Julius Sensat / Philosophy 941: Rawls / Notes / Installment 5: Reasoning in the original position (2)

First fundamental comparison (conclusion)

Last week we discussed Rawls's argumentation concerning the "first fundamental comparison" in the OP: that between Rawls's principles of justice and the principle of average utilitarianism. Rawls argues that it is appropriate in this comparison for the parties in the OP to use the maximin rule for decision under uncertainty. Use of this rule would lead to a choice of Rawls's principles because they represent the maximin option--the option whose worst possible outcome is best. Recall that Rawls claims that three conditions obtain that make the maximin rule appropriate (p. 98):
  1. The decision makers have no reliable basis for estimating probabilities of possible outcomes of their decision.
  2. The guaranteeable level (the worst outcome of the maximin option) is satisfactory, so that the parties care relatively little about possible gains above that level.
  3. The non-maximin option has a possible outcome that would be disastrous, that is, significantly worse than the guaranteeable level.
As we saw last week, Rawls argues that the second and third conditions are strongly satisfied for this comparison. What is the role of the first condition in his argument? Rawls tries to avoid relying on the first condition as much as possible. This may be because to support it he has to appeal to claims about probability that are somewhat controversial. None the less, he claims that it is sufficiently satisfied that the strong satisfaction of the second and third conditions make it still rational to rely on the maximin rule (p.102). I will try to give you a sense of why he thinks he can make this claim.

To see how having a basis for assigning probabilities could threaten the argument, it is best to go back to the thought experiment of last week. Recall that in this thought experiment, you are offered a choice between the following gambles on the outcome of a drawing of a ball from an opaque urn:

You are told only the balls in the urn are either red or black, but you are not told the numbers (they could be all red or all black or anything in between). Here G2 is the maximin option, and the guaranteeable level is $1000. To insure that the third condition obtains, suppose you owe the mob $1000, they are collecting tomorrow, and you have no other way of obtaining the money. To insure that the second condition obtains (that the guaranteeable level is highly satisfactory), assume that you are not in dire straights except for your debt to the mob and that there is nothing of very significant value to you, compared to repaying the mob, that you could secure with winnings above $1000.

We assumed that you had no basis for assigning probabilities. But suppose you did. Suppose for example that you were informed that there are 1 billion balls in the urn, and all but one of them are red. Intuitively, this would allow you to assign a very high probability to winning $10,000 in G1 and would seem to indicate that G1 is the rational choice.

More generally, many decision theorists say that when you can assign probabilities, you should follow the decision rule of maximizing expected utility rather than the maximin rule. According to this way of thinking, there are two possible states of the world that have a bearing on the outcome of your decision: either the world is such that a red ball will be drawn, or the world is such that a black ball will be drawn. Together each option open to you (G1,G2) and each state determine an outcome of decision, as shown in the following table:

  Red Black
G1 $10,000 $0
G2 $2000 $1000

To maximize expected utility, you need to be able to assign values or "utilities" to these outcomes and probabilities to the states. The utilities represent the relative importance to you of the various outcomes. Suppose you assign them as follows (for further discussion on arriving at these assignments, see the supplemental notes on von Neumann-Morgenstern utility theory):
 

  Red Black
G1 10 0
G2 8 7

Now suppose that you are told that there are 100 balls in the urn and all but one are red. Then it seems reasonable to assign probabilities as follows:

  Red Black
G1 .99 .01
G2 .99 .01

The expected utility of an option is a weighted average of the utilities of the possible outcomes of the option, where the weights are provided by the corresponding probabilities. So to calculate the expected utility of an option, you multiply the utility of each of its possible outcomes by the probability of the state that would lead to that outcome, and you add up the results.

Expected utility of G1 = probability of red × utility of $10,000 + probability of black × utility of $0 = .99(10) + .01(0) = 9.9 + 0 = 9.9

Expected utility of G2 = probability of red × utility of $2,000 + probability of black × utility of $1000 = .99(8) + .01(7) = 7.92 + .07 = 7.99.

According to the rule of maximizing expected utility, you should choose G1. And note that since the utility assignments reflect the assumptions that the second and third conditions of for use of the maximin rule obtain to a high degree, the proponent of expected utility maximization can in this case argue for a different choice, on the basis of the very high probability that a red ball will be drawn.

How does all this apply to Rawls's argument? If a party in the original position had a basis for assigning a very low probability that choice of utilitarianism would put his client citizen in one of the envisaged bad situations, along with a high probability that utilitarianism would put him in a desirable position, then perhaps an analogous argument to the one just explained could be given for choosing utilitarianism on grounds of expected utility maximization. However, the veil of ignorance would seem to rule out the possibility of having information that would justify such probability assignments.

Some decision theorists that use a Bayesian conception of probability argue that it is always rational to maximize expected utility. They say that you should assign probabilities based on the confidence you have that one state will obtain rather than another. They also say that in the absence of any objective information one is justified in assigning equal probabilities to the various outcomes. Some go so far as to say that in the original position you should assign equal probability of your client citizen's ending up in any position from the least to the most advantaged. To see what effect this would have, go back to our thought experiment with the urn and construct the following table assigning equal probabilities.
 

  Red Black
G1 .5 .5
G2 .5 .5

In this case, we get the following expected utilities:

Expected utility of G1 = .5(10) + .5(0) = 5
Expected utility of G2 = .5(8) + .5(7) = 7.5
So in this case we get a result that agrees with the maximin rule. So if the veil of ignorance is reasonable, then even if Bayesian theorists are right that probability assignments can make sense and be justified in the absence of definite information, thereby precluding the satisfaction of the first condition, this may not affect the decision if the second and third conditions obtain to a high degree.

Second fundamental comparison

The "second fundamental comparison" in the original position is that between Rawls's two principles and what he calls the "principle of restricted utility," which is constructed from Rawls's principles by removing from them the difference principle and putting in its place the principle of average utility constrained by a social minimum. Thus, the principle of restricted utility says that justice requires (1) the basic equal civil and political liberties guaranteed by Rawls's first principle, with (2) fair value for the political liberties, (3) fair equality of opportunity, (4) a guaranteed minimum allotment of social and economic benefits, and subject to all the foregoing, (5) arranging prospects so as to maximize average prospective welfare.

What is the rationale for making this comparison? Rawls says that it is a fundamental comparison because it compares justice as fairness with its strongest rival from the set of conceptions in which the priniciple of utility has a prominent role. If justice as fairness wins out in this comparison, we could conclude that the principle of utility should not play a fundamental role in a conception of justice. There is another reason for the comparison, I think. It allows us to focus specifically on the difference principle to determine its reasonableness. In the first comparison, justice as fairness wins out over utilitarianism in large part because of its explicit endorsement of the basic liberties, fair value of political liberties and fair equality of opportunity. To some extent, the difference principle went along for the ride. Here it must pull its own weight.

Of course, it is pulling its own weight in the context of agreement on the remaining components of justice as fairness: the first principle and the principle of fair equality of opportunity. In the second comparison there is agreement on these. This common content plays an important role in the argument. In a sense, the support for the difference principle comes in part from its claimed superior potential to cohere with those principles in a workable conception of justice.

Rawls groups the grounds for his conception into three categories: those based on publicity, reciprocity and stability, respectively.

Publicity

Recall that one of the constraints on the original position is the requirement on the parties to choose a conception of justice to be the public basis of social unity. One of the roles of such a conception is to provide the basis of persons' conception of themselves as citizens. In virtue of the common content (the overlap) in the competing conceptions of the second comparison, the parties know that in the well-ordered society for which they are fashioning a conception of justice, citizens are to regard themselves and recognize each other as free and equal. More particularly, because of the principles already agreed to, citizens are to regard themselves and recognize each other as equal partners in fair social cooperation. Rawls thinks that the difference principle coheres better with this conception.

Reciprocity

One of the reasons for this is that the public conception as determined by the common content provides a foothold for considerations of reciprocity. Among equal partners in in a cooperative venture, reciprocity seems appropriate as a basis for distribution. The difference principle expresses or incorporates an idea of reciprocity, whereas the restricted utility principle is "aggregative" and pays no attention to equality or reciprocity. The first comparison already exploited this fact with respect to equality. The current comparison brings reciprocity to the forefront.

Rawls's thought is that the equal partnership secured by the common content calls for a presumption in favor of equal distribution, unless inequalities work to everyone's advantage. This is an idea of reciprocity, and the difference principle implements one possible version of this idea. Rawls attempts to describe one way in which the difference principle does this in §§36.2-36.3, with reference to Figure 1 on p. 62. The idea seems to be that point D, selected by the difference principle, is the only efficient point at which those who are better off are not better off to the detriment of those who are worse off.

At any rate, Rawls goes on to maintain that the difference principle contains or expresses a second, deeper idea of reciprocity as well: the idea that social institutions are not to take advantages of contingencies of natural endowment, initial social position or good or bad luck over the course of life except in ways that benefit everyone. This is the sense in which the difference principle treats the distribution of native endowments as a common asset.

Notice that the principle of restricted utility itself could be interpreted as viewing the distribution of native endowments as a common asset.  However, it arguably does not deal with that distribution on the basis of some sort of reciprocity, but only through the aggregative requirement of maximizing average welfare.

Stability

Rawls thinks that in the second comparison, justice as fairness will be judged as better able to generate its own support than the principle of restricted utility, and thus will be preferable to the parties on stability grounds. While stability is more fully treated in part IV, he raises some relevant considerations here, in §§37-38. He explores the reasons derived from the conception of justice that citizens would have for complying with its requirements and for not seeking to alter the basic structure in a direction more to their advantage.

With respect to justice as fairness, it seems that the more advantaged are more likely to be discontent than the least advantaged. What reasons will be available to them in a well-order society for not seeking modifications in the basic structure that work to their further advantage? Rawls lists the following:

  1. Commitment to the status of citizens as free and equal gives all citizens a reason to embrace the reciprocity of the difference principle.
  2. The more advantaged can see themselves as having been benefited, first by their good fortune in the natural lottery, and second, by the basic structure, which invites them to benefit still further by training their talents and putting them to work for everyone's benefit. In §38, he compares the utility principle in this regard, saying that it asks more of the less advantaged (namely, to accept over the whole of their lives fewer social and economic advantages for the sake of greater advantages for the more advantaged), than the difference principle asks of the more advantaged.
  3. The difference principle contributes to a public culture that encourages mutual trust and the cooperative virtues, especially compared to the principle of utility, which is hard to apply and leads to suspicion and distrust. Further, by affirming the difference principle, the more advantaged convey to the less advantaged in the clearest possible terms their acceptance of reciprocity; they thereby foster trust and cooperation on the part of the less advantaged.
A point that Rawls makes in §38 seems relevant here in connection with stability and the strains of commitment. The absence of an idea of reciprocity in the utility principle puts more severe psychological strain on compliance. Rather than embracing the conception because of the reciprocity it expresses, citizens have to tap their capacity for sympathy or identification with the concerns and interests of others.

The social minimum

In the second comparison, both alternatives will include a social minimum. How are they to be determined? In justice as fairness, its level will be the amount necessary to maximize the prospects of the least advantaged. What about the minimum in the restricted utility principle? Rawls considers what some have suggested: the minimum necessary to guarantee a decent human life. Rawls compares the two with reference to the issue of stability. The idea is that a minimum is not sufficient if it imposes excessive strains of commitment. When are strains excessive? In §38, Rawls considers two ways in which strains might be excessive:
  1. A stronger form of excessive strain applies when citizens become sullen and resentful, and they are willing to take violent action in protest of their condition. The least advantaged are bitter; they reject society's conception of justice and see themselves as oppressed.
  2. A weaker form does not necessarily lead to violence. Rather, citizens are apathetic and cynical about political life. They withdraw into other activities. They are not hostile or rebellious, but they do not affirm society's conception of justice--it fails to engage their moral sensibility.
Rawls's claim is that the social minimum of the principle of restricted utility may be sufficient to avoid excessive strains of the first sort, but not of the second. As evidence, he would point to the apathy and cynicism that is so widespread in contemporary capitalist welfare states. Moreover, to elaborate on a point from §39, some forms of inequality generate problematic attitudes not only on the part of the deprived. The more advantaged can themselves become cynical and arrogant. The difference principle, on the other hand, has a better chance of avoiding both forms of excessive strain. This is due in part to the respect that the difference principle accords every citizen; even the least advantaged should be able to embrace a conception of themselves as full members of the public political world, in virtue of society's ideals and principles and the ways in which the more advantaged work to their good.

SENSAT HOME      |     PHILOSOPHY DEPT      |     UWM          Updated February 21, 2004