The range of John Rawls's contribution to ethics and political philosophy is vast. At the more abstract level, he wrote about the ideas of reflective equilibrium; moral and political constructivism; public reason; and argued that moral theory is largely independent of epistemology, metaphysics and the philosophy of mind.¹ More concretely, he defended the model of a property-owning democracy; critically reflected on the US constitutional history of free speech; and argued that “both Hiroshima and the fire-bombing of Japanese cities were great evils.”² What ties these and many other disparate concerns together is the idea of the original position. By considering the choice from the original position, Rawls holds, we can bring our more abstract commitments together to generate principles of justice that can then be applied to more concrete cases. Conversely, the original position can help ensure that our more abstract commitments are consistent with the considered particular judgments about which we are most confident. Considering the choice from the original position may “provide guidance where guidance is needed” (TJ, 18) and help us move in the direction of reflective equilibrium by serving as “a mediating idea by which all our considered convictions, whatever their level of generality … can be brought to bear on one another” (PL, 26; cf. JF, 39, 81). Justice as fairness represents “the hypothesis that the principles which would be chosen in the original position are identical with those that match our considered judgments and so these principles describe our sense of justice” (TJ, 42).

Although he did not introduce the idea of the original position until 1963, its roots can be found in a problem that Rawls faced in his 1950 dissertation, “A Study in the Grounds of Ethical Knowledge.” There, he frames his project by noting: “Two attitudes are common in present discussions of ethical knowledge: that expressed by various authoritarian views, and [that] put forth by a variety of opinions often labeled ‘positivistic’ ” (SGEK, 1). While the former “asserts that ethical principles must be taken on authority, or posited by an act of faith, or at least presupposed,” the latter holds that ethical norms “are simply the expressions of settled emotional dispositions, and, being such, they are not the kind of things that can be submitted to rational criticism and reflection” (SGEK, 2, 3). Even though Rawls was concerned with ethical (as opposed to political) theory at the time, he found this choice unacceptable for a strikingly political reason that is worth quoting at length:

Although both of these extremes stand in many ways opposed to each other, they have in common the following: they make reasoning in moral questions to no effect … In either case it is plain that the use of rational principles, as the only reasonable instrument of settling conflicts and guiding decisions, is given up.

Now what is the effect of these two views? Plainly we all have to make decisions, and plainly we need some guide. Therefore the effect of negating rational argument is to usher in other only too well-known techniques of guiding choices and deciding social policy. The result is to encourage in social life just those elements which, in democratic countries, we have tried to get rid of: the authoritarian, the arbitrary and the irrational.

The democratic conception of government looks to the law and not to the state as the primary source of authority; and it views the law as the outcome of public discussions as to what rules can be voluntarily consented to as binding upon the government and the citizens. The law is regarded as those rules which discussion has shown to be right and reasonable so far as the citizens, as a group of intelligent men, have been able to ascertain that fact. Rational discussion is not outlawed or held to be irrelevant. On the contrary, it constitutes an essential precondition of reasonable law. Democratic theory and practice must consider the process of reasoning as one of the very crucial points in its whole program.

(SGEK, 7–8)

Rawls's project was to rescue moral reasoning and the possibility of moral knowledge from the antidemocratic and nihilistic choice between dogmatism and skepticism.

The idea was to ground moral knowledge in “the collective sense of right of free and intelligent men and women.” The key thought was this: “Those principles of law and morals are finally authoritative which reasonable men can willingly adopt as their own, and because they feel them to be right and just after the widest inquiry has been made on the questions involved” (SGEK, 8). Although these judgments are fallible, they can serve as the basis from which to “construct” an ethical theory. Rawls discusses four necessary and sufficient criteria for an individual to be included among this group who provide the justification of principles. He or she must (1) be intellectually and emotionally mature; (2) have a reasonable degree of knowledge and education; (3) “be reasonable,” by which he means something like being “intellectually honest” – willing to learn from evidence, for example; (4) possess “a certain amount of sensitivity for the feelings and interests of other people; and a certain sympathy and understanding for human suffering and human problems” (SGEK, 32–37). Rawls is at pains to stress that this does not involve appeal to an arbitrarily selected group of authorities. In fact, this group includes “any man who possesses to the requisite degree the four properties just considered, irrespective of all other properties such as wealth, class, nationality, race, creed or religion, whatever they may be” (SGEK, 39).

Not all judgments made even by this class of persons provide a suitable basis for constructing an ethical theory, and Rawls places additional requirements on the judgments themselves. For example, he says, we should rely on judgments that are relatively stable:

it is obvious that the study of ethics, so far as it is a rational attempt to explicate common sense morality, could not exist unless there were more or less permanent patterns of judgment and appraisal to serve both as a material for theoretical construction and as a body of data against which the validity of the theory could be tested.

(SGEK, 49–50)

And they should be those about which we are most confident: “They express our deep-seated intuitive convictions which remain on reflection … the feeling of certainty is not temporary, but remains with us after criticism” (SGEK, 57, 58).

The “feeling of certainty” should not be taken to indicate infallibility, however. (Perhaps the term “confidence” would be more appropriate than “certainty.”) Even “the collective sense of right of free and intelligent men and women … is not final, or infallible” (SGEK, 8). Although these judgments serve as data for the construction of an ethical theory, they are not unalterable or foundational.

While an explication could hardly cause us to change all our judgments, it may, after we have reflected upon it, cause us to change some of our opinions. Therefore, not only may an ethical theory provide an answer where there is a genuine conflict, and so where there is no opinion at all; but it may actually change some accepted appraisal which was originally considered a part of the subject matter.

(SGEK, 92)³

Rawls notes that this feature marks an important contrast between ethics and natural science.⁴ So, to summarize: the “data” from which Rawls proposes that we construct an ethical theory consist in the stable judgments that reasonable men and women would make, under favorable circumstances, about which they are most confident after reflection and criticism. However, even these judgments are subject to revision in light of the development of the theory. By constructing principles in this way, Rawls hopes to rescue ethical knowledge from the challenge of dogmatism and skepticism. In addition, the construction of principles in this way can help to provide “criteria to evaluate difficult and doubtful cases wherein there is, as yet, no recognized solution” (SGEK, 298). Together, these elements constitute the core of what Rawls would later identify as “reflective equilibrium,” the approach that he defended and utilized throughout his career.

In order to be able to construct a theory on this basis, the data must show a certain consistency. Thus: “It is hoped, of course, that the class of reasonable men, as defined, will in fact agree in holding certain ethical opinions and agree in making certain ethical judgments. Otherwise ethical theory would have no subject matter from this source” (SGEK, 43). One obvious way in which the judgments of different people might diverge is if they failed to be impartial. Not surprisingly, therefore, in addition to the requirement that the judgments be relatively stable, Rawls also requires that they be impartial. He notes certain obvious factors which could lead to objectionable partiality, and therefore should not be present, such as “Anger, revenge, jealousy, fear and the like” (SGEK, 54). In addition to putting aside these emotions, impartiality also requires that “all of the relevant interests have been conscientiously reviewed in imagination” and the judgment “is not unwittingly or deliberately biased by a likelihood of winning personal gain” (SGEK, 54). If one's own interests are at stake, this is liable to distort the judgment away from impartiality. But Rawls apparently struggles to find a psychological mechanism that would adequately allow us to implement this requirement in real people. He notes

two ways in which the removal of distorting interests can be achieved. First, those interests can be removed from a person's character by repression or reorientation. Second, the person's interests can be satisfied fully, so that temptations are removed … Since a moral judgment involves a fair adjustment of interests, a competent judge must know these interests and their satisfactions directly; and this he cannot do if he has repressed them. Therefore, regarding the legitimate and ordinary interests which come up for adjustment, a competent judge will fully understand them only if he has made them an abiding satisfaction in his life … Yet on the other hand many interests, if satisfied, wreck havoc [sic] on the person himself as well as on society; and some interests of this kind come up for appraisal. Repression is the only course in such cases, and a fair weighing of them will have to depend upon our ability to know them in imagination. It is somewhat futile to urge that a fair moral judge of tyrants must be a person who has been a tyrant himself, since only tyrants know the great good of being a dictator.

(SGEK, 55–56)

Although not wholly without merit, it is obvious that the methods of repression and satisfaction, in addition to introducing a host of additional problems, some of which Rawls notes, cannot guarantee impartiality and agreement among the persons on whose judgments Rawls proposes we rely.

Rawls's first publication, “Outline of a Decision Procedure for Ethics,” was based on his dissertation and came out the following year. In both works, for the most part, Rawls assumes that ethical problems emerge when interests conflict.⁵ While both works are concerned with ethics in general, in “Outline” he seems more willing to identify such conflicts as problems of justice.⁶ A more important development concerns how impartiality is modeled. Rawls drops an explicit requirement that the judgments themselves be impartial and transfers this requirement to the “reasonableness” of the judges. The “sympathetic knowledge” that competent judges possess is of “those human interests which, by conflicting in particular cases, give rise to the need to make a moral decision” (CP, 3). A competent judge has this knowledge either by “experiencing, in his own life, the goods they represent” or “by means of an imaginative experience of it” (CP, 3). Most importantly, however:

A competent judge is required to have the capacity and the desire to lay before himself in imagination all the interests in conflict, together with the relevant facts of the case, and to bestow upon the appraisal of each the same care which he would give to it if that interest were his own. He is required to determine what he would think to be just and unjust if each of the interests were as thoroughly his own as they are in fact those of other persons, and to render his judgment on the case as he feels his sense of justice requires after he has carefully framed in his mind the issues which are to be decided.

(CP, 3)

In other words, Rawls simply stipulates that the judges render decisions through an impartial consideration of the various interests. There is no mention of repression and satisfaction as the psychological instruments by which impartiality is to be achieved. Rawls's model here obviously is the traditional idea of a perfectly impartial spectator, a view that he would later associate with classical utilitarianism (TJ, 23–24, 161–164).

“Justice as Fairness,” published in 1958, included several crucial developments. First, Rawls draws attention to the fact that he is concerned with justice as opposed to other moral virtues. In fact, developing his insights from “Two Concepts of Rules,” published in 1955, he focuses more narrowly still on “justice only as a virtue of social institutions, or what I shall call practices” (CP, 47). Second, Rawls explicitly rejects utilitarianism (at least in its classical form) as “unable to account” for “the fundamental idea in the concept of justice,” which is fairness (CP, 47). Instead, he proposes his “two principles of justice,” which, through various revisions, he would defend for the rest of his career: a principle of equal liberty (later, an equal scheme of basic liberties) and a principle encompassing fair equality of opportunity and the difference principle (CP, 48).

His justification shifts in significant ways, as well. Rawls now imagines the judges as participants in the practice who are rational and “by and large mutually self-interested” (CP, 52). As rational participants with potentially conflicting interests they need to identify principles for assessing and reforming their shared practice. But they understand that any such principles that they propose will be “binding on future occasions.” From this, Rawls claims, it follows that

each will be wary of proposing a principle which would give him a peculiar advantage, in his present circumstances, supposing it to be accepted. Each person knows that he will be bound by it in future circumstances the peculiarities of which cannot be known, and which might well be such that the principle is then to his disadvantage. The idea is that everyone should be required to make in advance a firm commitment, which others also may reasonably be expected to make, and that no one be given the opportunity to tailor the canons of a legitimate complaint to fit his own special condition, and then to discard them when they no longer suit his purpose.

(CP, 53)

The assumption of mutual self-interest is not intended to be “a general theory of human motivation.” Rather, it reflects the fact that issues of justice emerge only when the interests of different individuals conflict (CP, 56). Although this choice is based on self-interested considerations, once the principles or rules of the practice are chosen, compliance is assured by the duty of fair play, which is not itself chosen by the participants. The duty of fair play requires compliance with the chosen principles whether particular applications serve one's self-interest or not. This structure, Rawls claims, reflects our moral nature since we are moved by both self-interest as well as a commitment to the impartial application of the principles. Relying on a self-interested choice of the principles has another advantage. It allows us to “view the principles of justice as the ‘solution’ of this highest order ‘game’ of adopting, subject to the procedure described, principles of argument for all coming particular ‘games’ whose peculiarities one can in no way foresee” (CP, 57). However, he emphasizes, “this highest-order ‘game’ is of a special sort. Its significance is that its various pieces represent aspects of the concept of justice” (CP, 57–58).

Rational and self-interested individuals proposing binding standards for the evaluation of their shared practice would have to consider the possibility that their circumstances might shift in unpredictable ways. It would be irrational to propose principles that favor their current situation if they also impose a high cost under conditions that they might face in the future. This leads Rawls to propose a general criterion for the justice of a practice: “A practice is just if it is in accordance with the principles which all who participate in it might reasonably be expected to propose or to acknowledge before one another when they are similarly circumstanced and required to make a firm commitment in advance without knowledge of what will be their peculiar condition” (CP, 63). In addition, Rawls suggests that this will lead each to accept principles as if “his enemy were to assign him his place” (CP 54). But this does not follow. More precisely, this follows only on the basis of heroic assumptions about social mobility and the prospect of radical change in circumstances. If we consider a rigidly hierarchical social structure which is relatively stable, there may be little or no chance that those at the top will face such an extreme reversal of fortune that they would find themselves at the bottom. In that case, it might be in their rational self-interest to propose principles that would preserve their privilege. In A Theory of Justice, Rawls levels a similar criticism at Edgeworth's assumption that “the policy of maximizing utility on each occasion is most likely to give the greatest utility for any person individually.” Rawls replies that

the necessary assumptions are extremely unrealistic … it must be assumed either that men move from one social position to another in random fashion and live long enough for gains and losses to average out, or else that there is some mechanism which insures that legislation guided by the principle of utility distributes its favors evenly over time.

(TJ, 147–148)⁷

Of course, without these assumptions those at the bottom will find it in their own rational self-interest to propose principles that would radically reform their institutions. In such a case, participants cannot reach consensus concerning principles with which to reform their institution, and Rawls's method fails to deliver principles of justice.⁸

A veil of ignorance, of course, would ensure impartiality and generate consensus far more effectively than either repression and satisfaction or the prospect of social mobility and change. In a pair of papers published in 1963, Rawls first refers to the mutual acceptance of principles “in an original position of equal liberty” (CP, 77). In the first of these, “Constitutional Liberty and the Concept of Justice,” he continues to assume that “each person will be wary of proposing a principle which would give him a peculiar advantage … [because] he will be bound by it in future circumstances, the peculiarities of which cannot be foreseen and which might well be such that the principle is then to his disadvantage” (CP, 78). But in “The Sense of Justice,” the “original position of equal liberty” is explained this way:

In this position it is assumed that there is an absence of information; in particular, it is assumed that the parties do not know their social position, nor do they know their peculiar talents and abilities – that is, their native assets. Briefly, they do not know how they have fared in the natural lottery. Nevertheless, in the original position, knowing the possibility (or allowing for it) of different native endowments, it is rational for them to acknowledge the two principles of justice.

(CP, 113)

It would be irrational for the parties to propose principles strongly tied to their particular circumstances simply because they do not know what those circumstances are. Indeed, it would be impossible for them to do so. This uncertainty, he now claims, underwrites the idea that we may think of the principles “as those which a person would keep in mind if he were designing a social system in which his enemy were to assign him his place” (CP, 98). We no longer need to assume high levels of social mobility or likelihood of radical change in circumstances. However, Rawls offers virtually no discussion or explanation of the features of this original position.

In his initial characterization, the veil of ignorance is fairly thin. The parties still know their full conception of the good on the basis of which they assess the proposed principles. It was only in his 1969 paper “The Justification of Civil Disobedience” that he mentions (once again, without noting or justifying the change from earlier versions) that the parties in the original position “do not know their own particular interests and preferences or the system of ends which they wish to advance: they do not know their conception of the good” (CP, 178). At this point, it is unclear how the parties in the original position could make their choice since the thicker veil of ignorance prevents them from knowing their good. The previous year, Rawls had introduced the notion of primary goods. However, they were initially used not as a tool for the parties in the original position to cope with their lack of knowledge, but rather as a way of identifying the least advantaged social position (representative person) (CP, 158).

The various features of the original position that had accumulated in Rawls's publications over the years without much explanation were unified and given an explicit defense in TJ. Also, for the first time, Rawls presents a systematic argument that the parties would choose his two principles of justice over utilitarianism. The idea of the veil of ignorance is to “nullify the effects of specific contingencies which put men at odds and tempt them to exploit social and natural circumstances to their own advantage” (TJ, 118). We do not want the principles tied to any particular conception of the good any more than we want them tied to any particular social position or set of natural talents and abilities. The point of the virtue of justice, after all, is to identify fair institutions to mediate the conflicts among differently situated individuals with different conceptions of the good. In fact, since we want to force the parties to make their choice “solely on the basis of general considerations” (TJ, 118), we deny them any information that could be used to differentiate one person from another. Further, the parties do not know “the particular circumstances of their own society,” including “its economic or political situation” (TJ, 118). This veil of ignorance is thick (CP, 335–336). On the other hand, the parties do know “the general facts about human society. They understand political affairs and the principles of economic theory; they know the basis of social organization and the laws of human psychology” (TJ, 119). They know that they are choosing principles that will be used in a well-ordered society publicly to evaluate and potentially to reform the basic structure of society. They know that that society will likely contain a wide diversity of conceptions of the good and reasonable comprehensive doctrines. They also know that the members of the well-ordered society for whom they are selecting principles will be moral persons (or citizens): they have the capacity for a sense of justice and the capacity for a conception of the good. These two moral powers are of great importance and will be discussed below.

The parties in the original position are motivated by their conceptions of the good alone, although the veil of ignorance prevents them from knowing their specific contents. They compare and choose among the various principles based on their assessment of which will result in their doing as well as possible in terms of realizing the goals associated with their conception of the good. They “try to advance their conceptions of the good as best they can, and … in attempting to do this they are not bound by prior moral ties to each other” (TJ, 111). They choose “solely on the basis of what seems best calculated to further their interests so far as they can ascertain them” (TJ, 512). Although they are assumed to be “mutually disinterested,” this does not mean that they are “egoists, that is, individuals with only certain kinds of interests, say in wealth, prestige, and domination” (TJ, 12; cf. 111). They cannot assume that their conceptions of the good have these specific contents any more than any other. The reason that we conceive the parties “as not talking an interest in one another's interests” (TJ, 12) is because we aim to represent the circumstances in which there is a conflict among reasonable conceptions of the good that needs to be resolved fairly. As we have seen, those are the circumstances in which the virtue of justice is required. While the parties are not (directly) concerned to promote each others' goods, we do not assume they are opposed either, so Rawls stipulates that the parties are not moved by envy (TJ, 124). In sum, the motivation of the parties is quite different from that of the moral persons or citizens in a well-ordered society. Unlike the individuals for whom they are selecting the principles, the parties are not directly motivated by concerns of fairness or justice. The veil of ignorance is what forces their choice to be impartial and therefore fair. This simplified motivational structure is exactly what gives the original position its power to help us reach reflective equilibrium. If the parties were to consider which principles match their sense of justice, they would simply be reiterating our problem in reaching reflective equilibrium. By asking which principles best serve their interests, “we have asked a much more limited question and have substituted for an ethical judgment a judgment of rational prudence” (TJ, 39; cf. 512). Yet, because of the veil of ignorance, the answer to this question of rational prudence, we hypothesize, will also answer our question concerning the content of the principles of justice that we should endorse in reflective equilibrium.

When Rawls made the veil of ignorance thick enough to prevent the parties from knowing their conceptions of the good it became unclear on the basis of what values they could make a choice at all. In TJ, social primary goods play this role in addition to being used to identify any inequalities among the relevant social positions (TJ, 348–349).⁹

it is rational to want these goods whatever else is wanted, since they are in general necessary for the framing and execution of a rational plan of life. The persons in the original position are assumed to accept this conception of the good, and therefore they take for granted that they desire greater liberty and opportunity, and more extensive means for achieving their ends. With these objectives in mind, as well as that of securing the primary goods of self-respect (§67), they evaluate the conceptions of justice available to them in the original position.

(TJ, 380; cf. 125)

Although Rawls mentions the importance of primary goods for “framing” a rational plan of life, it would be easy to assume that primary goods are identified by their instrumental contribution to achieving one's determinate ends. And this may suggest that they may be tied closely to the content and popularity of various determinate conceptions of the good. In the “Preface to the Revised Edition” of TJ he noted that the original account “left it ambiguous whether something's being a primary good depends solely on the natural facts of human psychology or whether it also depends on a moral conception of the person that embodies a certain ideal” (TJ, xiii). In fact, Rawls says, the primary goods depend on a certain ideal of moral persons as “rational beings with their own ends and capable … of a sense of justice” (TJ, 11; cf. 17, 442, 491). This ideal (or model) is mentioned throughout TJ, but the connection to primary goods may be unclear. Even more importantly, the parties in the original position assume that they have certain “fundamental aims and interests” such as a “religious interest” and an “interest in the integrity of the person” and “a highest-order interest in how all their other interests, including even their fundamental ones, are shaped and regulated by social institutions” (TJ, 131).¹⁰ The fundamental aims and interests that the parties must be concerned with can be drawn from general sociological and psychological knowledge, for example, that many people have a deeply held religious faith. The idea of higher-order interests, however, was somewhat obscure. Apparently, his idea, as he would later put it, was that the parties would want to ensure that “their original allegiance and continued devotion to these ends are formed under conditions that are free” (CP, 228). As developed in his later work, the two moral powers – the capacity for a sense of justice and the capacity for a conception of the good – are presented as necessary for full participation in a scheme of social cooperation (PL, 19, 81, 103–104). The parties have two higher-order interests in protecting these powers so that full social participation is possible, regardless of the particular system of ends that they affirm. So, although the parties do not know the determinate conceptions of the good which they attempt to satisfy, they can rely on primary goods because they serve a wide variety of ends, including fundamental ends that they assume they might have and, most importantly, their higher-order interests associated with full participation in a scheme of cooperation.¹¹

In some ways, it might seem as though the thick veil of ignorance achieves its goal of ensuring unanimity too well. As Rawls notes, “we can view the agreement in the original position from the standpoint of one person selected at random” (TJ, 120). What, one might ask, is left of the idea of a contract? Rawls's version of the social contract differs in a number of respects from ordinary contracts. Most obviously, the veil of ignorance means that there is “no basis for bargaining in the usual sense” (TJ, 120). Furthermore, ordinary contracts are binding, when they are, because they were agreed to through an act of consent. They specify certain obligations that hold only because of the agreement. For Rawls, in contrast, there is a natural duty of justice that binds everyone regardless of any voluntary agreement that they may have made. The original position is designed to help us identify the content of the virtue of social justice, rather than to establish its binding character or importance.¹²

Yet, Rawls insists on retaining the rhetoric of a contract for a number of reasons, even though the agreement in the original position departs from normal contracts in these important ways.¹³ First, the idea of a contract carries with it the implication of publicity. Citizens know the outcome of the agreement and they use the principles in their public deliberations about the basic structure of their society (TJ, 15, 115). Second, Rawls believes that one of utilitarianism's main sources of attractions is that it relies on a maximizing conception of practical reason familiar from the context of individual choice. It does this by “extending to society the principle of choice for one man” (TJ, 24). However, in doing this, it “does not take seriously the distinction between persons” (TJ, 24). Issues of justice only arise in the context of multiple persons with potentially conflicting interests. It is important that this plurality be represented in the original position. As Rawls put it in 1968, “The initial situation must be one of group choice; it must not be that of one person, whether this person is a rational risk-taker or an impartial sympathetic spectator” (CP, 174). Finally, by retaining the rhetoric of a social contract, Rawls is able to emphasize that the parties in the original position must make their agreement in good faith, with the full expectation that they will be able to comply with its provisions once the veil of ignorance is lifted: “when we enter an agreement we must be able to honor it even should the worst possibilities prove to be the case. Otherwise we have not acted in good faith. Thus the parties must weigh with care wither they will be able to stick by their commitment in all circumstances” (TJ, 153). This is a stipulation that we make in setting up the original position, and it is essential in the justification of the two principles.

We are now in a position to examine the main grounds on which the parties in the original position would make their choice. The argument is complex, and here I can only discuss certain strands. Rawls presents his argument that the parties would accept his two principles in two stages: first, by a comparison with average utilitarianism, which focuses on his first principle, and then by a comparison with so-called “mixed conceptions,” which focuses on the second principle. Rawls claims that since the veil of ignorance prevents each of the parties from winning “special advantages for himself … the sensible thing is to acknowledge as the first step a principle of justice requiring an equal distribution” (TJ, 130). But this is not at all obvious. In fact, given their goal of choosing principles that maximize their prospects for achieving their conception of the good, it might seem that a more sensible choice would be average utilitarianism. And, indeed, Rawls acknowledges that “if we waive the problem of interpersonal comparisons of utility, and if the parties are viewed as rational individuals who have no aversion to risk and who follow the principle of insufficient reason in computing likelihoods … then the idea of the initial situation leads naturally to the average principle” (TJ, 143).¹⁴ To argue against this conclusion, Rawls must show why the parties would have a strong aversion to risk when making certain choices or when certain interests are at stake. It is not enough simply to stipulate that they would be risk averse. As he pointed out in a 1974 article, “Reply to Alexander and Musgrave,” such a stipulation “would indeed have been no argument at all” (CP, 247).

The argument just sketched for average utilitarianism assumes that the parties know nothing about the structure or content of their preferences, interests, or ends (cf. Harsanyi 1976). On this view, “The parties are conceived as having no definite highest-order interests or fundamental ends … They are, we might say, bare-persons” (TJ, 152). This model, therefore, involves a choice from an “initial situation” that differs from the original position, where the parties are assumed to “have certain highest-order interests and fundamental ends … which they must try to protect. Since they know that the basic liberties covered by the first principle will secure these interests, they must acknowledge the two principles of justice rather than the principle of utility” (TJ, 152). The parties would be willing to sacrifice potential gains in the satisfaction of other preferences in order to avoid the chance that their fundamental and higher-order interests would not be protected. In fact, an agreement in which they risked having their fundamental interests violated unnecessarily would be one that was not made in good faith. The parties know, for example, that they have a fundamental interest in practicing their religious faith (if they have one) and being able to affirm their moral and philosophical convictions. The parties would want to choose principles that protect liberty of conscience even if that meant giving up possible economic gains that could conceivably be had from restricting the religious liberty of some. The veil of ignorance would prevent them from knowing whether it would be their faith that would be suppressed. Even if they assumed it was unlikely that they would be the ones to suffer, it would not be rational for them to risk it. As Rawls explains in Justice as Fairness, “Were the parties to gamble in that way, they would show that they did not take seriously the religious, philosophical, and moral convictions of the persons they represent. Indeed, they would show that they did not understand the nature of religious belief, or philosophical or moral conviction” (JF, 105).

The “strains of commitment” would be too great if the worst happened and their fundamental interests were not protected. This is the force of saying that certain interests are fundamental. Similarly, the parties would want to ensure that the conditions necessary for the development and maintenance of the two moral powers are satisfied. The parties would not risk undermining the conditions that develop their capacity to participate fully in a scheme of social cooperation. Choosing such principles risks almost completely undermining their prospects for achieving their good, and since that is the goal of the parties, they would avoid such principles at all costs.¹⁵ When their fundamental and higher-order interests are at stake, the parties will be extremely conservative in their choice of principles. This is not due to a peculiar aversion to risk that we assign to the parties. It is because of the generic structure of their interests. If we were to model their preferences formally, we would be forced to say that “the marginal utility of these fundamental interests is infinite” (CP, 228).

It is true that in practice, utilitarianism might also provide protection for the basic liberties. In fact, it seems quite likely to do so if we assume that the utility functions of citizens reflect their fundamental interests and that the basic liberties protect these interests. If this is the case, that is, if utilitarianism can provide support for protection of the basic liberties, “we should be cheered” (JF, 107) since it shows a certain overlap between utilitarianism and justice as fairness. However, this does not show that the parties would choose average utilitarianism instead of the two principles. The principles, remember, are to be used in public deliberation about the institutions of the basic structure. Although we may assume that the protection of fundamental interests would have a very high utility, average utilitarianism would have us weigh this against the preferences that some citizens might have against the protection of the basic liberties (of others). The parties would want to protect the basic liberties directly and explicitly “rather than have them depend upon what may be uncertain and speculative actuarial calculation” (TJ, 139). Furthermore, the two principles contribute to stability by “announcing to one another once and for all that even though theoretical computations of utility always happen to favor the equal liberties (assuming that this is indeed the case here), they do not wish that things had been different” (TJ, 139). The basic liberties directly protect our fundamental interests and provide the conditions for the development and maintenance of our higher-order interests. Because of the structure of their motivation, the parties in the original position would place great weight on protecting the basic liberties. The key contrast between average utilitarianism and the two principles is that the latter, but not the former, provide a direct guarantee that the basic liberties will be protected equally.

Rawls's argument from the original position has become associated with the idea of the “maximin” rule for choice under uncertainty. It is obvious why. Going back to “Justice as Fairness” in 1958, he suggested that we think of the principles as those a person would choose “if he were designing a practice in which his enemy were to assign him a place” (CP, 54), and in TJ he states that “it is useful as a heuristic device to think of the two principles as the maximin solution to the problem of social justice” (TJ, 132). Furthermore, there is a structural similarity between maximin and the difference principle.¹⁶ Yet, for all of this, Rawls's use of maximin reasoning has been widely misunderstood. Contrary to the widespread assumption, Rawls argues that “the maximin rule is not, in general, a suitable guide for choices under uncertainty” (TJ, 133). And as we have seen, we cannot simply posit that the parties would be risk averse; we need an argument as to why, given their situation, choosing in a way that is analogous to maximin would be rational. The strains of commitment argument sketched above depends on a particular structure of fundamental and higher-order interests. For that very reason, however, we have grounds for a conservative choice only when those interests are at stake. When Rawls talks about the parties protecting themselves against the worst possible outcome, it is not the possibility of having a below-average (or even the smallest) share of wealth and income that he is primarily worried about (as long as the share is above a certain minimum). Rather, it is the prospect of not being able to satisfy one's fundamental or higher-order interests. This may be obscured in discussions of maximin that assign numerical values (utility levels) to the various possible outcomes. Once the basic liberties have been secured and everyone is assured a minimal share of resources, the fundamental interests have been satisfied. The question is whether a basic structure satisfying a conception of justice might have “positions we could not accept” (not simply ones that we would disprefer to others). If so, the parties cannot “in good faith” agree to that conception (JF, 103; cf. CP, 250–251). But the strains of commitment argument is much weaker in selecting among conceptions that each secure our fundamental and higher-order interests.¹⁷

A mixed conception is one that accepts Rawls's first principle of justice but substitutes a different principle for his second. Considering the choice between the two principles and such a mixed conception allows us to assess the grounds for supporting the second principle of justice. Assuming that it provides an adequate minimum share of resources, a mixed conception will satisfy our fundamental and higher-order interests and therefore keep the strains of commitment within acceptable limits. This undercuts the grounds that we saw above for using maximin as a heuristic. The implication is that, contrary to widespread belief, “in arguing for the difference principle over other distributive principles … there is no appeal at all to the maximin rule for decision under uncertainty” (JF, 43, n3; cf. 94–95).¹⁸ The reasons the parties would choose the second principle are very different from the grounds for selecting the first, and Rawls presents them much later in TJ (§49), making them easy to overlook. There, the parties are faced with the choice between the two principles and a mixed conception that (like all mixed conceptions) accepts the first principle of justice, and accepts fair equality of opportunity, but substitutes for the difference principle “the principle of average utility constrained by a certain social minimum” (TJ, 278).¹⁹ Since this and many other mixed conceptions protect our fundamental and higher-order interests, it is no surprise that they “are much more difficult to argue against than the principle of utility” (TJ, 278). And, in fact, “some mixed conceptions are certainly adequate enough for many purposes” (TJ, 310). Using the terminology that he introduced only later, it seems clear that Rawls would say that this mixed conception would be a member of the “family of reasonable political conceptions of justice” (PL, 442; cf. xlvi–xlvii, l–li, 6). A society committed to this mixed conception could certainly be legitimate, even if Rawls believes it would not be as fully just as one committed to the two principles. Note, however, that this mixed conception is underspecified. We have not yet determined the level at which the social minimum is set. One way that we might set it is by the difference principle.²⁰ In that case, there is a virtually complete agreement concerning which institutions are just. Indeed, in TJ §43 when Rawls sketches an institutional arrangement that is designed to satisfy the difference principle, it is clear that it would satisfy the mixed conception as well.²¹

Given that the two conceptions are so similar, we must look at secondary considerations to choose between them. Here, I will mention two advantages that justice as fairness has over the mixed conception. First, even though the mixed conception differs sharply from utilitarianism, the idea of utility maximization still plays a subordinate role. This means that it must resolve longstanding issues in the definition of utility and make interpersonal comparisons (TJ, 281–285). This is not merely a theoretical problem, although it is that, too. The principles are to be used in public deliberations concerning the justice of social institutions. Assessments of aggregate utility are much more opaque and therefore controversial than simply identifying the prospective share of primary goods of the least advantaged social position. It also raises the prospect of requiring that we make so-called “shameful revelations” about our preferences (Wolff 1998). A second consideration is that “the difference principle includes an idea of reciprocity [that] distinguishes it from the restricted utility principle” (JF, 122). Starting from a baseline of equality, the difference principle allows inequalities only when everyone gains. Average utilitarianism may allow further inequalities beyond this point so that the more advantaged gain even more while the less advantaged lose (compared to the corresponding positions under the difference principle). At this point, gains are no longer reciprocal. Rawls argues that reciprocity is a deep-seated psychological tendency in human beings, and, other things equal, a conception that is able to tap into it is likely to be more stable than one that does not (TJ, 405–441; JF, 124–126). Focusing on the public application of conceptions and on their stability would not be decisive if fundamental and higher-order interests were at stake. But given that these two conceptions do secure these interests, it seems appropriate that such secondary considerations become relevant, and they both seem to favor the two principles. Thus, while there are very strong grounds for the parties to endorse the first principle (and its priority) that are revealed by a comparison with utilitarianism, there are also grounds, albeit weaker, for endorsing the second principle, revealed by a comparison with mixed conceptions.

To conclude, Rawls argues that the principles chosen from the original position

are objective. They are the principles that we would want everyone (including ourselves) to follow were we to take up together the appropriate general point of view. The original position defines this perspective, and its conditions also embody those of objectivity: its stipulations express the restrictions on arguments that force us to consider the choice of principles unencumbered by the singularities of the circumstances in which we find ourselves. The veil of ignorance prevents us from shaping our moral view to accord with our own particular attachments and interests. We do not look at the social order from our situation but take up a point of view that everyone can adopt on an equal footing. In this sense we look at our society and our place in it objectively.

(TJ, 453)

Although I cannot discuss the point here, when presented as a form of political constructivism in Political Liberalism, Rawls continues to claim that the original position establishes a form of objectivity. (See PL, 102–116.) The choice from the original position answers the challenge that Rawls addressed in his dissertation: how is it possible to have an objective form of moral knowledge? Rawls concluded his dissertation by claiming that he had refuted authoritarianism and positivism “by offering a counterexample; that is, by taking a class of moral judgments and showing that they can be reasoned about and justified” (SGEK, 346). Rawls remained devoted to pushing back the forces of skepticism and dogmatism. The moral knowledge that comes from considering the choice from the original position, which he considered and developed over his career, was his refutation.

1 On reflective equilibrium, see TJ, 18–19, 42–45; PL, 95–97, 381–385; JF, 29–32. On constructivism, see CP, 303–358; PL, 89–129. On the relationship between moral theory and other areas of philosophy, see CP, 286–302.

2 On property-owning democracy, see JF, 135–179. On free speech, see PL, 340–363. On Hiroshima, see CP, 565–572.

3 Cf. SGEK, 293–294: “unless we provided for the possibility that our common sense judgments could be reformed in the light of ethical principles, ethics would be identical with the sociology of morals.”

4 Rawls is aware that some may object that “Scientific theories control their data, and exercise a coercive power over observations” (SGEK, 93). But, he argues, “there is a limit as to how far a theory can discard observations relevant to it. Otherwise, it would not be a theory at all, but an opinion stubbornly maintained in the face of contradicting reports … Thus, while it is true that a physical theory may control its data, the relation is entirely different from that which may exist between an ethical theory and its data” (SGEK, 94–95).

5 See, for example: “If one considers that ethics is concerned with a fair adjustment of a variety of conflicting interests, then an impartial judgment will have, as one of its characteristics, an awareness of all of the relevant interests involved” (SGEK, 53).

6 See, for example, CP, 13. In the dissertation the connection to justice, although present, was not as pronounced. See, for example: “Caring for a person's interests may result in caring for his interests irrespective of those of other people, so that acting from love may very easily lead to injustice, and what to all appearances is arbitrary favoritism. Therefore, so far as the disposition is a direct inclination to help another's interests, we cannot assign it a definite value until we know to what extent it is likely to work for injustice” (SGEK, 161).

7 See also Rawls's discussion of this issue in “Constitutional Liberty and the Concept of Justice” (1963) in CP, esp. 80–85.

8 David Gauthier faces a similar problem in chapter 7 of Morals by Agreement. There, he argues for a proviso that is not “the object of an agreement among rational individuals, but [is] a precondition to such an agreement” (1986, 192). This proviso “prohibits bettering one's situation through interaction that worsens the situation of another” (1986, 205). This determination requires a comparison between an individual's prospects when participating in an institution with his or her prospects without participating. Rawls rejects the idea of determining what one's prospects would be outside of society altogether, for example, in “Distributive Justice” (1967), CP, 135; cf. PL, 278.

10 Rawls wavers between referring to these as “higher-order” and “highest-order” interests. See “Social Unity and Primary Goods,” CP, 365. In Political Liberalism, he consistently uses “higher-order” and I follow that usage.

11 Rawls argues that the basic liberties, in particular, are necessary for the protection of these higher-order interests (PL, 299–324). See also the discussion of the moral powers in Mandle 2000, 55–70.

12 It is true that in TJ, Rawls considers the justification of the natural duties from the original position (TJ, 293–301). However, this occurs after the choice of principles has already been made. Furthermore, reframed as a political conception of justice, justice as fairness would allow other justifications of this natural duty.

13 See CP, 249, for a defense of the idea of a contract or agreement rather than mere choice: “In general, the class of things that can be agreed to is included within, and is smaller than, the class of things that can be rationally chosen” (cf. JF, 102).

14 Rawls argues against the use of the principle of insufficient reason. The parties cannot rationally assign an equal probability to each possible outcome. I will simply note that the parties do not even know what the various social positions will be, let alone the likelihood of their occupying each of them (TJ, 134; CP, 248).

15 This is similar to the reason that Rawls holds that the social basis of self-respect is “perhaps the most important primary good” (TJ, 386). Self-respect is “the sense that one's plan is worth carrying out” (TJ, 155). If the parties were to choose principles that did not provide the social basis of self-respect, their prospect of successfully carrying out their plan and achieving their good, no matter what it was, would be fatally compromised. Although I cannot explore this here, Rawls argues that his two principles provide stronger support for self-respect than rival conceptions of justice (TJ, 477–480).

16 Rawls notes the importance of distinguishing these two “very different things” in “Some Reasons for the Maximin Criterion” (1974), CP, 225. See also the passage he added to the revised edition of TJ at 72: “It is undesirable to use the same name for two things that are so distinct.” On the other hand, Rawls himself apparently conflates the two in the “Reply to Alexander and Musgrave,” CP, 241.

17 This is not to say it is nonexistent. In Justice as Fairness, Rawls connects the ideas of reciprocity and the strains of commitment. He argues: “in asking the less advantaged to accept over the whole of their life fewer economic and social advantages (measured in terms of utility) for the sake of greater advantages (similarly measured) for the more advantaged, the principle of utility asks more of the less advantaged than the difference principle asks of the more advantaged. Indeed, asking that of the less advantaged would seem to be an extreme demand. The psychological strains that may lead to instability are bound to be greater” (JF 127). He then notes that the strains may result either in people who “see themselves as oppressed” and “are ready as the occasion arises to take violent action in protest” or, more mildly, result in people who “grow distant from political society” and become “withdrawn and cynical” (JF, 128). Presumably, the most serious violations of the basic liberties generate the first type, while more mild violations of the principle of reciprocity generate the latter. See the discussion of reciprocity below.

18 In “Some Reasons for the Maximin Criterion,” Rawls seems to make a “strains of commitment argument” for the use of maximin reasoning when choosing between the two principles and a mixed conception (CP, 229–230). However, note that there the mixed conception he is discussing does not have a guaranteed social minimum. Therefore, fundamental interests may well be at risk.

19 Note that considering this choice will not give us a justification of fair equality of opportunity since both principles accept it. In the case of the difference principle it is important that it be paired with fair equality of opportunity because the difference principle is a structural principle. It does not apply to individuals but to social positions defined by the institutions of the basic structure. But the parties in the original position are concerned with their individual prospects. While the difference principle identifies permissible structural inequalities, fair equality of opportunity requires that each individual have a fair prospect of occupying the various social positions.

20 Rawls speculates that there might be an implicit reliance on the difference principle to specify this level (TJ, 278–279). However, see Waldron 1986.

21 On this account, in addition to securing the basic liberties and ensuring fair equality of opportunity, the government has the responsibility to keep markets functioning efficiently without the formation of monopoly powers, to bring about full employment, to provide public goods, to limit inheritance and bequests, and to ensure an appropriate “social minimum” (TJ, §43). “But once a suitable minimum is provided by transfers, it may be perfectly fair that the rest of total income be settled by the price system, assuming that it is moderately efficient and free from monopolistic restrictions, and unreasonable externalities have been eliminated” (TJ, 245).