Democracy as the Search for Justice: A Defence of the Democracy/Contractualism Analogy



This article advances a partially epistemic justification of democratic authority by defending what David Estlund has called the democracy/contractualism analogy: the idea that democracy can track justice due to crucial similarities between effective democratic politics and the hypothetical deliberations employed by contractualist liberals to explicate or construct correct principles of justice. According to this analogy, the collective decision making definitive of democracy is best conceived as an attempt to realise the process of intersubjective justification that (according to contractualist liberals) defines what is just; therefore, any tendency democracy might have to produce just outcomes could conceivably be attributed to the success of such an attempt. This article explains why Estlund's arguments against the analogy fail. It argues that in a society in which the modes of reasoning specified by the analogy are widely entrenched, there is a justifiable presumption that majoritarian voting will have a tendency to result in reasonable outcomes.

Why should liberals be democrats? One popular answer is predicated on the intrinsic value of democratic procedures. By empowering citizens to rule, democracy honours the ideal of autonomy that is itself the cornerstone of liberal political morality. But we need only look to cases of democracy gone awry – in which democratic procedures are marshalled towards the production of grossly illiberal outcomes – to wonder if honouring autonomy can really be the whole story for democracy's justification. For the liberal case for democracy to be secure, it must also include some account of democracy's epistemic value: its tendency to achieve outcomes that are acceptable by the standards of liberal justice. But on what basis could democracy be expected to do this? In his book Democratic Authority, David Estlund locates a candidate answer to this question within the literature on deliberative democracy, an answer grounded in what he calls the democracy/contractualism analogy: the idea that democracy can secure justice due to crucial similarities between good democratic politics and the idealised choice situations employed by contractualist liberals to explicate or construct correct principles of justice. On this analogy, the collective political decision making definitive of democracy is best conceived as an attempt to realise the very process of intersubjective justification that (according to contractualist liberals) defines what is just; therefore, any tendency democracy might have to produce just outcomes could conceivably be attributed to the success of such an attempt. Estlund considers and rejects this analogy. In so doing, he powerfully resists the prevailing inference from contractualist liberalism (the idea that correct principles of liberal justice can be construed as the conclusions to a hypothetical choice situation among reasonable citizens) to deliberative democracy (the idea that politics should ideally operate along the lines of such choice situation). Therefore, according to Estlund, while liberals should be democrats for many reasons, they should not be democrats on the basis of a democracy/contractualism analogy, and they should not be the kinds of deliberative democrats suggested by the analogy.

In this article my aim is to undermine Estlund's argument against the democracy/contractualism analogy, and to suggest in so doing that a particular version of the analogy is, in fact, a plausible explanation for why democracy is the ideal institutional arrangement for contractualist liberals. In the first section, I ask why contractualist liberals might contend that democracy possesses epistemic value, and I present the proposal that a democracy/contractualism analogy (DCA) could justify this contention. In the second section, I will argue that Estlund's rejection of this proposal relies upon a flawed interpretation of the nature of the particular contractualist situation (T. M. Scanlon's) in reference to which his argument is advanced. By defending an interpretation of Scanlon's contractualist situation according to which it is best conceived as an account of how we ought to understand and engage in first-order moral deliberation, I show how it can ground a partial analogy between the kind of reasoning that transpires among participants in a contractualist situation and the kind of reasoning that citizens should ideally employ in their deliberations with one another as citizens of liberal democracies. In the third section, I will explain why the conception of participation at work in Scanlon's contractualist situation – in which parties seek to identify mutually justifiable principles, free from prior dogmatic commitments about what justice substantively requires – is precisely the participatory ideal that actual citizens should strive to realise. Finally, in the fourth section, I will remark on how the democracy/contractualism analogy could successfully account for why democracy has epistemic value. I will suggest that when the practices of reflective citizenship of the kind specified by the analogy are entrenched, there may be a justifiable presumption that (a majority of) citizens will tend to affirm reasonable, as opposed to unreasonable, political proposals, and that this presumption plausibly grounds the argument for democratic voting as the best procedure for political decision making in a liberal society.

Introducing the Democracy/Contractualism Analogy: What is at Stake?

David Estlund's Democratic Authority is one of the most ambitious recent attempts to answer two questions that have dominated normative democratic theory since Rousseau: what is the justification for democracy, and why is it the kind of justification that citizens committed enduringly to one another's moral status should accept? Estlund persuasively argues that voting cannot be justified on the mere basis of procedural fairness; if all that mattered were procedural fairness, why not flip a coin to determine which policy to enact? Moreover, the intrinsic value of democratic procedures, flowing from their affirmation of citizens' autonomy, is evidently insufficient to ground their justification entirely; were those procedures to result consistently in illiberal outcomes, our faith in their justifiability would be rightly shaken. As Estlund (2008) compellingly argues, the justification of democracy most plausibly depends – at least partially – on the idea that democracy has some epistemic value. Part of what justifies democracy, in other words, is the moral quality of the decisions it tends to reach.

In the chapter of Estlund's Democratic Authority that concerns me here, Estlund's mission is to show why democracy's modest epistemic value – its modest tendency to produce reasonable outcomes – is not to be explained by what he calls a democracy/contractualism analogy: the claim that democracy can ‘produce outcomes that are right by contractualist standards … by promoting the similarity (in certain respects) of actual procedures to the procedure in the hypothetical contractualist situation’ (Estlund, 2008, p. 239). While Estlund fails to distinguish between the different versions of a DCA, there are several ways to understand what it might entail. The first part of Estlund's discussion mostly addresses the idea of what we can call a structural analogy, according to which proper democracy secures justice because the structure of democratic decision making mirrors the structure of some kind of idealised choice situation (such as Jürgen Habermas' [1998b; 2000], in which participants have unlimited time to deliberate, or Brian Barry's [1996] Scanlon-inspired deliberation, in which every participant is equipped with veto power). This idea, Estlund properly detects, is a dead end; it would be institutionally infeasible and normatively undesirable to give every actual, breathing citizen a veto over every policy. I will not dwell on this aspect of the discussion due to its evident fruitlessness, and instead move to a different kind of analogy that Estlund discusses, what I call a reasoning analogy. On this analogy, what is mirrored in actual politics is not the structure of an idealised situation, but rather the processes of reasoning in which parties to that situation engage.

Whom is Estlund talking about? In his original article on which the relevant chapter is based, he claims to be discussing ‘a central strand in theories of deliberative democracy’, though this formulation is not present in the article's current form (Estlund, 2003, p. 387). The roster he has in mind consists broadly of contractualists, including John Rawls (2005), Jürgen Habermas (1998a; 1998b; 2000), Brian Barry (1996), Joshua Cohen (1989; 1996) and William Nelson (1980). All these thinkers rely upon some kind of idealised contractualist situation to explicate the content of justice – whether it is the original position (Rawls), the ideal discourse situation (Habermas), the ideal deliberation (Cohen), what Barry calls a ‘Scanlonian original position’ (drawing on Scanlon's contractualist ethics) or what Nelson refers to as the democratic ‘tests’ a principle must pass – and, Estlund adds, believe that a deliberative democracy modelled in some way on that situation is the best way to achieve just outcomes. Because Estlund directs his discussion explicitly at Barry's Scanlonian version of the democracy/contractualism reasoning analogy, I accordingly restrict my discussion to this particular version of the analogy. But if his rejection of this argument fails, as I believe it does, it does not follow that every possible version of the reasoning analogy is thereby vindicated. Such a sweeping argument would require careful analysis of each other thinker's work, which I cannot undertake here.

The simple insight underpinning the reasoning analogy I defend can be grasped by distinguishing two crucial questions. First, how do we determine what laws are demanded by justice? Second, how should we determine what laws we are actually going to implement? The standard answer contractualist liberals offer to the first question is an account of a procedure of reasoning. The simple but arresting insight of the DCA is this: given that the morally proper end of legislation is indeed justice, the same procedure of reasoning by which we arrive at conclusions about what justice requires could be part of the way we determine what laws we are actually going to implement. In other words: if we all want the laws that are enacted to be just laws, then let the process of reasoning that enables us to identify what justice requires be the same process of reasoning that guides how we decide what to implement; if we do this, there is a hope that our conclusions will secure justice as opposed to injustice, for we will have engaged in the very process of reasoning that leads to the identification of ideally just principles and policies. That liberalism is itself predicated upon an affirmation of citizens' moral powers transforms this hope into a justified presumption.

This is the basic, rough shape of the best possible argument for a DCA along Scanlonian lines. Note a few points about the analogy. First, it does not presume that democracy secures justice by ‘an invisible hand’ (pace Estlund, 2003, p. 387). Just as a contractualist situation can ‘go wrong’ when philosophers make mistaken inferences, deductions or judgements in the reasoning process, so too can democracy ‘go wrong’ when citizens and lawmakers make mistaken inferences, deductions or judgements in the reasoning process. My claim is a qualified and modest one: when the reasoning procedure specified by the analogy is democratically entrenched, this could ground democracy's tendency to arrive at substantively reasonable outcomes – and thus secure democracy with the modest epistemic value that Estlund rightly thinks is essential to its justification (and do so in a more plausible way than he does [Estlund, 2008, p. 167]).1 Second, the analogy does not presume that there is no reasonable disagreement about justice. On the contrary: what unites contractualist liberals is not their agreement on all substantive matters of policy (though no doubt there are certain demands of justice it would be unreasonable to deny); rather, it is their agreement of how to arrive at judgements about such matters. Moreover, the legitimacy of law plausibly does not depend on its achievement of perfectly just outcomes, since citizens are bound reasonably to disagree about what outcomes meet that exacting standard. What matters is that democracy has a tendency to reach outcomes within the set of reasonable candidates for the perfectly just outcomes (Rawls, 2005, p. 450). That citizens will disagree is, of course, why a decision procedure – like democracy – is necessary. The DCA explains the conditions of democratic deliberation that ought always to preface this decision procedure if we are to have confidence that it will have a tendency to reach reasonable outcomes.

Defending the Democracy/Contractualism Analogy

I will now examine Estlund's argument against the version of the DCA I seek to defend, a Scanlon-inspired reasoning analogy. I believe that his argument against the reasoning analogy hinges upon an exegetically and philosophically suspect account of the role of the idealised choice situation in Scanlon's contractualism, and that it fails for this reason. I will demonstrate its failure by denying one of Estlund's central claims: that participants in a Scanlonian choice situation must not explicitly address ‘the primary question’ of what justice requires.

Scanlon's contractualist principle stipulates that ‘an act is wrong2 if its performance under the circumstances would be disallowed by any set of principles for the general regulation of behavior that no one could reasonably reject as a basis for informed, unforced general agreement’ (Scanlon, 1998, p. 153). He contrasts his approach with that of Rawls, noting that ‘the fates of others’ become relevant to parties in the original position since the oppressed and marginalised in society may include those for whom the deliberating parties are trustees. But Scanlon emphasises that on his view the involvement of others in the decision making is reflected in ‘a different kind of motivation, namely the aim of finding principles that others, insofar as they too have this aim, could not reasonably reject’ (Scanlon, 1998, p. 191). Their aim is to determine what correct moral principles require through attempting collectively to identify principles acceptable to all, given their common motivation to live ‘in unity with our fellow creatures’, as J. S. Mill said, a characterisation of moral life that Scanlon explicitly endorses (Mill, 1987, p. 303).

Note that the idea of an idealised contractualist situation appears only in the earliest announcement of Scanlon's theory, where he refers to the idea of a ‘choice situation’ in which individuals possess ‘full knowledge of their situations’ and ‘a desire on each of their parts to find principles which none could reasonably reject insofar as they too have this desire’ (Scanlon, 1982, p. 127). Barry develops the idea of such a choice situation into the notion of a ‘Scanlonian original position’, in which the idea that each participant can reject unreasonably burdensome proposals is conceived as a veto power possessed by each (Barry, 1996, pp. 67–72).

Estlund identifies an interpretive problem in explaining how the process of principle selection in Scanlon's choice situation is to work. What exactly do the parties to Scanlon's situation ask themselves? The trick, Estlund contends, is that they cannot be construed as asking themselves ‘the primary question’: the question of ‘what we owe each other’ or, equivalently in Scanlon's view, the question of what ‘can be justified to others’ (Scanlon, 1998, p. 7). (For the purposes of this discussion, we can equate this with the question of ‘what justice requires’ and with the question of what principles are ‘mutually justifiable’.) That is because – Estlund believes – Scanlon's contractualist situation is intended to play a distinctive role as a device that helps us as philosophers answer the primary question. But if parties to the situation are simply asking that very question themselves, why do we need them?3 To assign parties the primary question would imperil ‘contractualism's distinctiveness’ (Estlund, 2008, p. 251). This pushes Estlund to wonder whether parties are simply pursuing their own self-interest, rejecting proposals every time one frustrates a personal aspiration. But that cannot be quite right either, for Scanlon explicitly says that the parties ‘are assumed not merely to be seeking some kind of advantage but also to be moved by the aim of finding principles that others, similarly motivated, could not reject’ (Scanlon, 1998, p. 5). Estlund thus sets himself the challenge of conceiving how parties could be ‘moved’ in this way while nevertheless not asking themselves the primary question – a challenge that must be accomplished, he thinks, if Scanlon's contractualist device is to play any function distinctive from our own first-order moral reasoning.

Estlund outlines a solution with two elements. First – what he calls ‘reasonable self-service’ – we should conceive of parties as motivated to reject proposals ‘only if they themselves have personal reasons against them’ (Estlund, 2008, p. 248). In other words, ‘the contractual participants do not reject proposals on the grounds that they are reasonably rejectable by someone or other’ – this would be the primary question – ‘but only for their own personal reasons’ (Estlund, 2008, p. 248). Second, we should conceive of each party as ‘prepared not to press a personal reason he has against a proposal if this would leave only alternatives to which others had objections at least as weighty’ (Estlund, 2008, p. 251). By combining these two elements, Estlund believes that ‘the idea of reasonable accommodation among the contractors could be brought in without giving them the primary question’ (Estlund, 2008, p. 252).

Estlund's strategy is promising, and it could work – if only it were not for the following question: where do the alternatives that the parties consider come from? This unmentioned question haunts Estlund's entire discussion. Unless there is some designer dropping proposals into the initial situation, like newly cracked fortune cookies for parties to ponder, then proposals must come from participants. How could this be? Estlund does not say. Perhaps, as suggested by his conception of ‘reasonable self-service’, parties press their own interests within reasonable limits. What does this mean? Estlund's discussion gives us little guidance; all he says is that contractors accommodate one another purely by deciding not to veto a self-burdening proposal when alternative proposals are even more burdensome to others. We do not know whether he thinks any accommodation would transpire in the phase of proposing principles.

When we shift from the matter of rejecting proposals to the matter of advancing proposals, Estlund's distinction between reasonable self-service, on the one hand, and addressing the primary question, on the other hand, collapses. We have no reason to think that Scanlon's stipulation of parties as ‘moved by the aim of finding principles that others, similarly motivated, could not reject’ applies only to their role in rejecting principles; such a presumption would be decidedly ad hoc (Scanlon, 1998, p. 5). Now when it comes to conceptualising the process of rejection in the contractualist situation, it is perfectly coherent to stipulate that when a participant rejects a proposal, he or she will only reject it on the basis of personal reasons: if a proposed principle deeply frustrates someone's commitments or interests, only that contractor will be the one who rejects it. Likewise, it is perfectly coherent to say that a participant will take others' interests into account by refraining from vetoing self-burdening proposals that burden others even more. But when we turn to the matter of a participant who proposes a principle that advances his or her interests, we must include the proviso within reasonable limits in order for that participant still to be the kind of agent who possesses the aim of finding mutually justifiable principles. And this modifying clause can mean only one thing: that he or she is advancing his or her interests by proposing principles that do not unreasonably burden the interests of others. But to propose principles that advance one's interests within reasonable limits, and to propose principles that no one could reasonably reject, are one and the same.

Estlund writes:

[T]ruly analogous participants will not address the question of justice itself, but only their own interests so far as they can be reasonably pressed. This is not egoism, but nor is it a sufficient orientation to the common good to support the tracking claim under circumstances of real and proper democratic choice … [J]ustice would not be directly addressed by participants who were analogous to the hypothetical contractors (Estlund, 2008, p. 241).

But this cannot be right. Pressing one's interests only so far as reasonable means proposing principles that advance one's interests but in a manner that does not unreasonably burden the interests of others. It is not enough to advocate a principle of distributive justice with the aim of advancing one's own economic interests without demonstrating why it could not be reasonably rejected.

With this in mind, the distinction between advancing one's own interests within reasonable bounds and searching for mutually justifiable principles disappears. There is no middle ground between pursuing one's self-interest and pursuing mutually justifiable principles when it comes to the formulation of proposals. Moreover, Estlund's analysis is misleading in so far as it suggests that a participant is not advancing one's interests when he or she proposes mutually justifiable principles. After all, to search for mutually justifiable principles – principles that no one could reasonably reject – is to search for principles that are justifiable to oneself, as well as to others. Part of explaining that a principle is mutually justifiable means showing why it permits us ‘to govern ourselves in a way that others could not reasonably refuse to license’ (Scanlon, 1998, p. 157). Principles that would considerably and unnecessarily burden one's own (reasonable) aims are not mutually justifiable. Therefore, to identify mutually justifiable principles just is to identify principles that advance one's own interests within reasonable bounds. I cannot think of a principle for the general regulation of behaviour, or the distribution of benefits and burdens in society, that advances one's interests in reasonable bounds but would not be described as mutually justifiable. In this way, Estlund's distinction between ‘reasonable but self-serving motives’ (Estlund, 2008, p. 244) and the motives to find principles that no one, similarly motivated, could reasonably reject, is illusory when it comes to the matter of proposing principles.

Before proceeding to sketch an interpretation of Scanlon according to which we should see parties as asking themselves the primary question – and explaining why we should not regard this as unfortunate for the theory – I want to address one way in which Estlund could respond to my criticisms and defend the thesis that parties do not ask themselves the primary question. This would be to suggest that, while parties are accommodative at the rejection stage, we should conceive of parties as strictly self-interested when it comes to the stage of proposing principles. This, at first, seems to fit with the rest of Estlund's account: party A proposes a monstrously self-serving principle that would require the subordination of the rest of humanity to A's aims; B, burdened by A's proposal, vetoes it; A politely declines to object in response, recognising the reasonableness of B's rejection; and the self-serving principle thereby slides into the rejection pile.

One crucial problem with this approach is that it is not clear how we end up determining the set of principles that lie beyond reasonable rejection if all that is happening is that parties propose self-serving accounts with no attention (at the stage of proposals) to reasonableness. The situation would consist in the multiply repeating proposal and rejection of the same candidate principles. So we would need an account of how each party proceeds once his or her prudentially optimal principles are rejected. What criteria would A use when making his next gambit? Do parties concede territory, as it were, with each iteration of proposals – using information from the previous round of rejections to dilute the self-serving character of each fresh proposal as they bargain their way to justice? If so, how far do they dilute them, and in accordance with what criteria? If the criteria are moral, then there is pressure to think that the parties could just cut to the chase and propose mutually acceptable principles – which would undermine Estlund's thesis that they are not asking themselves the primary question. I believe that these questions render this proposed interpretation unworkable. But even if they could be answered adequately, this would nevertheless be an extremely exegetically unattractive interpretation of Scanlon's contractualist situation. Scanlon (1998, p. 5) insists that we should regard parties as ‘moved by the aim of finding principles that others, similarly motivated, could not reasonably reject’. It is highly implausible to think that this is somehow an error; Scanlon makes the comment precisely when he is contrasting his initial situation with others in the literature: ‘What distinguishes my view from other accounts involving ideas of agreement is its conception of the motivational basis of this agreement’ (Scanlon, 1998, p. 5). Yet the idea that such parties are characterised as only prudentially rational by (my extended version of) Estlund's interpretation of the proposal phase clearly sits in serious tension with this characterisation. It is worth searching for a rival interpretation of Scanlon's contractualist situation that avoids this tension, and indeed that avoids the serious problems that I have shown to afflict Estlund's reading.

Re-conceiving Contractualism

Reinterpreting Scanlon

If my criticism of Estlund is sound, then we should regard Scanlon's contractors as asking the primary question. This is the only way, I have argued, to make sense of Scanlon's contention that they are ‘moved by the aim of finding principles that others, similarly motivated, could not reasonably reject’ (1998, p. 5). Intriguingly, Estlund himself admits that this quotation ‘seems to give them the primary question’ – though he only admits this in a footnote in his 2003 article; it does not appear in Democratic Authority. ‘The exegetical problem raised by these quotes in Scanlon cannot be pursued here’, Estlund continues in that footnote – a puzzling statement, given that the right way to understand Scanlon's account is precisely the focus of his discussion (Estlund, 2003, p. 401n).

Recall that what motivates Estlund is a worry that assigning the primary question to contracting parties will end up imperilling ‘contractualism's distinctiveness’ (Estlund, 2008, p. 251). If parties are simply asking the primary question, why do we need to think about them? We can just ask ourselves. The choice framework's role as a device through which conclusions about right and wrong are constructed or determined depends, he thinks, on this not being the case. As a result, Estlund goes to great lengths to interpret Scanlon in a way that retains the distinct character of the reasoning within his contractualist situation. But his efforts, as I have shown, are strained. Moreover, they are unnecessary. For there is nothing incompatible between the claims (a) that parties to Scanlon's contractualist situation ask themselves the primary question and (b) that Scanlon's contractualist situation represents an account of how correct moral principles are best identified. That is because Scanlon's contractualist situation just is a framework for understanding how moral reasoning – the process of arriving at determinate conclusions about right and wrong – properly proceeds. It is an account of first-order moral reasoning, through which determinate moral principles are identified or, if you prefer, constructed.

To understand this way of interpreting Scanlon, consider some structural similarities between his view and the view of moral justification offered by John Rawls. Glossing over details for the sake of simplicity,4 I think we can regard Rawls' project as beginning by identifying (a) an account of citizens' moral status as free and equal and (b) a companion account of their interests as free and equal (such as their interests in developing and exercising their moral powers, and in primary goods). He then seeks to determine (c) the principles by which social institutions should be arranged if they are to respect this understanding of persons as free and equal. The original position is a process of reasoning that takes us from where we start – with (a) and (b) – over to (c). ‘The leading idea’, Rawls writes in 1980, ‘is to establish a suitable connection between a particular conception of the person and first principles of justice, by means of a procedure of construction’ (Rawls, 1980, p. 516). And while the later Rawls hedges on his meta-ethical conviction that principles are constructed by this process (as opposed to discovered by it or, as the later Rawls comes to say, ‘represented as the outcome’ of it), the structure is clear: we begin with some account of normative ‘raw materials’, including an idea of the basic moral status of persons, and we reason our way from that account to specific principles (Rawls, 2005, p. 93).

Scanlon's account can be understood similarly. As Scanlon himself says:

The very structure of the [contractualist] test is already a moral principle that constrains the kind of norms that can pass it. It is already a moral principle that everybody counts – that we should be able to justify our norms to everyone (quoted in Voorhoeve, 2009, p. 182).

Scanlon thus begins with an account of the moral status of persons according to which they are equally entitled to justification. In virtue of this moral status, Scanlon believes – and here I am quoting Michael Ridge – that ‘everyone has reason to make room for the … agent-relative concerns of others’ (Ridge, 2001, p. 481). Scanlon thus proceeds to offer us an account of objective but nevertheless agent-relative reasons that each person has: reasons to seek food and shelter, pursue certain projects, devote time to one's family, develop relationships, etc. (Scanlon, 1998, p. 204). What Scanlon's contractualist deliberation does is serve as the framework of reasoning between these normative raw materials from which we begin – an account of the moral status of others and a companion account of persons' agent-relative reasons – to reach specific conclusions about right and wrong. It describes how we should reason our way from these raw materials to arrive at conclusions about what we owe to each other. And we do this by aiming to devise a set of principles that, given the objective agent-relative reasons for action people have, does not burden anyone in a way that could be avoided by a rival set of principles. We do this, in other words, by asking the primary question. Just as Rawls' original position takes us from abstract normative raw materials to determinate conclusions, so too does Scanlon's contractualist situation. Each is an account of how someone who endorses the fundamental ideas of the theory – who is morally committed, in Scanlon's case, to the equal claim of all to justification – should reason his or her way to conclusions about what he or she ought to do. Indeed, DCA pioneer Barry's influential proposal that we import Scanlon's contractualist account into Rawls' theory would only make sense if their accounts of contractualist reasoning played similar roles (Barry, 1996, pp. 67–72).5

There are many virtues in interpreting Scanlon in this way. First, it is in many respects a more natural reading. We need not strain to find a description of what goes on ‘inside’ the contractualist situation that marks it out as distinct from what goes on ‘outside’ in the first-order activities of moral deliberation; instead, we can embrace the idea that Scanlon's descriptions of the ideal agents in each are identical. Just a page before Scanlon describes the parties as ‘moved by the aim of finding principles that others, similarly motivated, could not reasonably reject’ (Scanlon, 1998, p. 5), he says, remarking on the phenomenology of moral experience:

When I ask myself what reason the fact that an action is wrong gives me not to do, my answer is that such an action would be one that I could not justify to others on grounds I could expect them to accept (Scanlon, 1998, p. 4).

My reading does not force us to invent some way in which these two descriptions differ. The idealised processes of reasoning that carry on inside the contractualist deliberation are the same processes of reasoning that we ourselves undertake when we do moral deliberation properly.

Second, my interpretation helps to make sense of Scanlon's mysterious contention that reasons of ‘fairness’ can be invoked against principles in the contractualist situation. This is because principles that ‘arbitrarily favor one person over others’ offend against the root idea of the equal standing of all persons that serves as the foundation of Scanlon's contractualism (Scanlon, 1998, p. 212).6

Third, this interpretation helps immunise Scanlon from the common objection that his theory is objectionably circular because it relies on substantive judgements. This common objection begins by recognising that the reasonableness of any given rejection in Scanlon's deliberation needs to be assessed somehow – namely, by assigning weights to agent-relative reasons people have for and against a proposed principle, and seeing which person's reasons are strong enough to win out. But if this is true – and it is – a considerable amount of important work thereby transpires at the level of assigning moral weights to the different reasons, assignments which cannot be explained by the final principles of morality that are settled upon since they are necessary in reaching those final principles. What this objection concludes is that this makes the idea of a contractualist situation entirely dispensable: all we need to do is look at the realm of reasons for action, judge the relative strengths and then see what principles are demanded or disallowed by the strongest reasons (Southwood, 2010, pp. 61–9).

However, once we see that Scanlon's account does not cast the contractualist situation as a specialised device, but rather as a framework for understanding moral deliberation at its best – a framework that directs and orients these judgements in reasoning – this objection misses the mark. Scanlon is explicit about the role of judgement in his theory: ‘According to my version of contractualism, deciding whether an action is right or wrong requires a substantive judgment on our part about whether certain objections to possible moral principles would be reasonable’ (Scanlon, 1998, p. 194).

But this is not circular; it simply comes with the territory of all accounts of first-order moral reasoning (utilitarian, virtue-ethical and Kantian alike). Rawls' account, too, depends on judgement in determining how to rank different principles lexically. That there are such burdens of judgement in any theory is precisely why there is reasonable disagreement about what conclusions are properly reached – but this is an issue for any view (Rawls, 2005, pp. 56–7). Once Scanlon's account is understood as a description of idealised moral reasoning, the objection that such reasoning depends on substantive judgements becomes utterly unsurprising, and thus loses its force.7 Moreover, Scanlon himself says that while ‘the possibility of tightening contractualism’ as a general moral theory to reduce the role of judgement is implausible at such a high level of abstraction as assessing ‘what we owe to each other’, such tightening is ‘a feasible aim with respect to some specific areas of morality’ (Scanlon, 1998, p. 218). It is possible that the justice of social institutions is one of these areas.

Fourth, the most important reason to think that Scanlon would endorse this interpretation is that he does endorse it. Consider a restated form of the aforementioned circularity objection posed by Alex Voorhoeve:

It seems that to get some determinate answers to questions of right and wrong, we must first decide which things count as valid ground for rejecting a principle. But this would appear to leave the procedure devoid of any content: any moral principle you pull out of the contractualist hat is put there at the outset when you decide which things count as grounds for rejecting a principle (Voorhoeve, 2009, p. 182).

In response to this, Scanlon replies – and this is the kicker:

[W]e can't avoid questions of judgement, and this could lead to the charge that the contractualist reaches conclusions only because he has helped himself to lots of substantive ideas along the way, and that the theory itself therefore doesn't really yield new answers. It may be that this criticism is justified. There are different desiderata of a moral theory. I am inclined to think that providing a way of cranking out novel principles is overrated … What I mean to do is offer a way of understanding what moral thinking is (Scanlon, 2009, p. 183, emphasis added).

Scanlon goes even further, disputing the idea that parties to his contractualist situation are best conceived (as Estlund recurrently does) as a ‘hypothetical agreement’ that therefore must be modelled in some special way distinctly from our normal moral deliberation. Instead, Scanlon argues:

To describe [my theory] as a ‘hypothetical agreement’ is already mistaken. It isn't about what people would agree to under certain conditions. Rather, what matters is what it would be reasonable for them to reject under certain conditions, that is, if they, too, were trying to find principles that others could not reasonably reject (quoted in Voorhoeve, 2009, p. 184, emphasis in original).

That such conditions are hypothetical therefore only serves to indicate that they are idealised – they characterise the conditions of faultless reasoning – not that they refer to the operations of mysterious, idiosyncratically motivated agents who undertake some special contract with one another. Rather, they refer to our moral reasoning about what we owe to each other, at its best.

Pessimistically, Michael Otsuka writes that Scanlon's contractualist situation ‘is bought at the price of a reduction in the usefulness of the contractualist device as a means (distinct from ordinary forms of moral reasoning) of arriving at the right principles of justice’ (Otsuka, 2003, pp. 5–6). But Otsuka here makes the same mistake as Estlund: assuming that Scanlon's contractualist situation was a specialised device to serve as a ‘means’ to philosophers external to our first-order moral reasoning. On the contrary, it is an account of how we ought to undertake first-order moral reasoning, one rooted in Scanlon's thesis that ‘the idea of justifiability to others is taken to be basic’ to what morality is, and thus how we should undertake the search to identify its demands (Scanlon, 1998, p. 5).

From the Idea to the Practice of Reasonable Rejection

Having presented an interpretation of Scanlon's contractualism that regards it as a description of how we ought to understand and engage in first-order moral deliberation, I want to note some implications of this interpretation that are relevant for the defence of a Scanlon-inspired DCA. Scanlon's discussion of deliberation is clearly focused on the idea of deliberation within a single person's mind: each asks him or herself whether it could be reasonable for someone to reject the principle upon which he or she is about to act, and draws a personal judgement about its justifiability. But while Scanlon is right that such judgements are ones that ‘each of us must make for him- or herself’, he also emphasises that ‘interaction with others plays a crucial role in arriving at well-founded moral opinions’ (Scanlon, 1998, pp. 393–4). What I want to do here is show how Scanlon's account of moral deliberation presents a useful organising framework for those interactions in actual political life.

Scanlon enjoins us to search for principles that no one could reasonably reject. But the could is crucial. After all, if participants in the moral deliberation really are the kinds of participants who seek mutually acceptable principles, why would there need to be any reasonable rejections that transpire at all? Who in the deliberation would have proposed a principle that was reasonably rejectable, and why would he or she have proposed it if he or she had the aim of finding mutually justifiable principles? If everyone reasons faultlessly in moral deliberation, and if we assume that all the addressed questions have determinate answers, then this will not happen. But of course, real deliberations will include moral mistakes. And when such mistakes are made, the right response in deliberation is to condemn them on the basis of reasons – to ‘reasonably reject’ them. The powerful image of people who stand and actively reject proposals on the basis of their unreasonableness is indispensable to the democratic recreation of Scanlon's process of reasoning. Such active rejections serve two vital roles in deliberation. The first is to expose malevolence: that people are wilfully acting by norms they cannot justify to others. But the second – what I want to pursue here – is epistemic. When we ask ourselves why someone in a real-life deliberation who is morally motivated might have proposed a principle that was reasonably rejectable, the answer is clear: because he or she did not know it was unreasonable.

I want to develop the proposal that Scanlon's contractualism provides not just a framework for individual moral deliberation, but constitutes a powerful proposal for collective moral deliberation. Recall how this framework operates. When a proposal is advanced, he claims, the first task is to identify whether any participant has a justifiable objection to the proposed principle. If a participant does – if the proposal burdens him or her in a way that alternative arrangements could avoid without burdening another participant more – the proposal must be rejected. In making such a determination, it must also be considered whether the failure to permit this principle would have far worse effects on others; in such a case, those others might respond with objections to the proposed prohibition of the proposed principle (Scanlon, 1998, pp. 202–3).

This idea of a dialectical exchange of considerations that results in a judgement about whether a proposed rejection of a proposed principle is reasonable is also reflected in Scanlon's discussion of generic reasons, or reasons that are grounded in ‘commonly available information about what people have reason to want … in virtue of their situation, characterised in general terms’ (Scanlon, 1998, p. 171).8 It is on the basis of these generic reasons that the proposed rejections of principles are justified – are deemed reasonable rejections. Consider how the process of generalisation transpires. Participant A advances a proposal whose adoption would advance her interest, but believing that it does not unreasonably burden others. Participant B then contends that A's proposal does, in fact, burden B's interests in a manner that an alternative arrangement would not; B then proposes this alternative arrangement. Participant C then objects that the new arrangement is overly burdensome to her, and that a further alternative is available.

Now fill in the example with a very easy and pretty implausible (but conveniently clear) case. Perhaps A proposes that her Catholicism be the religion of the state, thinking that it would allow her to practise her Catholic beliefs easily, and being unaware that there are others who would object strongly; B, a Protestant, responds by suggesting that they generalise the proposal to make Christianity the official religion of the state, so as not to burden B's interests in practising her own sect's faith; and then suppose C, a Buddhist, suggests they generalise the principle further to be a principle of religious freedom, whereby all can exercise their religion of choice. They have now arrived at a principle that no one could reasonably reject, and specifically through the identification of the generic reasons.

The upshot of the idea of generic reasons is that an agent who rejects a proposed principle must explain the burden that the acceptance of such a principle would inflict upon him or her in terms to which all could coherently relate. For example, if a person felt compelled to consume a particular kind of food in accordance with her religious beliefs, and a principle was proposed that would have the consequence of banning the consumption of that food, she would have to explain her plight in mutually comprehensible terms by claiming that a principle forcing someone to sacrifice a core practice endorsed by that person's conscience, in the absence of further evidence that the practice's exercise burdens others in any comparable way, is a reasonably rejectable principle. But in order to endorse her situation as such, I maintain, one would have to comprehend the basic role of the particular food in her religion, and the fact that she actually endorsed it as central to her core beliefs, in order to understand her argument, or to confirm that her particular situation would indeed be protected by the generic value to which she is appealing. Cases of this kind abound from political experience; the justifiability of serving pork in state school cafeterias would never have been considered an issue of justice before some began to question it. In such cases, each side must perform important mental work: the objector must do the work of re-describing her situation under a generic reason, like dietary autonomy or liberty of conscience; and everyone else must realise that the practice in question is correctly protected by the generic reason – even if the practice is one predicated on specific values they do not share, or even that they potentially repudiate in their own lives.

Indeed, it is precisely because the particular claims subsumed under generic reasons are often subject to intense epistemically reasonable disagreement that the process of working out what kinds of claims are or are not covered by generic reasons is particularly onerous. Consider Scanlon's example of a generic reason: privacy. People have many different reasons for favouring privacy, but their underlying normative accounts of its importance vary considerably. Therefore, someone who objects to a situation because it would subject her to shame and humiliation – a conservative Muslim woman in a Western airport forced to undergo a strip search and thereby expose bare skin to passing men, for example – needs to re-describe her needs in mutually comprehensible terms, in generic values. This re-description will be especially important since many Westerners do not object to such strip searches on privacy grounds. They are thus less inclined to see the objection as covered by the generic reason of privacy, and precisely since so many of them disagree reasonably (in the epistemic sense) with the underlying principle that women ought not expose bare skin to passing men. But the re-description is motivated by her aim to find mutually justifiable principles: to find principles that protect her interests within reasonable bounds and are defended in a way that demonstrates their reasonableness by being cast in a general language that all in the moral community can share.

The notion of a generic reason is intuitive to members of liberal cultures, accustomed as we are to conceptual taxonomy – subsuming particular complaints, grievances and claims under general pre-established classificatory categories. But the word ‘pre-established’ does a lot of work here, and it is not obvious in advance of political experience what these pre-established categories are. They need to be worked out through social and political practice. As Nancy Fraser notes, ‘Until quite recently, feminists were in the minority in thinking that domestic violence against women was a matter of common concern and thus a legitimate topic of public discussion’ (Fraser, 1992, p. 129). Examples like these make it difficult to suggest that the identification of generic reasons is a single task to be accomplished once and for all in a single moment of theoretical reflection; after all, when so many philosophers of the past were mistaken in their views on women and race, who is to say what we today are mistaken about? It is more plausible, then, to describe the identification of generic reasons as a process whose conclusions must continually be regarded provisionally. Fittingly, Scanlon explicitly endorses the thesis that the identification of generic reasons is an ongoing process through which moral agents constantly reflect on the particular things that particular people have reason to care about, and the generalised patterns that emerge from reflecting on these things as we consider what limits the contractualist ideal of mutual justifiability places on our pursuit of them.

This constructive discussion of the ‘dynamics’ of Scanlonian reasonable rejection helps us construe the practical instantiation of Scanlon's moral deliberation as a search for moral learning among participants. I say that this is ‘constructive’ because Scanlon does not devote much space to the practical development of moral commitments on the basis of reasoning. But he clearly recognises its importance:

[T]he shaping role of the aim of justifiability to others is a dynamic one. There is no fixed list of ‘morally relevant considerations’ or of reasons that are ‘morally excluded’. The aim of justifiability to others moves us to work out a system of justification that meets its demands, and this leads to a continuing process of revising and refining our conception of the reasons that are relevant and those that are morally excluded in certain contexts … [It] is a continuing process, not a fixed list of results (Scanlon, 1998, pp. 157–8).

Thus Scanlon's contractualist situation constitutes a promising framework for how morally motivated actors can expose one another's moral mistakes and learn from one another in actual political life.

It may seem strange to spend so much energy in an essay on democracy in interpreting the work of T. M. Scanlon, whose work is not addressed to democratic theorists. But aside from the distinctive value of Scanlon's work, Estlund has made this focus necessary. Estlund has, in many eyes, hammered a nail in the coffin of the democracy/contractualism analogy. But this is mistaken: the nail did not reach its target, and the contents of the coffin remain alive. Debunking Estlund's flawed rejection of the DCA is the key to resuscitating it as a serious contender in the ring of candidates for why democracy might have epistemic value.

Securing Justice through Democratic Citizenship

The basis of an analogy between the contractualist situation – in which idealised moral reasoners address the primary question – and actual democratic politics – in which we address the primary question, endeavouring to emulate the open-mindedness and thoughtfulness of ideal participants in so doing – should by now be considerably clearer. The insight I have sought to establish is that practices of ongoing discussion, moral learning and thoughtful reflection have epistemic value. How do they have epistemic value? When citizens apply intelligence to their task of finding mutually justifiable principles, constantly interrogating their assumptions and learning from others, they are more likely to succeed in identifying mutually justifiable principles. Therefore, the entrenchment of these practices of conscientious citizenship can better ensure mutually justifiable policy outcomes. And given that the best political decision-making process in a liberal society is one that produces mutually justifiable outcomes, these practices of citizenship should be at the heart of society's decision-making process. This is the democracy/contractualism analogy: a plausible candidate for deliberative democracy's epistemic value.

But what makes it democratic? It should be clear what makes it deliberative. But where does democracy fit into this story? Estlund's discussion presented me with an opportunity to entertain the possibility that in a certain kind of society – one in which citizens are conscientiously committed to the achievement of liberal justice – there might be a tendency, owing to the practices of those citizens, for citizens to develop normatively reasonable political convictions, such that they would not seek to impose their comprehensive doctrines on one another. And if there could be a tendency for citizens to be reasonable in this way, such a tendency could ground the presumption that having those very citizens make decisions would result in reasonable outcomes. If we take this presumption and conjoin it with the idea that some decision is needed in the face of disagreement, then democracy – conceived as the collective authorisation of laws through voting by those subject to those laws – becomes a plausible candidate for how decisions should be made in a liberal society, a society in which citizens share the aspiration to treat one another as free and equal through the enactment of mutually justifiable laws.

Two closing clarifications are important. First, I have attempted to resuscitate one argument for why deliberative democracy has instrumental value. But this is fully compatible with the claim that democracy enjoys other sorts of value, such as its intrinsic value in publicly honouring citizens' status as equals (Christiano, 2008), or the distinct non-instrumental value it secures by placing citizens in valuable relationships with one another (Beerbohm, 2012). All I have done here is endorse Estlund's proposition that part of democracy's justification must be epistemic, and defend (pace Estlund) one way of understanding that epistemic value. What serves as a sufficient justification for democratic procedures remains unanswered. Second, my argument is not that the majority will necessarily have come to the most reasonable solution on every issue; the burdens of judgement guarantee contestation on that very point. While I have only roughly sketched the normative case for a democracy/contractualism analogy in this article, it is plausible that a fuller exposition would rest more explicitly on the claim that the entrenchment of the practices of liberal democratic citizenship – practices of reflection and self-scrutiny – serves not to guarantee that a majority of citizens will converge on the most reasonable conception of justice, once and for all, but rather to ensure that unreasonable laws and policies are regularly avoided, and that this is sufficient for democracy's justification. This fuller exposition would necessarily be accompanied by an investigation of the organisational implications of this argument: how institutions and civic practices can be devised intelligently to bring the moral best out of democratic citizens by inducing the right forms of deliberative engagement. Even with such an account in place, it is important to recognise that we will never be able to say, collectively, that the most reasonable solution has been found to every problem we face. The fact that the disagreement about justice is reasonable means that we agree on a common criterion – reasonableness – and that a successful process of reflection and argument in accordance with that criterion will tend towards better conclusions. Of course, we will come to differ during the long course of that process, as the burdens of judgement kick in. But it will have pointed us in the right direction: towards justice. And in a society in which a preponderance of citizens are engaged in practices that point their reasoning in this direction, there is a justified presumption that democratic decision making will tend towards reasonable rather than unreasonable conclusions.


I am grateful to audience members at the Harvard University Graduate Conference in Political Theory. I am also grateful to Malte Frøslee Ibsen, David Miller, Jeremy Waldron, and the anonymous reviewers of Political Studies for helpful feedback.

  1. 1

    Estlund believes that democracy's tendency effectively to avoid primary bads (war, famine, genocide, political collapse, etc.) is what confers democracy with modest epistemic value. But this empirical claim – which he explicitly chooses not to defend – is contestable; majoritarian votes have unjustifiably burdened the interests of minorities so frequently in democracy's history that I cannot see how this could be the basis of democracy's modest epistemic value. When democracies have successfully and enduringly secured justice, they have potentially done so precisely because their populations were concerned with justice's achievement in precisely the manner that my version of the DCA enjoins.

  2. 2

    Scanlon aspires to explain a broader terrain of morality than justice – hence ‘wrong’ instead of ‘unjust’ – but the terrain he explains indeed includes justice.

  3. 3

    Note that this is a different concern from the idea that Scanlon's situation is circular. While Estlund conflates these two objections, the latter concerns the substantive role of judgement in Scanlon's theory with respect to how reasons are to be weighed against each other. Even if Estlund's characterisation of Scanlon's contractualist situation were correct – even if we could adequately conceive of parties as asking themselves something different from the primary question – the distinct worry of circularity would remain. I shall address this issue later on.

  4. 4

    For example, I am deliberately ignoring how the idea of society as a fair scheme of cooperation figures in the Rawlsian reasoning process.

  5. 5

    It may appear that Rawls and Scanlon are crucially disanalogous in that Rawls' parties cannot be conceived as asking themselves the primary question. But we can re-describe what goes on inside the original position in a way that eliminates this disanalogy. The original position is a matter of thinking about ourselves as free and rational agents, and then asking what principles secure our fundamental interests as free and rational agents, free from the morally arbitrary contingencies that render our choices heteronomous. These principles are the principles of justice. Recall that in the stirring final passage of A Theory of Justice, ‘to see clearly and to act with grace and self-command’ from the perspective of the original position (and thereby aspire to ‘purity of heart’) is precisely what Rawls enjoins us to do (1999, p. 514). Thus when appropriately described, even Rawls' own procedure of construction can be viewed as one of asking the primary question – and undertaking the reasoning that leads one to settle on answers to that question.

  6. 6

    I thank an anonymous reviewer for suggesting this point.

  7. 7

    It is true that Scanlon is concerned to be sure that his framework is not ‘unnecessary’ (1998, p. 213). But here he is concerned not with the role of judgement, but with the description of moral motivation: we must not describe idealised reasoners in such a way that they reason with substantively developed views of justice already on hand. That would make the framework unnecessary. But this proves my point: for moral deliberation would also be ‘unnecessary’ if we had already decided and made up our minds unreflectively about what we owe to each other. Scanlon's parties, like us, should be armed with two things only: a capacity to make substantive judgements about reasons, and the willingness to search for mutually acceptable principles. These are a far cry from having a determinate account of justice already on hand. We still need a reasoning framework to get us there.

  8. 8

    Among such generic reasons, Scanlon argues, are ‘reasons to want to avoid bodily injury’, ‘to be able to rely on assurances they are given’, ‘to have control over what happens to their own bodies’ and so on, including being able to ‘give special attention to our own projects, friends, and family’ (Scanlon, 1998, p. 204).


  • Jeffrey Howard is Lecturer in Political Philosophy in the Department of Government at the University of Essex in Colchester, UK. While a doctoral candidate at Oxford University, where he studied at Nuffield College, he served as College Lecturer in Political Theory at both Trinity and Merton Colleges. His research interests include contemporary liberalism, democratic theory, moral psychology and the philosophy of criminal law. He holds an MPhil from Oxford and a BA from Harvard. Jeffrey Howard, Department of Government, University of Essex, Wivenhoe Park, Colchester CO4 3SQ, UK; email: