This paper rereads David Hume's Dialogues Concerning Natural Religion as dramatising a distinctive naturalist account of political toleration. I have a double purpose in mind: first, to complete and ground Hume's fragmentary explicit justification of toleration; second, to unearth a potentially attractive alternative to more recent, and especially Rawlsian, approaches to toleration. My project is therefore both interpretative – an attempt to understand Hume's thought – and archaeological – an attempt to bring some neglected riches to light. To make my case, I first sketch in two pieces of background: the problem of toleration (section 1), and the intertwined influences of naturalism and scepticism in Hume's thought (section 2). I then address Dialogues directly and develop a new interpretation of Dialogues part 12 as political drama (section 3), before considering some comparisons between a Humean and a Rawlsian justification of toleration (section 4).

1. Toleration

The problem and theory of toleration is rooted in the early modern recognition that pluralism is not temporary. In modern societies, people will have deep differences over the most important moral, political, social, and especially religious questions and values. Attempts to suppress these differences are disastrous: if they temporarily succeed, the result is tyranny, conspiracy, and the Inquisition; when they fail, the result is religious war. The idea of toleration, then, is the idea of leaving a space of accepted difference, for instance between Protestants and Catholics, while maintaining a shared core of belief or of moral and political practice. That is, toleration is the on-its-face paradoxical idea of choosing not to enforce some of what we think most important.

Within that broad idea, there are many particular theories of toleration, and many taxonomies have been developed to handle that range. Preston King, for instance, divides theories of toleration by their objects, and distinguishes between ideational, organisational and identity toleration.Footnote ² Catriona McKinnon divides by justification, and distinguishes between theories grounded in scepticism, in value pluralism, and in the ideal of reasonableness.Footnote ³ Rainer Forst divides by the relation between tolerator and tolerated, and distinguishes between permission, coexistence, respect, and esteem conceptions of toleration.Footnote ⁴ For my own purposes, I shall make use of three taxonomic questions to distinguish theories of toleration:

1. 1. What justifies the shared core of belief or practice in the face of difference?

2. 2. What must we refrain from enforcing or remain neutral about?

3. 3. Who gets in?

Question 3 is the mirror-image of question 2: it asks what we must not refrain from enforcing or remain neutral about – what are the limits of toleration? (A question often pressed by asking, Must we tolerate the intolerant?)

Two examples will make the point of these questions clearer, as well as providing useful materials for my later argument. First, consider John Rawls's theory of toleration:

1. 1. A shared core of belief and practice is justified by reasonable agreement between free and equal citizens, dramatised by a thought experiment – the original position – which asks each of us what political regime she would choose to live under if she knew neither which of its roles she filled, nor her own answers to the deepest moral, political, social, and religious questions. The shared core is what it is fair to require of one another despite our deep differences – that is, what would be chosen under these conditions of fair choice.

2. 2. We must remain neutral, in the public or political sphere, about our comprehensive conceptions of the good: we must choose not to enforce our answers to the deepest questions. However, Rawls does not intend to imply scepticism about those answers. We are fully to hold and assert the truth of what we believe in private.

3. 3. Anyone who can limit their public discourse, and their arguments about the use of state power, according to the ideal of reasonableness, gets in. In Rawls's own U.S. context, for instance, premillenial dispensationalist Christians get in, just so long as they make no appeal to The Revelation of St. John the Divine when trying to direct public power – because they cannot reasonably expect other citizens to accept such arguments.Footnote ⁵

Second: toleration is a central issue for Hume, both in his moral and political philosophy, and in his history; but his explicit discussion of toleration is fragmentary and incomplete, and one of my two purposes here is to offer a completion for it. For the moment, then, consider Hume's explicit account of toleration:

1. 1. A shared core of belief and practice is justified in two ways. First, by a sceptical argument derived from Pierre Bayle: no-one can be sure enough of the truth of her own beliefs to have rational warrant to enforce them. Second, and more prominently, by a historically-informed pragmatism. Hume repeatedly argues in his History of England that attempts to end pluralism of religious belief, for instance by Mary, Elizabeth I, or Philip II, are disastrous in practice. They result in tyranny and war; so, toleration is ‘a compromise born out of long-lasting suffering’.Footnote ⁶

2. 2. We must remain neutral between Christian confessions. Hume argues for specifically religious toleration, on, again, pragmatic grounds: religious difference is the central problem for peace in his own time.

3. 3. Who gets in? Hume is frustratingly unclear: Protestants, Catholics, and perhaps deists do; but the exact limits of the space of toleration are not well-drawn.

This is an unsatisfying theory of toleration, for at least three reasons. It fails to answer my question 3. It focuses only on religious toleration, which, although an important question now as for Hume, is not the only problem of difference. And, most importantly, the justifications Hume gives for toleration are weak. The sceptical argument does too much: if no-one can be sure enough of her own beliefs to have rational warrant to enforce them, then no-one can be sure enough of her belief in toleration to have rational warrant to enforce that.Footnote ⁷ Either there are cases where we must enforce despite a lack of warrant to do so, and the proposed justification of toleration fails to show that our deepest moral, social and religious claims are not among those cases; or enforcement always requires warrant, and toleration is no more warranted, nor therefore enforceable, than any other social and political practice. Either way, sceptical toleration is self-defeating. The pragmatic justification of toleration, on the other hand, doesn't do enough. It offers at best a modus vivendi justification, that mutual toleration is the least worst option in current circumstances. In different circumstances – a less equal balance of power between competing confessions; better technologies of surveillance and control – intolerance might be the better bet. We want more out of a justification of toleration than the mere thought that, right now, no-one can safely get away with intolerance. I conclude that Hume's explicit justification of toleration needs, at least, to be supplemented.

So far, I have sketched the problem of toleration; offered three taxonomic questions for distinguishing theories of toleration; and given the examples of Rawls and of Hume's disappointing explicit theory, which I hope to complete by offering a political interpretation of Dialogues. Others have interpreted Dialogues politically,Footnote ⁸ but have gone wrong, in my view, in not seeing the politics in the wider context of Hume's thought, and especially in relation to his naturalism and scepticism. I therefore now continue to my second piece of background.

2. Naturalism and Scepticism

Hume was interpreted by his contemporaries, and up to the end of the Nineteenth Century, as a playful sceptic who takes the empiricism of Locke and Berkeley to its logical conclusion, gleefully destabilises our beliefs, but has no positive doctrine of his own to put in their place. Norman Kemp Smith's classic 1905 article ‘The Naturalism of Hume’Footnote ⁹ argued that, on the contrary, Hume is a rigorous naturalist who subordinates reason to inescapable instinct and feeling, and therefore rejects scepticism on the grounds that, although it is rationally irrefutable, it is impossible for creatures like us to believe it. We have natural beliefs, not subject to reason, in both metaphysical and moral arenas. More recent interpretation, for instance by David Fate Norton,Footnote ¹⁰ has tended to accept that ‘naturalist or sceptic’ is a false dichotomy. Hume is both; the hard question is, How are his naturalism and his scepticism related?

I shall approach that question by making a distinction between two different kinds of scepticism. The first is the familiar, Cartesian kind, about which the first thing to say is that Descartes is not a sceptic. He is a radical anti-sceptic who makes use of sceptical moves and tropes for his own purposes, and ‘the sceptic’ is a rhetorical character in his project of reconstructing all knowledge on firm foundations. Descartes's favourite metaphor is architectural: current knowledge is a condemned house which we must demolish in order to rebuild. Descartes's character ‘the sceptic’ has two roles in this drama. She is the demolition firm Descartes brings in to knock down the old house, and she is the opponent to beat for Descartes's new methods for gathering and defending knowledge. If those methods can defeat the sceptical monster Descartes has raised, they can certainly defeat the less extreme problems of ordinary scientific practice. The main point to draw from these obvious remarks is that to be a Cartesian sceptic is not a possible way of life for us, and is not intended to be. This character, ‘the sceptic’, plays two particular roles in Descartes's project of reconstruction, but is not presented as a real human possibility. No-one can adopt this form of scepticism as a way of life.

My second kind of scepticism is Pyrrhonism, and the first thing to say about it is that it does offer a way of life which we might adopt (or, more cautiously, it is intended to offer such a way of life). Pyrrhonism is a spiritual and therapeutic discipline aimed at ataraxia – tranquility – via epoche – suspension of judgement about opposing knowledge claims. The sceptical adept frees herself from disturbance by reminding herself of the equally compelling alternatives to each particular claim of knowledge that she encounters. Her mantra is neither one nor the other. The Pyrrhonian's focus, then, is on how to live without disturbance, not on how to know with certainty. Sextus Empiricus characterises the Pyrrhonian life as follows:

In what follows, I shall emphasise the contrast Sextus draws here between holding opinions and being subject to guidance and necessitation by nature and feeling.

How is this relevant to Dialogues? The character Philo can be understood as a Pyrrhonist. In Dialogues part 1 he initially appears to be a Cartesian sceptic, denying that there is any rational warrant for any belief whatsoever (actually being introduced to such a person would be rather like being introduced to Hamlet). Cleanthes makes the obvious objection to this position as a way of life:

Philo immediately retreats:

One way of reading this retreat is that Philo is advocating Hume's own mitigated or Academic scepticism.Footnote ¹⁴ But I want to focus, instead, on the echo of Sextus: we must live, and act, and converse; we are not able to be utterly inactive. On this reading, Philo is drawing the reader's attention to the immediate and inescapable demands of human life – the guidance and necessitation of nature and feeling. Pyrrhonists, according to Sextus, ‘are disturbed by things which are forced upon them; for we agree that at times they shiver and are thirsty and have other feelings of this kind.’Footnote ¹⁵

The Pyrrhonist does not hold opinions;Footnote ¹⁶ she is merely subject to certain demands. The answer to Cleanthes's common-sense argument against Cartesian scepticism is that belief is not the only thing which shapes life; it is not even the most important thing. Life is shaped by the necessary demands of nature, even in the absence of opinions or beliefs. There is no rational warrant for acting as if one really is a vulnerable human in a recalcitrant world, but no such warrant is required. The answer to the Cartesian sceptic's question is to recognise the inescapable demands of hunger, thirst, action and sociability.

The relationship between naturalism and scepticism, then – at least here, if not in all of Hume's work – is the Pyrrhonian one: scepticism insists on the inescapability of nature, rather than insanely trying to ignore it, or irrelevantly insisting on rational warrant. If this is right, then the irony of Kemp Smith's interpretation is that, by arguing against a sceptical reading of Hume, he turns Hume into a Pyrrhonian sceptic.

In summary: Pyrrhonism, unlike Cartesian scepticism, is a way of life. The lesson Hume dramatises with Philo's Pyrrhonian arguments in Dialogues 1 is that belief is not the only thing which shapes life: we are inescapably subject to the demands of the world and of being human. I now have my two pieces of background, and can go on to my reading of Dialogues.

3. Dialogues as Political Drama

Dialogues Concerning Natural Religion (first published posthumously in 1779) is a dialogue – a play script – closely modelled on classical philosophical dialogues, and especially on Cicero's On the Nature of the Gods.Footnote ¹⁷ It is divided into twelve parts plus a prologue: part 1 I have already discussed; parts 2–8 consider design arguments for the existence of God; part 9, a priori arguments; parts 10–11, the problem of evil. Part 12 is ambiguous, and my main focus here. The piece has four characters, two of whom I have already mentioned: Philo the jesting sceptic; Cleanthes the natural theologian and advocate of the design argument; Demea the orthodox, authoritarian Christian, often the butt of Philo's jokes; and Pamphilus, Cleanthes's ward and the narrator of the piece. Pamphilus takes little direct part in the Dialogues, offering only a framing narrative at the beginning and a partisan summation at the end.

Dialogues has typically provoked two interpretative questions, apart from the details of its arguments. First, Who is Hume? Which character argues for Hume's own position? Every character except Demea has been identified as Hume's distinct representative.Footnote ¹⁸ John Nelson has argued that Dialogues is a philosophical autobiography, with each character representing a different stage in the development of Hume's thoughtFootnote ¹⁹ (and Demea, perhaps, representing the priggish teenage Hume who made up lists of virtues to quantify his own and his classmates' moral worthiness). However, the majority view is that Hume's representative is Philo: he gets the lion's share of the lines and he wins most of the arguments. This identification raises the second interpretative question: If Hume is Philo, how are we to explain Philo's startling U-turn in Dialogues 12? In the main body of Dialogues, Philo has destroyed every argument for the existence of God put to him. The conclusion the reader expects is that, at best, we have no compelling reason for belief in God or for adopting religious practices. But in part 12, Philo suddenly announces:

This reversal has provoked a huge range of interpretations. Hume, masked as Philo, has been claimed to be an atheist, a deist, an agnostic, even a Spinozan pantheist.Footnote ²¹ It has been suggested that Hume thinks that belief in God is an inescapable natural belief like belief in causation or in a continuing self;Footnote ²² or that he is a sceptical fideist who strongly distinguishes faith from reason and ‘seeks to commend faith to us by emphasising the impotence of reason’.Footnote ²³ Part 12 has been read as just an ironic reemphasis of Philo's victory over religious belief, intended to be read in a heavily sarcastic tone of voice.Footnote ²⁴ It might be tempting to read it as advocating the view put into the mouth of Epicurus in section 11 of the first Enquiry: public religious performance for the sake of civil peace, regardless of private unbelief.Footnote ²⁵

Luckily for me, I don't need to resolve this second question, because I think that the first question – who is Hume? – is the wrong question. Dialogues is a play. Hume was a master at writing treatises and essays in a single voice – not always, or perhaps ever, his own voice – and had a deep concern for literary form and rhetorical strategy. The dialogue form he chose for this piece is not accidental: the message is in the whole, in the interactions of all the characters, not in any one character. Shakespeare does not appear in his own person in Hamlet; there is no need to ask whether Rosencrantz or Guildenstern is Tom Stoppard; and it is similarly a mistake to expect Hume to be behind just one of the masks he adopts in Dialogues.Footnote ²⁶

I suggest, further, that Dialogues is not only a drama: it is a political drama. There are two initial clues of political intent. First, questions of education for public life run all through Dialogues. Pamphilus, the narrator, is Cleanthes's student. Demea starts the conversation in Part 1 by stating a theory of education: his purpose in teaching his children is to tame their minds ‘to a proper submission and self-diffidence’Footnote ²⁷ so that they will be ready to receive the religion he will eventually impart to them without the self-confidence for, or tools of, rational defence. In general, the other characters are very aware of Pamphilus's presence and their educative effect on him. Second, Dialogues adopts the political vocabulary of Hume's explicitly political History of England, Natural History of Religion, and many of his essays on politics and religion: the vocabulary of superstition and enthusiasm, which Hume regards as pathologies both of belief and of social action.Footnote ²⁸ The main argument for a political reading, however, is that it organises and makes sense of what was previously difficult to understand.

I therefore offer the following way of accounting for Philo's U-turn, understanding the Dialogues as a whole, and articulating Hume's implicit theory of toleration: Dialogues dramatises a theory of toleration grounded in political Pyrrhonism. It is a picture of tolerant politics in microcosm which displays Philo and Cleanthes, who are friends, but who apparently disagree about almost everything, and especially over religious questions and values. They disagree about almost everything, but there are two exceptions.

First, both accept a ‘true religion’Footnote ²⁹ defined by exclusion. It excludes superstition, which for Hume is grounded in fear and weakness, and leads to tyranny; it excludes enthusiasm – fanaticism – which is grounded in passionate over-confidence, and leads to violence. Perhaps most importantly, it excludes the authority of priests and religious institutions:

True religion claims no authority over civil life. The main point of Dialogues 2-11 is that no claims about the character or commands of God (or the gods) have rational warrant. Hume's further point in part 12 is that those who claim to know anything about God's character and commands should be excluded from civil power. Demea is not to be allowed to impose his ‘established doctrines and opinions’.Footnote ³¹ True religion, then, is in practice indistinguishable from no religion. If it has any religious features, they are forms of words or public rituals, with which Philo is willing to play along for politeness's sake, but which are to have no influence on the communal use of power. There is an echo here of the Epicurean position described above: public performance is distinguished from private belief, and the latter is to have no influence on the former. But in this version, it is Demea's belief in a God who makes knowable commands which is excluded, rather than Epicurus's disbelief.

The second point that Philo and Cleanthes agree on is the source of morality. For Demea, the source of morality is the authoritative command of God, as heard and interpreted, of course, by Demea. For Cleanthes and Philo, the source of morality is the demands we are subject to as humans. The source is nature, not command: we must live, act, and converse; we must suffer thirst and feel cold; we must make our way as vulnerable bodies who need each other; we cannot help but react to selfishness and benevolence, or to politeness and contempt, as we do; we can be educated to react to justice and injustice similarly. Cleanthes and Philo, that is, share Hume's own naturalistic ethics as set out in the second Enquiry.Footnote ³² Cleanthes briefly defends religion in its ‘proper office’, when it makes no claims about supernatural authority and merely, silently ‘enforces the motives of morality and justice’.Footnote ³³ But Philo rapidly dismisses even this last half-hearted defence, and both come to agree that, at best, religion has no effect on morals: the effect of popular religion is ‘factions, civil wars, persecutions, subversions of government, oppression, slavery’,Footnote ³⁴ and ‘the smallest grain of natural honesty and benevolence has more effect on men's conduct than the most pompous views suggested by theological theories and systems’.Footnote ³⁵ When religion interferes with morals, as Demea wants, the result is pathological violence and tyranny.

This microcosm of friendship dramatises a macrocosmic politics: Hume is aiming, as a political Pyrrhonist, at social as well as personal tranquility. His therapeutic project is directed at politics as well as at personal life. I now return to my three, taxonomic questions about toleration, in order to set out the implicit theory I have uncovered in Dialogues.

1. 1. What justifies the shared core of belief or practice in the face of difference? What's inescapable for us: necessity, not belief. We cannot avoid the necessities of conversation, hunger, thirst, and living together. That core must resist certain centrifugal forces: the pathologies of belief and social action which are superstition and enthusiasm, and the authority-claims of priests and other interpreters of supernatural command.

2. 2. What must we refrain from enforcing or remain neutral about? The deliverances of priestly speculation and assertion. There are to be no divine commands in social life.

3. 3. Who gets in? Not Demea: he left in a huff at the end of part 11. And not, therefore, the orthodox authoritarians – in particular, the traditionalist members of the Kirk who had prevented Hume from getting university positions in Glasgow and Edinburgh – whom he represents. But both Philo the sceptic and Cleanthes the theist do get in, because they agree that distinctively religious claims are to have no effect on common life. They agree that the ground of politics and ethics is the demands of nature, not belief.

We can further articulate this theory by comparison with the other example I gave in section 1, Rawls. Both Hume and Rawls make a distinction between public and private: what we say and do in public, and especially in relation to public power, must be limited for the sake of civil peace. But there are two important differences between them. They differ, first, over the location of real life. For Rawls, our real lives are in our local, rich conceptions of the good and in the private communities – churches, for instance – held together by commonalities of belief. Politics, in contrast, is thin and constrained because we cannot reasonably expect others to share those rich conceptions of the good, those beliefs. This is one way in which Rawls opens himself to well-known communitarian critiques.Footnote ³⁶ For Hume, the contrast is the other way around: real life is human social life, the life in public. The speculative use of reason to support supernatural commands – reliance on belief rather than necessitation by nature – is strained, artificial, and dangerous. Giving in to the immediate demands of humanity is natural and easy. Second, as that implies, the regimes of toleration implied by Rawls and by Hume are differently justified. For Rawls, the justification appeals to shared belief in the ideals of the free and equal citizen and of reasonableness. For Hume, it appeals to shared necessity: the shared demands of being human in a recalcitrant world.

4. Nature as Justification

I said at the start of this paper that I aimed both at interpretation and at archaeology. I now move to the second of those aims, and try to show that the theory of toleration I have discovered in Hume is an attractive alternative to contemporary Rawlsian theory. Of my three taxonomic questions about the justification, the inclusions and the exclusions of a regime of toleration, the first is fundamental, and I concentrate on it here. I shall set out Rawls's strategy for justification, contrast it with Hume's naturalistic alternative, and consider some advantages and possible disadvantages of the latter. I intend only to make the contrast between Hume and Rawls clearer, and to offer some considerations which might tend to make us choose Hume of the two, not to refute Rawls.

A justification of some proposition or action is a decisive reason for believing that proposition or performing that action. The practice of justification, then, is concerned with discovering and conveying reasons, and showing that they outweigh or trump conflicting reasons. The distinction between Hume's and Rawls's accounts of the justification of toleration is a distinction between the kinds of reasons they take to be decisive for what we ought to believe and/or do about toleration.

Rawls's account of the justification of a moral or political regime changed over his career, in ways too complex to describe here. What became increasingly clear over the course of those changes, however, was that Rawls saw justification as an activity internal to a system of beliefs. Rawls's decisive reason for endorsing beliefs about toleration is consistency: denying those beliefs is inconsistent with beliefs about the freedom and equality of citizens; these beliefs about citizens are so central to our political culture that we cannot or will not give them up. The immediate practical task of political philosophy is to discover in our culture a basis for public agreement on first principles of justice, and those principles will embody toleration by being neutral between competing religious and secular conceptions of the good. We justify toleration by showing that our shared core of belief in the free and equal citizen requires us to leave space for difference over our answers to deep questions about value, and to refrain from enforcing our own answers to those questions.

Rawls's constructivist approach to the justificatory task is to give accounts of (1) the liberal-democratic concept of the citizen as free, equal, and possessing moral powers; (2) the well-ordered society; and (3) the original position. Rawls then moves from (1) to (2) – from an ideal of the citizen to a social ideal – via a process of construction involving (3). The original position dramatises commitments already implicit in our most firmly-held political beliefs, by showing us what treating citizens as free and equal requires – that is, by showing us how (1) the ideal of the person and (2) the social ideal fit together. The result of this process – justice as fairness – is not true by correspondence to any pre-existing moral order, but is objective in the special sense that it derives from a suitably constructed point of view, represented by the original position. The reason Rawls discovers for defending a tolerant regime, then, is that we are already committed to beliefs which we are unwilling to give up, and which are inconsistent with intolerance. Crudely: if Rawls is asked, Why should I believe in toleration?, he will respond, Because not to do so would render your political belief-system incoherent. His justification is consistency.Footnote ³⁷

Hume's justification of toleration, as I have discovered it in Dialogues, appeals not to consistency but to nature: we are vulnerable humans in a world which makes unignorable demands on us. I take it that these demands, unlike beliefs, are non-cognitive and non-referential. There is a difference in direction of fit: beliefs aim from us at the world; demands are the world taking aim at us. Demands are also reason-giving: if I am subject to a demand, I have reason to do certain things and not to do certain others. In some ways, all I can do is to resign myself to what is going to happen regardless of my beliefs and desires – I will eventually die, for instance. In other ways, I can respond intelligently to the various demands of nature by, for instance, doing what I can to maintain my health. In general, nature's demandingness makes it such that there are necessary conditions of achieving or maintaining one's good, and that these conditions are not liable to be annulled by what one believes about them. Demands, then, are causal rather than (merely) evidential, and the Humean concern is with what the world's demands are and what they do to us. The requirements we are subject to as humans are centering forces: they pull us toward a certain form of life which resists the centrifugal distortions of belief-motivated supersition, enthusiasm, and authoritarianism.

So far, my appeals to nature's demands have been impressionistic; it's time to do some clarifying and generalising work. The demands of nature which are important for political justification fall into two connected classes, needs and vulnerabilities. A need is a necessary condition of achieving some X, and a vital or basic need is a necessary condition of achieving X, where X = not suffering some serious harm.Footnote ³⁸ Humans need food and shelter in the sense that we will suffer the serious harms of starvation and exposure without them. Humans also need the company and conversation of other humans in this sense: without them, we suffer serious developmental and psychological harms. Vulnerabilities are those characteristics of humans which entail that we have needs. They are the specific ways in which we can suffer harm, including the various ways in which our bodies can be damaged, our personalities deranged, and our relationships with others rendered intolerable. We are vulnerable to being cut, broken, maddened, and enslaved. As the last possibility suggests, some of our most important vulnerabilities are to one another: some of the worst kinds of suffering, such as slavery, are inflicted by other humans. Further, as Thomas Hobbes argues, our equal vulnerability to one another is one of the central facts for political justification:

No matter what small advantages you enjoy over other humans, they can always gang up on you, and they know where you sleep. These remarks about human needs and vulnerabilities are, in a sense, banal; they have to be insisted on only in the context of a contemporary Rawlsian paradigm which understands justification as an appeal to the internal consistency of political belief-systems.

We are now in a position to compare the Humean with the Rawlsian account of the justification of toleration. I want to emphasise two advantages of Hume, before considering one possible disadvantage. Hume's first advantage is that the Humean strategy for justifying toleration has far wider scope than the Rawlsian. Rawls's account of justification is particularistic or anti-universalist: we start from where we are, embedded in a particular political culture, and do not attempt to extend its values to all times and places. In Rawls's later work it became increasingly clear that ‘we’ are the heirs of a liberal-democratic constitutional tradition – perhaps, even, just that tradition in the United States – and no-one else:

Hume's naturalistic justifications, in contrast, offer reasons to all humans: those who do not share the inheritance of the Declaration of Independence do share human needs and vulnerabilities. This is attractive, at least, for the practical reason that many of our most pressing disputes about toleration, and about political justice in general, are not internal to the United States and its penumbral culture.

The Humean's second advantage is her ability to weigh or rank different reasons against one another. Reasons of consistency can pull either way: if I have two beliefs which are inconsistent – say, belief in the liberal ideal of the citizen and belief in an intolerant theocratic regime – I can make my belief-system coherent by dropping or modifying either belief. The Rawlsian has no obvious basis for claiming that I must drop my second, intolerant belief and become a fully consistent liberal, instead of dropping my belief in the citizen and becoming a fully consistent theocrat.Footnote ⁴¹ The Humean, in contrast, can distinguish reasons derived from human needs and vulnerabilities from reasons derived from beliefs, and insist that the former are trumps (i.e. lexically prior), or at least that they have considerably more weight. This has the advantage of avoiding the ugly conclusion that, for instance, a consistent theocratic regime is as well-justified as a consistent liberal regime. It is not, to the extent that the theocratic regime harms people.Footnote ⁴²

As already noted, the Humean emphasises the centering force of nature's demands. On the Rawlsian picture, there are multiple islands of stability and coherence in the webs of people's beliefs, including islands which do significant harm. On the Humean picture, there is one centre which is focussed by the gravity of need and vulnerability and threatened by centrifugal forces.

One possible disadvantage of the Humean account is that the political proposals we can make on the basis of needs and vulnerabilities are far less specific and detailed than those the Rawlsian can make on the basis of consistency with the ideal of the citizen. The original position is designed to articulate this liberal conception of politics, and is supposed to result in Rawls's precise principles of justice. In contrast, the Humean seems to be left with the rather less precise suggestion that we refrain from hurting people. The demands of nature do not apparently justify any policy proposal as specific as, for instance, that the basic structure of society should satisfy the difference principle in its distribution of goods.Footnote ⁴³ However, I am not certain that this is a big difference, nor therefore much of a disadvantage. If the original position allowed Rawls to show that commitment to the liberal ideal of the citizen entails commitment to toleration, then to compete, the Humean would have to offer some similar derivation, and that looks unlikely. But Rawls is explicit that he cannot show such an entailment: the original position allows us to make a reasonable choice between some major alternatives in the history of moral and political thought, not to derive the principles of justice from the ideal of the citizen.Footnote ⁴⁴ So, the Humean is not in a worse position: she also can appeal to history, and especially to the history of tolerant and intolerant regimes, to argue for the superiority of the former in relation to human needs and vulnerabilities. She does not need to derive an account of a particular regime solely from those needs and vulnerabilities.

A great deal more could be said in defence of Rawls and the Rawlsian strategy. I have not tried to refute Rawls, but only intend, as noted above, to offer some considerations which might incline us to the Humean naturalistic rather than the Rawlsian mode of justification.