is nothing in respect of which the disagreement between Lib and Can over N can be authoritatively adjudicated. Thus stated, the formulation of political relativism follows those used in discussions of epistemic relativism, with the notion of that in respect of which disagreements can be authoritatively adjudicated, replaced by one of extra-systemic truth. Then the question, however, is whether anything in the political case can play the arbitrating role of truth does in realist analyses of epistemic justication. If the role is allotted to agreement, or orientation to agreement, it is not a condition such as a transcendental condition on the intelligibility of making normative judgements that one be orientated towards agreement. No doubt there is an orientation at least to understanding, since that is spurned on pain of unintelligibility. But that orientation does not, in itself, carry any further commitment to agreement, even in an ideal world. Still less does any commitment to agreement, which may be highly localized, do the job of extra-systemic arbitration. One can of course substitute for truth the notion of validity, as the goal to which normative discourse is orientated. But either that notion is abstracted from the outcomes of real discursive processes, or it is not. If it is, it functions essentially like a moral norm, laid down in advance of politics, which discursive processes for example, those of politics need to respect, as either a goal or a side-constraint. Clearly, in view of what has already been said, the validity of these normative constraints on the processes will itself need to be established, presumably in some way independent of the processes themselves. If, conversely, they are real processes, they will be subject to the same vagaries as any real process of collective deliberation. Then the notion of validity itself will be contingent on discursive outcomes. It is quite hard to see any standpoint from which the dispute between Lib and Can could be authoritatively adjudicated, other than real political or quasi-political processes. In some cases, of course, the dispute will be notional rather than real, if for example it results in no conicts over jurisdiction or resource allocation. While the conict between modern western societies, say, and the Aztecs regarding human sacrice is notional rather than real, conicts elsewhere are real enough. But there is no purely theoretical object which delivers authority either as a repository of legitimacy in itself, or as a matrix for creating valid norms.

Conclusion
No doubt all theorizing has to take some points as xed and others as negotiable. Liberal political design takes morality as xed, and politics as negotiable. It imputes to individuals a certain signicance in constructing political principles from its preferred starting-point. This is one way of doing political theory. But the notion that it is the denitive way of doing it needs to be justied. One can equally well take the facts of corporate existence as given, and moralitys standing and content as variable. What is xed will clearly limit the scope of the negotiable. There is no reason to think that justication should play no role in the activity of

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

464

European Journal of Political Theory 9(4)

politics, including the political activity of theorizing about politics. But it does not follow that political design must be subject to moral norms, for example because it is held that power must be subjected to a regime of strong moral justication. The holism sketched here itself predicts a lack of t between the subject-matter of justication, and the justicatory resources which are, in large measure, furnished by that subject-matter. As far as that goes, both the basis for extraterritorial theorizing and its inconclusiveness are already implicit in the possibility of making an immanent critique. Notes
1. Blank distinctions between source and type may be questioned, on the grounds that reflecting on the source of normativity may yield conclusions about the type to which it belongs: Kantians, in particular, will incline to see the two as closely related. 2. This is most obviously true of democratic politics, but in principle applies to any system in which the executive is thought of as acting under authorization. 3. For the term political design, see Glen Newey (2001) After Politics: The Rejection of Politics in Contemporary Liberal Philosophy. London: Palgrave. 4. Justice, seen as a pivotal normative concern in political design, might be regarded as straddling the line between moral and specifically political normativity. But very often justice is understood simply as the political branch of moral normativity. 5. E.g. in Rawls (1980) Kantian Constructivism in Moral Theory, Journal of Philosophy 77: 51572, but also frequently elsewhere. 6. Dworkin contends that utilitarianism must rule out external preferences, and derives from this a utilitarian defence of the right to pornography. See Dworkin (1981) Is there a Right to Pornography?, Oxford Journal of Legal Studies 1: 177212. But why must my desire to stop you getting the porn you crave give way to my desires regarding my getting it? I may not care either way about that. 7. It is a further question what kind of normativity characterizes politics. It may be a version of bindingness. But there is no reason to assume that this must be so. A more relevant category is likely to be that of practical necessity. 8. The term trump is due to Ronald Dworkin; see his Rights as Trumps, repr. in Dworkin (1977) in Jeremy Waldron (ed.), Theories of Rights (Oxford: Oxford University Press 1984), pp15367. 9. I here summarize an argument developed more fully in my paper Just Politics, Critical Review of International Social and Political Philosophy; also in Emanuela Ceva & Enrico Rossi (eds.), Justice, Legitimacy and Diversity (London: Routledge 2012). 10. The justification of political arrangements requires that they not be reasonably rejectable; so either, implausibly, there can be no reasonable disagreement about political questions, or political arrangements predictably fail the justification test. 11. In my view Rawlss third option, perfect procedural justice, is chimerical, as I argue in Just Politics. 12. I ignore the complication that, when the relevant contexts are non-extensional, it is doubtful whether truth-functional equivalences are constructible even in principle. 13. Rousseau, Discourse on the Origins of Inequality in V. Gourevitch (ed.) (1997) Rousseau: the Discourses and Other Early Political Writings, pp. 151 ff. Cambridge: Cambridge University Press.

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

Newey

465

14. Hobbes (1996) Leviathan, ed. Richard Tuck, chs 1618. Cambridge: Cambridge University Press. 15. Hobbes introduces the notion in Leviathan and develops it in Hobbes, ed. William Molesworth, Elementorum Philosophiae Sectio Secunda De Homine in Molesworth ed., Opera Philosophica vol. II pp1132 (London 1839). 16. E.g. non-voluntarist theories, such as associationalism, which deny that political obligations require an originating agreement. See John Horton (2011) Political Obligation, 2nd edn. London: Routledge. 17. One response to this would be to deny that political obligations, which are not universal, should be identified with legal obligations, which are incident on citizens and resident aliens alike. 18. R. P. Wolff (1970) In Defense of Anarchism. New York: Harper & Row. 19. Quine (1951) Two Dogmas of Empiricism, Philosophical Review 60: 2043. 20. I argue this at greater length in Newey (n. 3), ch. 2. 21. Compare, for instance, the idea, due to Harold Lasswell, that politics concerns who gets what, when, how. No doubt political issues often revolve around these distributive questions. But if that is definitional of politics, it will make incoherent the intelligible claim that some issue involving distributive questions (such as the criteria for awarding honours or public recognition) should not be political. 22. I assume that this underlies Paul Kellys claim that endemic disagreement about its scope weakens the force of my contention that politics is an autonomous sphere of human activity: if disagreement is possible, it must be possible to claim that politics is, say, applied morality. (Kelly (2005) Liberalism, p. 106. Cambridge: Polity.) However, the autonomy of politics comes out in the fact that, since such claims are disputable and disputed, they can only constitute one move in a wider political argument. Any particular claim about the nature of the political can be countered by another one. But that fact, by itself, does not end the political argument. 23. See Dworkin (1987) What is Equality? Part 3: The Place of Liberty, Iowa Law Review 73: 154. 24. This seems to be true even of avowedly non-voluntarist accounts of political membership, such as associationalist ones. It may be that membership is thought of, in these accounts, as independent of an act of will by the members. But, insofar as they seek to provide a justification of membership, they are providing reasons in relation to which members can regard their incorporation in the association as assented to and therefore free. 25. Scanlons sophisticated contractualist theory holds that people are motivated to reason to seek agreement on terms that nobody can reasonably reject. This is not meant to explain why we should seek agreement on these terms: only unreasonable persons would demur from this. See Scanlon (1998) What we Owe to Each Other. Cambridge, MA: Belknap Press. 26. Williams (2005) In the Beginning was the Deed, p. 6. Princeton, NJ: Princeton University Press. Cf. Williams (2002) Truth and Truthfulness: An Essay in Genealogy, pp. 225 ff. Princeton, NJ: Princeton University Press. 27. Williams acknowledges the problems with the formulation, though he thinks that they arise because of the vagueness of produced by. While it is true that the aetiology of peoples acceptance of modes of justification is sometimes unclear, a further and larger problem arises not because the causal mechanisms are not clear, but whether these mechanisms should be described as coercive.

Downloaded from ept.sagepub.com at US BUREAU OF RECLAMATION on July 5, 2013

You might also like