I
G. A. Cohen has argued that
(C): ‘[A normative] principle can reflect or respond to a fact only because it is also a response to a principle that is not a response to a fact. To put the same point differently, principles that reflect facts must, in order to reflect facts, reflect principles that don't reflect facts.’Footnote 1
The terms used in C are defined as follows. A normative principle (following Cohen's usage hereafter referred to, for short, as a ‘principle’) is ‘a general directive that tells an agent what (they ought, or ought not) to do’.Footnote 2 A fact is ‘any truth other than (if any principles are truths) a principle’.Footnote 3
Cohen gives the following example to illustrate C. Suppose someone affirms the principle ‘we should keep our promises (call that P) because only when promises are kept can promisees successfully pursue their projects (call that F)’.Footnote 4 Cohen argues that this person has to concede that she affirms the further principle P 1 which says that ‘we should help people to pursue their projects’.Footnote 5 It is P 1 that makes F a ground for P. Fact-sensitive principles are principles which have facts among the grounds for affirming them.Footnote 6 It is Cohen's thesis that the scrutiny of fact-sensitive principles will always reveal a deeper fact-insensitive principle that explains why a fact grounds a certain principle. According to Cohen, Rawls denies C. Furthermore, the Rawlsian construction is inevitably fact-dependent. As Rawls concedes, the denizens of the original position (hereafter OP) cannot make a decision in the absence of facts.Footnote 7 Since, according to C, someone who affirms a principle in light of facts is committed to another principle, Rawlsians are mistaken about the structure of their beliefs. There is a fact-insensitive principle Rawlsians are committed to but have failed to explicate.
I shall argue that Rawls does not deny C, but argues for an altogether different thesis. To distinguish Cohen's thesis from Rawls's thesis, I'd like to introduce some terminology. A principle is metaphysically fact-insensitive if its truth doesn't depend on any facts. A principle is epistemically fact-insensitive if our belief in it doesn't depend on knowledge of facts. Principles, such as Rossian prima facie duties, which we are inclined to affirm without knowledge of facts, are examples of epistemically fact-insensitive principles. Cohen's central claim is that there are metaphysically fact-insensitive principles. I shall argue that Rawls's central claim is that metaphysically fact-insensitive principles and epistemically fact-insensitive principles need not coincide, and in fact epistemically fact-insensitive principles can be metaphysically fact-sensitive. I shall also argue that Rawls acknowledges the existence of principles which underlie OP, but these principles are not principles of distributive justice.
Rawls's constructivism and his comments on the basis of which Cohen argues that Rawls denies C are best understood and evaluated within a general framework of moral theorizing. Accordingly, in section II, I present a brief account of how we develop our moral theories, and what our goals in theorizing are, which is in line with Rawls's approach. In section III, I offer an alternative interpretation of the passages from Rawls that Cohen uses in arguing that Rawls denies C. Section IV develops this interpretation further and defends Rawls's claims about the relationship between facts and principles. I argue that what Rawls rejects is approaches which rule out the use of facts in the derivation of higher-level moral principles from lower-level ones, and the use of facts in justifying a theory to us. Rawls, like many utilitarians, denies that it is an objection to a moral theory that it fails to capture our intuitions about cases which cannot arise in our world. This is not because a theory's failure to capture such intuitions doesn't make a practical difference, but because Rawls, like some utilitarians, refuses to take such intuitions at face value. It is possible that even though Rawls does not deny C, OP only has force if C is false. Therefore, a separate defence of OP and the use of facts in it are necessary. This task is taken up in section V. I argue that Rawls does not deny that metaphysically fact-insensitive principles lie behind OP and that the use of facts in OP does not entail that the principles which emerge out of OP are epistemically fact-sensitive principles.
Three caveats. My defence of Rawls's constructivism applies only to the version of constructivism in A Theory of Justice. In later works, the role of constructivism within Rawls's theory, and the questions Rawls addresses in his theory, change substantially. The version of constructivism I defend is also the one Cohen takes issue with. Second, my defence is limited to what Sharon Street has called ‘restricted constructivism’.Footnote 8 Restricted constructivism is not a full-fledged metaethical position and applies only to a restricted domain within practical reason. Third, my goal is not to defend the content of Rawls's theory, but to defend its structure, which is what is at stake in Cohen's critique. Accordingly, I assume that Rawls's derivation of his principles from OP is correct, and assume that Rawls's theory has the theoretical virtues which I argue in section II we require from moral theories.
II
In ‘Outline of a Decision Procedure for Ethics’ Rawls describes his project as discovering principles that explicate the considered judgements of competent judges. These principles would be
such that, if any competent man were to apply them intelligently and consistently to the same cases under review, his judgments, made systematically nonintuitive by the explicit and conscious use of the principles, would be, nevertheless, identical, case by case, with the considered judgments of the group of competent judges.Footnote 9
The process of discovering moral principles, on this account, is inductive. We begin by individual judgements. We then develop lower-level principles which account for these judgements. Finally, we seek higher-level principles that bring together our lower-level principles. This cannot be the whole picture, because, as Rawls came to see by the time of Theory, we also hold higher-level principles. In some instances we may be more confident of these general principles than our specific judgements. Therefore, we need a structure that systematizes our moral beliefs at different levels of generality.
When the principles we already hold or those we have discovered conflict with our evaluative judgements, we need to make revisions. We are not equally certain of all the judgements we make and all the principles we hold. We feel some of them may be the products of our upbringing, our culture, or reflect the views of the social groups to which we belong. Theorizing acts as a critical apparatus that eliminates some of these judgements. Once we have uncovered the higher-level principles underlying our considered judgements and lower-level principles, we may revise some of our judgements in light of these principles or we may revise our principles. At each point of conflict between our considered judgements and principles, we face a choice. We may either revise our considered judgements or the principle we have formulated, depending on which of these we find to be more credible on reflection. Through this process of revision from both ends, we reach what Rawls has called ‘reflective equilibrium’.Footnote 10 The theory we are pursuing is not one that can account for most of our considered judgements, but the one that accounts for them in reflective equilibrium. After theorizing we end up with a set of beliefs different from the one we began with and of which we are more confident.
In theorizing, we also want a deeper explanation of the moral beliefs we hold. As Sidgwick points out, even when we have accepted principles which formulate our beliefs, and we do not doubt their truth, we want a ‘deeper explanation of why it is so’.Footnote 11 Different moral theories such as contractualism and utilitarianism provide us both with principles that they claim match our judgements and with an explanation of these judgements.Footnote 12 Sidgwick, for instance, puts forward the explanatory hypothesis that common-sense morality is a ‘machinery of rules, habits, and sentiments, roughly and generally but not precisely or completely adapted to the production of the greatest possible happiness for sentient beings generally’.Footnote 13 Rawls's theory can (roughly) be interpreted as saying that our intuitions about and lower-level principles of distributive justice are accounted for by the principles which are justifiable to each individual. The explanation we seek is one that also has justificatory force. The principle or set of principles that accounts for our beliefs should have independent appeal. Once we are aware of these principles we should be able to endorse them and our moral beliefs. By a moral theory, then, I understand a single principle or a set of higher-level principles that account for other lower-level principles and intuitive judgements, and have explanatory and justificatory power.
By way of summary, I would like to propose this analogy. We can think of our belief system in economic terms. Each inconsistency, unexplained moral phenomenon and case we cannot respond to is a cost. Each belief is an asset. The values of beliefs vary in terms of the level of credibility we attribute to them and their connectedness. We want to maximize the value of our assets and minimize our costs. Faced with two inconsistent beliefs, which have equal levels of connectedness, we will discard the one that has less credibility. Faced with two inconsistent beliefs that have equal levels of credibility, we will discard the one that has less connectedness, because discarding it will require less revision in our belief system. There may be elements of coherentism on this view, but it is compatible with there being beliefs that enjoy such high levels of credibility or connectedness that we consider them to be fixed. We have three goals: (a) to minimize inconsistencies; (b) to conserve the beliefs we already hold as much as possible; and (c) to maximize the extension of our beliefs, that is, to be able to respond to as many cases as possible. This procedure does not ensure that the beliefs we end up will be true. However, through this procedure we will have conducted our inquiry rationally, and our beliefs will be justified.
III
Rawls's most significant remarks concerning the employment of facts in moral theories, which are also cited by Cohen to argue that Rawls denies C, occur in his response to a possible objection to the difference principle. The objection is that the difference principle makes ‘the justice of large increases or decreases in the expectations of the more advantaged depend upon small changes in the prospects of the worst-off’.Footnote 14 For instance, the difference principle may allow huge inequalities if they are necessary for improving the condition of the worst-off even though the improvements in how the worst-off fare are infinitesimally small. Rawls responds to this objection by arguing that such cases will not arise. In defence of his use of a factual claim he says:
We should also observe that the difference principle not only assumes the operation of other principles, but it presupposes as well a certain theory of social institutions. In particular . . . it relies on the idea that in a competitive economy (with or without private ownership) with an open class system excessive inequalities will not be the rule . . . Now the point to stress here is that there is no objection to resting the choice of first principles upon the general facts of economics and psychology. As we have seen, the parties in the original position are assumed to know the general facts about human society. Since this knowledge enters into the premises of their deliberations, their choice of principles is relative to these facts. What is essential, of course, is that these premises be true and sufficiently general. It is often objected, for example, that utilitarianism may allow for slavery and serfdom, and for other infractions of liberty. Whether these institutions are justified is made to depend upon whether actuarial calculations show that they yield a higher balance of happiness. To this the utilitarian replies that the nature of society is such that these calculations are normally against such denials of liberty.
Contract theory agrees, then, with utilitarianism in holding that the fundamental principles of justice quite properly depend upon the natural facts about men in society. This dependence is made explicit by the description of the original position: the decision of the parties is taken in the light of general knowledge.Footnote 15
In this passage, Rawls conflates two different perspectives. He shifts back and forth between our perspective, that is, the perspective of someone trying to decide whether the principles which emerge out of OP formulate her convictions, and the perspective of the denizens of OP. In mentioning the fact that there will not be excessive inequalities in competitive economies, Rawls is showing that certain distributions which we would consider unjust will not be considered just by the difference principle. He then mentions that the choice of the denizens of OP is sensitive to facts. However, the denizens of OP do not assume that such inequalities will not be the rule in order to choose the difference principle. In fact, even if they do not make Rawls's empirical assumption, they would still adopt the difference principle. If maximin is the correct decision rule in OP and the denizens of OP are not motivated by envy, then, they have no reason to object to inequalities which may arise even if they do not assume that there will not be excessive inequalities in competitive economies. It is we evaluating the principles the denizens of OP have chosen who have to make this empirical assumption so that we are assured the difference principle does not give rise to inequalities we find objectionable.
The agreement between contract theory and utilitarianism cannot be about the use of facts in OP, because utilitarianism, with the exception of Harsanyi's version of it, does not use such a device. Furthermore, utilitarians don't think that their principle is made true by certain facts obtaining. The parallel Rawls draws in the passage quoted isn't between his two principles of justice and utilitarianism, but between contract theory and utilitarianism. According to both contract theory and utilitarianism, principles of distributive justice depend upon facts. Standardly, utilitarians don't offer utilitarianism as a principle of justice. Nevertheless, they have an account of justice. Rawls outlines the utilitarian strategy of accounting for principles of justice, which is based on Mill's account, as follows:
[F]rom a utilitarian standpoint the explanation of these precepts [of justice] and their seemingly stringent character is that they are those precepts which experience shows should be strictly respected and departed from only under exceptional circumstances if the sum of utility is to be maximized. Yet, as with all other precepts, those of justice are derivative from the one end of attaining the greatest balance of satisfaction.Footnote 16
Rawls's parallel strategy is to argue that principles of distributive justice are the principles which would be chosen in OP in light of facts, and that our common-sense precepts of justice and beliefs about the stringency of demands of justice are accounted for by contract theory.Footnote 17
Contract theory and utilitarianism agree that we can assume certain facts hold when testing the principles they have proposed against our intuitions or in deriving higher-level principles from lower-level ones. They both agree with a theoretical strategy that says: ‘Here are your fundamental principles of justice: When conjoined with facts about the world, they will generate your evaluative judgements. It is irrelevant whether they will fail under factual assumptions which do not actually hold.’Footnote 18 As I shall elaborate below, both utilitarianism and Rawls claim that moral beliefs, such as our beliefs about distributive justice, which are epistemically fact-insensitive, can be metaphysically factsensitive.
Cohen also quotes Rawls's statement that ‘Conceptions of justice must be justified by the conditions of our life as we know it or not at all’ to bolster his claim that Rawls denies C. I'd like to offer an alternative interpretation of that passage for the sake of completeness. In that paragraph, Rawls is saying that if false assumptions are necessary for principles to be justified to those governed by them, then this is unacceptable. And if false assumptions are needed for the denizens of OP to choose a principle, then this is also unacceptable. What Rawls says there may commit him to the claim that principles may depend on facts, and that principles of distributive justice do, in fact, depend on facts, but Rawls clearly is not arguing that all principles are fact-sensitive. What Rawls says here seems to be nothing more than ‘No sound principle is grounded in false beliefs’ rather than ‘All principles are grounded in facts’.Footnote 19
IV
Cohen is interested in the logical relations between facts and principles. Accordingly, he assumes that the interlocutors in the examples he uses to illustrate his thesis have a clear grasp of what their principles and the grounds for these principles are.Footnote 20 Cohen's interlocutors have, as it were, carried out their theorizing and have a view regarding the structure of moral reality. Cohen is defending the thesis that there are metaphysically fact-insensitive principles. Rawls, however, is defending a thesis about how we should go about our theorizing. Rawls's thesis is that when theorizing we can assume that certain facts hold, and principles which are epistemically fact-insensitive need not be metaphysically fact-insensitive.
If we return to Cohen's example I cited in section I, we are to imagine someone who affirms the principle that we should keep our promises (P) because ‘only when promises are kept can promisees successfully pursue their projects’ (F). Given her previous reason, this person, as Cohen argues, has to also affirm the principle ‘we should help people to pursue their projects’ (P 1). F's obtaining makes P true. Someone who makes this claim commits herself to affirming P1, which is a higher-level principle. To illustrate Rawls's position, I would like to examine the point of view of someone in the process of moral theorizing, and examine the possible ways in which P, F and P 1 may be related in the belief structure of a typical inquirer who is in the process of theorizing.
In section II, I suggested that our aim in theorizing is to develop a unified account of the moral beliefs we hold at different levels of generality. On this account, the support for holding P may be the generalization from the observation of the wrongness of individual instances of breaking promises to P, the independent plausibility of P, or its derivation from higher-level principles. In the case of a typical inquirer, the main support for P comes from its apparent plausibility. If this is the case, what is the function of P 1? P 1 is a higher-level principle than P. Therefore P 1 accounts for more evaluative judgements. P1 may play a role in interpersonal justification or a justificatory role in the beliefs of a single individual. To see the role of P1 in interpersonal justification imagine we are trying to convince someone that we should keep our promises. This person holds the set of moral beliefs E. E does not contain any judgements regarding promising, thus does not contain P. Suppose further that the evaluative judgements in E are best accounted for by P 1. By showing that P is entailed by P 1, which accounts for this person's other judgements, we draw support for P. This may lead a person who holds E to revise his principles and come to hold P and P1.
P1 may also play a justificatory role for a single individual. A person who already holds P may come to hold P 1 given F and P 1's ability to account for other lower-level principles this person holds. A person who holds both P and P 1 independently will be more confident of his principles if F holds. Similarly, a person who holds only P 1 may come to hold P when F obtains. The transfer of conviction in this account is multi-directional. We may come to hold lower-level principles because they are entailed by higher-level principles we hold, we may come to hold higher-level principles because they account for our lower-level principles, or we may become more confident of our principles because they mutually support each other.
Given that the transfer of conviction is multi-directional, it is difficult to identify which principles are epistemically fact-sensitive and which ones are not. Suppose a person finds lower-level principles P 2, P 3 and P 4 plausible. There is a higher-level principle P 5 that she also finds plausible. P 5 accounts for P 2, P 3, and P 4 when F holds. If she believes that F holds, then her confidence in P 2, P 3, P 4 and P 5 will have increased. In one sense, all of these principles are epistemically fact-insensitive, because she holds them without reference to facts. However, her increased confidence in them is due to facts. It is also possible that P 5 conflicts with one of the lower-level principles when not-F holds. In such a case, the set containing P 2, P 3, P 4 and P 5, or to be more precise the consistency of this set of beliefs, would be epistemically fact-sensitive even though the principles by themselves are not.
Alternatively, she may find P5 neither plausible nor implausible on its own, but come to hold it given its ability to account for P 2, P 3 and P 4. Sometimes we arrive at our higher-level principles inductively. We move up to higher-level principles when we are able to subsume our lower-level principles under these higher-level ones. This is possible only when we establish conceptual connections or facts enable us to subsume our lower-level principles under higher-level principles. In so far as our higher-level principles are not given, we may need facts to develop them from lower-level ones. If higher-level principles are arrived at inductively by relying on facts, they will be epistemically fact-sensitive. This is another way of construing the Rawlsian reliance on facts and the reliance on facts that, according to Rawls, exists in utilitarianism. Rawls and utilitarianism both agree that (a) we can use facts in moving from lower-level principles to higher-level ones; (b) we can rely on facts in ensuring a match between higher and lower-level principles; and (c) we can hold epistemically fact-sensitive sets of principles.
It is worth reiterating that I am not denying the existence of metaphysically fact-insensitive principles. The point is rather that there is, often, a mismatch: metaphysically fact-insensitive principles will tend to be higher-level and epistemically fact-sensitive, whereas epistemically fact-insensitive principles will usually be lower-level principles, and not metaphysically fact-insensitive. It is this thesis that Rawls and utilitarians, as Rawls interprets them, hold. Suppose the account of promising in Cohen's example is correct.Footnote 21 Then the transfer of conviction in the case of a typical inquirer would not correspond to the structure of moral truth. Speaking metaphorically we can say that truth flows from higher-level principles to lower-level principles whereas conviction flows in both directions, and often from lower-level principles to higher-level ones. Usually, our theorizing begins from lower-level principles and intuitive judgements. We arrive at higher-level principles by employing facts. Furthermore we are often more confident of lower-level principles than we are of higher-level principles. Consequently, if it was shown that promises do not necessarily contribute to people's successful pursuit of their projects, we would remain committed to P, and withhold judgement on P 1.
The role of facts in theorizing I've attributed to Rawls is exemplified by Brad Hooker's version of rule-consequentialism. Hooker argues for his version of rule-consequentialism by claiming that it can account for our lower-level principles.Footnote 22 This argument relies on facts, and we will hold the principles he proposes in light of facts. However, he claims that it is his theory of rule-consequentialism and not our lower-level principles that hold in all possible worlds.Footnote 23 In other words, Hooker claims that it is the epistemically fact-sensitive higher-level principles he proposes which are metaphysically fact-insensitive. Rawls wants to remain as non-committal as possible on metaphysical questions. Nevertheless, he gestures towards a strategy similar to Hooker's, but then takes it back. He suggests that even though the principles that emerge out of OP are contingent since they depend on facts, the conditions OP embodies may be held to be necessary truths, but – here he takes it back – it is better to regard them ‘simply as reasonable stipulations’.Footnote 24
Should we prefer a theory that is less reliant on facts in capturing our lower-level principles and intuitions about specific cases? I believe there is no clear-cut answer to this question. A variety of considerations has to be taken into account. For instance, we need to consider how certain we are of the facts that have been employed. We aren't likely to have qualms about employing facts we are highly certain of in our theorizing. Another relevant point of comparison between a theory which does not employ facts and one which does is how well our goals in theorizing are served by these theories. We need to ask whether the theory in question provides responses to difficult cases, or whether it gives us a more systematic account of our moral beliefs. We also need to consider the scope of the facts we employ in our theorizing. Some facts about human nature and society apply only during certain periods of human history. We might hesitate to employ such facts in our theorizing. However, reliance on facts which have characterized human life and society in all epochs does not seem problematic. We can perhaps impose a more general requirement. Using the language of possible worlds, we can say that we do not require our moral theories to capture our intuitions about all possible worlds, but to capture our intuitions about a range of possible worlds that are close to ours.Footnote 25 In short, there is no answer prior to specific theories that suggests we should prefer less fact-dependent theories.
Imagine this scenario. One philosopher says to her opponent, ‘Yes, under the usual conditions of human life, your theory captures and systematizes all of our moral beliefs, and provides illuminating answers in difficult cases. My theory accounts only for our intuitions about promises. But your theory can't capture our intuitions about cases regarding promises made in a world in which people's memory only reaches back to the previous week.’ That this theory has a claim to capture our intuitions about cases which may arise in all possible worlds, or at least in more possible worlds than its rival, does not seem to be a reason which commends it. Furthermore, given the success of the more general but fact-dependent theory, we are likely to doubt the reliability of our intuitions about promises in a world so different from ours.
The method of deriving higher-level principles from more specific principles or evaluative judgements through the employment of facts is consistent with the existence of higher-level principles we already have. As we have seen, Rawls rightly maintains that we have considered judgements at all levels of generality among which we seek reflective equilibrium.Footnote 26 The picture is one where principles at various levels of generality mutually support each other, and facts are brought into the mix. A procedure whereby we derive all of our specific judgements from a few higher-level principles without employing any facts would be needed to show the superiority of theorizing which does not rely on facts. Alternatively, these higher-level principles would have to be strong enough to override our lower-level principles and intuitive judgements. There is no way of ruling out the possibility of discovering such principles. However, there are no likely candidates at present. Alternatively the project of theorizing as I presented it in section II may be rejected. This would be an impoverishment of moral and political philosophy which few would find palatable.
This account of the relationship between facts and principles sheds light on another passage quoted disapprovingly by Cohen. Rawls writes:
Some philosophers have thought that ethical first principles should be independent of all contingent assumptions, that they should take for granted no truths except those of logic and others that follow from these by an analysis of concepts. Moral conceptions should hold for all possible worlds. Now this view makes moral philosophy the study of the ethics of creation: an examination of the reflections an omnipotent deity might entertain in determining which is the best of all possible worlds.Footnote 27
In this passage Rawls probably has Leibniz's ethics in mind.Footnote 28 The idea in Leibniz is that since God is omniscient and omnipotent he would create the best of all possible worlds. The ethics of creation is the inquiry into the principles that would have guided God's creation and his choice of the best of all possible worlds. The study of ethics as Rawls conceives it is an inquiry into the principles that underlie the evaluative judgements we make in this world. Rawls is not denying that there are metaphysically fact-insensitive principles. He only claims that we can take facts as part of the data for our theories. What Rawls's theory seeks to explain is the evaluative judgements we make in this world, and accordingly facts can be used in our theorizing. His theory does not seek to capture our moral judgements about cases which could not arise in our world. If our higher-level principles were given, or derived solely by establishing conceptual links between them and our lower-level principles and evaluative judgements, and if these judgements and lower-level principles were either self-evident, or somehow reflected some timeless and unchanging truth, then they would hold in all possible worlds. This is what is meant by the reference to possible worlds.Footnote 29
V
In the previous two sections, I argued that Rawls does not deny C and defended the way he conceives of the relationship between facts and principles. Even if this defence is successful the truth of C might threaten the conceptual coherence of OP. To defend the structure of Rawls's theory we need to show that no such threat exists. In this section, I shall argue that constructivist procedures which are reliant on facts can capture epistemically fact-insensitive principles and the principles which underlie OP are distinct from the principles which emerge from it.
It is true that the denizens of OP cannot make any choices in the absence of facts. However, this does not mean that a fact-dependent constructivist procedure like OP cannot capture our epistemically fact-insensitive principles. This point can be illustrated with Kant's categorical imperative procedure. The categorical imperative is a constructivist procedure. It asks whether a certain maxim can be universalized. In thinking whether a maxim can be universalized we employ our knowledge of facts. For instance, when the procedure is applied to promising, it rejects a maxim such as ‘I am to make a deceitful promise whenever it serves my purpose’. The test of this maxim assumes that people learn from past experience, and remember the past.Footnote 30 In this way, the procedure is fact-sensitive. We may, nonetheless, hold its outcome without reference to any facts as an epistemically fact-insensitive principle. I believe that, for most people, the duty not to make false promises is one that they hold independently of facts and is one of the starting points in moral theorizing. It is for most of us an epistemically fact-insensitive principle that is captured by a fact-sensitive constructivist procedure.Footnote 31
It will be asked why bother with the constructivist procedure if it all depends on whether its outcome matches one's considered convictions. This is not the case. The constructivist procedure is not required to capture all of our intuitions. In fact, it is expected that in some cases when the principles which emerge out of the constructivist procedure conflict with principles we currently hold, it is the principles that emerge from the constructivist procedure that we come to hold. Nevertheless, the constructivist procedure has to be able to capture our strongly held convictions, which we may hold independently of facts. For this reason, it was necessary to show that the dependence of constructivist procedures on facts does not entail that the principles they produce will be epistemically fact-sensitive principles.
OP itself also has to be acceptable to us and play an explanatory role if it is going to play a significant part in our theorizing, and make us give up some of the principles we hold prior to theorizing. At this point a second objection kicks in: ‘According to C, whenever a fact grounds a principle, there's another principle which explains why that fact grounds the principle in question. You grant that OP is dependent on facts. So, aren't the principles that underlie OP your fundamental fact-insensitive principles of justice? Aren't you mistaking principles which are formulated in light of your fact-insensitive principles and facts, for your ultimate principles of justice?’
To respond to this challenge, the Rawlsian can grant that fact-insensitive principles lie behind OP. However, she can deny that those principles are her fact-insensitive principles of distributive justice. The Rawlsian adopts OP in light of his (possibly) fact-insensitive principles of pure procedural justice, and the principle that people ought to be treated as equals. It has been observed by many commentators that some versions of utilitarianism, rights-based theories like Nozick's and egalitarian theories like Rawls's, offer competing interpretations of how people are to be treated as equals.Footnote 32 In this context, Rawls's constructivism can be interpreted as offering support for a controversial interpretation of equality through the device of OP, which relies on uncontroversial beliefs about procedural justice and the fundamental equality of human beings. In a letter to Nagel, quoted in Nagel's ‘Equality’, Rawls makes precisely this point:
Suppose we distinguish between the equal treatment of persons and their (equal) right to be treated as equals. (Here persons are moral persons.) The latter is more basic: Suppose the Original Position represents the latter re moral persons when they agree on principles and suppose they would agree on some form of equal treatment. What more is needed?Footnote 33
This, then, is the structure of Rawls's theory. We have principles of procedural justice and the principle ‘People ought to be treated as equals’, which are employed in the construction of OP. These principles can be epistemically and metaphysically fact-insensitive. The denizens of OP choose principles of distributive justice in light of facts. The principles of distributive justice which emerge may correspond to both our epistemically fact-insensitive and epistemically fact-sensitive principles. In cases where there are conflicts between our existing moral beliefs and the principles that emerge out of OP, factual assumptions may be used to achieve a match. If factual assumptions are required to achieve this match, then our moral beliefs when we have reached reflected equilibrium will be an epistemically fact-sensitive set of beliefs.
Is the preceding account compatible with Rawls's own understanding of his theory? I have argued that Rawls does not deny C, and puts forward an altogether different thesis than not-C. The following passage from Rawls gives further support to this interpretation and shows that Rawls does not deny that moral principles, which may not be fact-sensitive, lie behind OP. Rawls writes:
[B]oth general facts as well as moral conditions are needed even in the argument for the first principles of justice . . . In a contract theory, these moral conditions take the form of a description of the initial contractual situation. It is also clear that there is a division of labor between general facts and moral conditions in arriving at conceptions of justice, and this division can be different from one theory to another.Footnote 34
Here, Rawls clearly accepts that moral beliefs are embodied in OP and nothing he says suggests that those beliefs are held in light of facts, or that they are true only because certain facts obtain. This shows that Rawls, at least in Theory, did not hold the view Cohen attributes to him, and the theoretical structure I've attributed to Rawls is compatible with his own understanding of his theory.
Rawls also does not deny that we have beliefs about distributive justice prior to his theory. The goal of his theory is to systematize and give further support to these beliefs, and revise some of them. As we saw, Rawls holds that the difference principle ‘relies on the idea that in a competitive economy (with or without private property) with an open class system excessive inequalities will not be the rule’.Footnote 35 Commenting on this passage, Cohen argues that if Rawls ‘appraised facts differently, he would reject the difference principle, because it permitted too much inequality’.Footnote 36 From this, Cohen concludes that there is a fact-insensitive background principle to which Rawls is committed, but has failed to articulate. It is true that Rawls is committed to an epistemically fact-insensitive principle according to which too much inequality is objectionable. However, this doesn't commit him to the claim that too much inequality is objectionable in all possible worlds.
That we find some inequalities objectionable is quite explicit in Rawls's account. His goal is to show that his theory captures these beliefs. If his theory failed to capture this belief, then we would not have achieved reflective equilibrium. The belief set which includes the difference principle, and the belief that too much inequality is objectionable, is an epistemically fact-sensitive set of beliefs. Let us suppose that this fact holds, and Rawls's theory is successful in capturing our moral beliefs in this world: every inequality we find objectionable in this world is condemned by the difference principle. What should give way in a possible world in which huge inequalities are necessary to improve the condition of the worst-off? There are two different ways of treating this case. First, we can take our intuitions about such worlds at face value. If we still find such inequalities troubling, then the theory has to give way. This is the approach to which Rawls and utilitarians, as Rawls interprets them, object. Alternatively, we may refuse to take such intuitions at face value. We may query whether we would have the same intuition in a world in which huge inequalities were necessary, and how reliable our intuitions about worlds very different from ours are. This would amount to pointing out that the data by which the theory is to be tested, our intuitions in that possible world, are epistemically unavailable to us. This is the route which Rawls favours. Note that this is a weaker claim than the one I attributed to Hooker in section IV. It does not argue that the theory applies in all possible worlds, but merely shifts the burden of proof onto those who claim that their intuitions hold in all possible worlds.
VI
I have tried to show that in the passages Cohen uses to argue that Rawls denies C, Rawls does not deny C, but argues for a different claim. Rawls believes that we may legitimately use facts in deriving higher-level principles from lower-level ones, or hold fact-sensitive sets of principles. Along with this, Rawls believes that epistemically fact-sensitive principles are not necessarily metaphysically fact-insensitive principles. This allows Rawls to claim that principles of distributive justice are metaphysically fact-sensitive though they may be epistemically fact-insensitive.
I then examined the possibility that even though Rawls does not deny C, the truth of C poses difficulties for OP. There were two possible challenges. One was that because constructivist procedures rely on facts, the principles that emerge out of such procedures would have to be epistemically fact-sensitive principles. In response to this challenge, I showed how Kant's categorical imperative procedure could generate the duty that we should not make false promises, which is an epistemically fact-insensitive principle for most people. The second challenge was that Rawlsians are mistaken about the structure of their beliefs. What they take to be principles of justice are versions of their fundamental fact-insensitive principles that have been watered down with facts. In response to this challenge, I argued that the principles which underlie OP are principles of procedural justice and the principle that people ought to be treated as equals, whereas the principles which emerge are principles of distributive justice. Therefore, the principles that emerge out of OP are not the straightforward application of some principles to facts.
This account of the relationship between facts and principles also provides Rawls with a way to resist Cohen's charge that the fundamental error of Theory is ‘that it identifies the first principles of justice with the principles that we should adopt to regulate society’.Footnote 37 Rawls can grant that it may seem, pre-theoretically, that principles of distributive justice are distinct from other concerns such as stability, publicity and feasibility that enter into the deliberations of the denizens of OP. However, it may turn out that our principles of distributive justice, which are epistemically fact-insensitive, reflect such concerns. This would be analogous to the utilitarian claim that even though considerations of justice seem distinct from considerations of utility, they can be shown to flow from considerations of utility.Footnote 38
