Whereas consequentialism (e.g. utilitarianism) takes the good to be prior to the right -- defining right action as the promotion of valuable consequences -- deontological ethics takes the right to be prior to the good -- claiming that it can be wrong to perform optimal acts.  Deontologists appeal to three main anti-consequentialist ideas.

• categorical rules (prohibiting or enjoining conduct)
• individual rights (interpreted as side constraints)
• principles of distributive justice

KANT
Immanuel Kant (1724-1804) was a systematic philosopher who produced complicated accounts of the possibility and nature of all knowledge, both theoretical and practical.  His account of morality, developed in his Groundwork for the Metaphysics of Morals (1785),  is deeply rationalistic, appealing to our rational natures and subordinating our empirical natures, especially our emotions and inclinations.  He thinks that moral requirements are demands of reason that can and should constrain our empirical selves but that nonetheless spring from our own rational wills.

Kant's anti-utilitarian sympathies are clear.  He rejects all other moral theories that base an agent's duty on any kind of appeal to happiness (405, 432-33, 443-44).  Also, in his doctrine of the good will, he professes indifference to the actual consequences of an agent's actions (394, 399), provided only that his intentions are correct.

THE GOOD WILL
Kant begins the first section of the Groundwork by focusing on the value of a good will.  A good will is the only thing good without qualification (393).  Other goods can be misused and produce bad outcomes; the good will is good in itself (393-94).  This means that the value of the good will does not depend upon its consequences (394).

If happiness were our ultimate end, then we would have been better-off acting on instincts that reliably secure happiness than by letting reason guide our conduct (395).  But since we have the capacity for practical reason, its function must be not merely to secure happiness, but to produce a will that is good in itself (396).

But what is a good will?  It must be a will guided by practical reason, for no ulterior purpose.  The good will is a will that conforms to duty (the demands of practical reason) for the sake of duty (397).  To illustrate the good will, Kant considers various people whose actions conform to duty.

1. The prudent shopkeeper for whom honest dealing is the most prudent policy (397).
2. The honest shopkeeper for whom honest dealing is what morality requires (397, by implication).
3. The sorrowful agent who resists suicide, though he has no desire to live (398).
4. The person who is beneficent from sympathy (398).
5. The person who is beneficent, because it is his duty, despite the fact that personal sorrows have deadened his sense of sympathy (398).
6. The person who is beneficent, because it is his duty, despite congenital indifference to the plight of others (398).

Must the performance of duty be painful to display a good will?  Must pro-moral sentiments taint a good will, and does a good will require indifference or contra-moral sentiments?

The good will does not require indifference or contra-moral sentiments; it's just that its operation is especially clear in such cases (397).  The problem with the Sympathetic Benefactor is not the presence of sympathy per se, but rather the fact that sympathy is what determines the agent's will.  This is a problem, Kant thinks, because the coincidence between sympathy and duty is accidental and unreliable (398).  In other cases, sympathy may prevent one from doing one's duty (e.g. punishing a wrongdoer or requiring a child to undergo a painful but necessary medical treatment).  If so, there's nothing wrong with performing beneficent actions that express one's sympathy, provided one's sympathetic sentiments are suitably regulated by correct moral beliefs.

But even if pro-moral sentiments need not taint a good will and a good will does not require indifference or contra-moral sentiments, it is nonetheless true that Kant thinks that the Grudging Moralists can and do display a good will.  Is this virtue or continence? Does genuine virtue require that the agent's emotions and inclinations conform to her sense of duty?

THE A PRIORI GROUND OF DUTY
If a good will conforms to duty for the sake of duty, we are still left wondering what duty consists in.  But Kant thinks that his discussion of the good will tells us something about this.  If the motive of duty cannot be empirical, then duty itself must be based on reason and on nothing empirical.  This is what he means when he says that the ground of duty must be knowable a priori (388-89, 410).  This may seem strange.  Surely, my duties (e.g. my contractual duty to sell you my widgets) depend upon empirical facts about what I have done (e.g. that I signed a contract to sell you my widgets).  Kant does not deny this.  But he thinks that my duties apply independently of my willingness to comply.

Moral requirements, Kant thinks, must depend upon general, rather than contingent and variable, features of agents.  They must depend upon features of agents as such.  But what is it to be a moral agent?  A morally responsible agent cannot be determined to act by her strongest desires or emotions; she must be able to step back from her desires, distinguish between the intensity and authority of her desires, deliberate about the appropriate objects of desire, and regulate her actions in accord with her deliberations.  This leads Kant to conclude that moral requirements depend upon features of rational agents as such.

Kant captures this idea in terms of his distinction between hypothetical and categorical imperatives.

Now all imperatives command either hypothetically or categorically.  The former represent the practical necessity of a possible action as a means of attaining something else that one wants (or may possibly want).  The categorical imperative would be one which represented an action as objectively necessary in itself, without reference to another end [414]

THE CATEGORICAL IMPERATIVE
If moral requirements are not to be based on variable empirical conditions, then they must be universal.  To achieve this sort of universality, Kant requires an agent be able to will her maxims (= subjective principles of action) to be universal.  His reasoning seems to be something like this.

1. Because moral requirements are inescapable, they express categorical imperatives.
2. Hence, moral requirement must apply to rational agents as such, independently of their empirical motivations.
3. Hence, moral requirements must be universal.
4. Hence, an agent's maxims must be capable of serving as universal laws.

F1: Act only maxims that you can will to be a universal law of nature (421).

F2: Treat humanity, whether in your own person or that of any rational agent, always as an end in itself and never merely as a means (429).

F3: Every rational being should be regarded as an autonomous legislator in a kingdom of ends (431-33, 438).

Some actions are so constituted that their maxims cannot without contradiction even be thought as a universal law of nature, much less be willed as what should become one.  In the case of others this internal impossibility is indeed not found, but there is still no possibility of willing that their maxims should be raised to the universality of a law of nature, because such a will would contradict itself [424].

Kant thinks that the case of false promises involves maxims whose universalization is inconceivable (403, 422).  I cannot will that my maxim of keeping promises only when it suits my interests be universal, because if everyone acted on this maxim, promises would often not be kept and the general level of trust necessary to sustain the practice of promising would not obtain.  Thus, a general practice of false promising would prove self-defeating.

Here, as in other contexts, Kant is especially concerned about those who would make an exception of themselves to otherwise legitimate rules (424).  In rejecting special pleading and free-riding, Kant may seem to appeal to the question "What if everyone did that?".

However, there appear to be many perfectly innocent activities that are not universalizable in this sense.  No one could will to perform any activity that is part of some larger division of labor (e.g. practicing philosophy or selling but not producing widgets), because, if everyone performed that one activity, no one would perform the other activities in the division of labor necessary to produce the products that sustain the division of labor.

Fortunately, the conceivability interpretation of F1 appears not to be basic.   Some maxims whose universalization is conceivable cannot be willed to be a universal law.   A common interpretation of F1 and consistency in the one's will results if we ask if we can consistently accept the consequences of everyone acting with our motives (cf. 403, 423). There are stronger and weaker interpretations of F1 depending on the range of consequences one must consider in universalizing.

• Empirical universalization asks whether I can accept what would be the actual or probable consequences of everyone's acting on my maxim.
• Counterfactual universalization asks whether I can accept the consequences of everyone's acting on my maxim in all possible circumstances or worlds.

But we have reason to wonder whether even counterfactual universalization provides a good interpretation of F1.  Like the Golden Rule, counterfactual universalization makes one's moral duties hostage to one's antecedent desires in a way Kant clearly wants to avoid.

How then should we interpret F1?  Kant thinks that our duties must be determined by features of us as moral and, hence, rational agents (408, 412, 425-27, 432, 442).  So we should interpret F1 as asking what rational beings can consistently will.  But this claim is ambiguous.  F1 might be interpreted as asking what rational beings can consistently will, that is, what someone who is rational can consistently will.  This test can depend on the contingent interests and desires possessed by rational beings, and so counterfactual universalization is one way of articulating it.  Alternatively, F1 might be interpreted as asking the different question about what rational beings as such can will, that is, what someone can will insofar as she is rational.  On this interpretation, F1 asks what we can will, not insofar as we have particular, contingent wants and interests, but what we can will insofar as we are rational beings.

This second interpretation has some interesting implications.  On this interpretation, I ask whether -- insofar as I am a rational being -- I can consistently will the universalization of my maxim.  You and I are presumably no different from each other solely insofar as we are rational beings.  Rational beings are different from one another in countless ways, but not insofar as they are rational.  Different maxims will survive counterfactual universalization depending on the contingent interests and desires of the rational agent who tries to universalize.  Not so under this interpretation.  Because all agents are alike insofar as they are rational, the results of this sort of this test do not depend on who performs it (427).  Nor is it clear that universalization is essential to F1.  We need only ask what a rational being would will, qua rational being; it shouldn't matter whether you or I ask the question, and we shouldn't need to ask what if everyone did that.

But we may wonder whether there is anything that a purely rational being would will.  I can understand what it is for a rational being to will or choose various actions and outcomes on the basis of interests and preferences.  But what would an agent stripped of all such interests and preferences will or choose?  It may seem that there is no basis left on which to will or choose.

FROM UNIVERSALITY TO HUMANITY
But Kant thinks that insofar as one is rational one will make rational nature one's end (428).  Happiness, Kant claims, can have only conditioned or instrumental value; rational nature alone has intrinsic value that does not depend on contingent and variable circumstance.  So only rational nature could be valued regardless of one's contingent circumstances and sentiments.

The Humanity formula seems to have more content than the Universality formula, and many have taken the Humanity formula to express Kant's credentials as a deontologist.  Is the Humanity formula incompatible with consequentialism?  What is it to treat someone or something as a means?  As a mere means?  If I torture an innocent child because this is the only means of getting a terrorist (her mother) to reveal the location of a terrorist bomb, do I use the child as a mere means?

RAWLS
In A Theory of Justice John Rawls develops an egalitarian conception of social justice which he defends by appeal to a hypothetical social contract.  Traditional social contract theory answers the question

Under what conditions is state authority justified?

Under what conditions is the basic structure of society just?

THE TWO PRINCIPLES OF JUSTICE
Let's begin with Rawls's principles.  There are two.

1. Each person is to have an equal right to the most extensive total system of equal basic liberties compatible with a similar system of liberty for all [Equal Basic Liberties].
2. Social and economic inequalities are to be arranged so that they are both (a) to the greatest benefit of the least advantaged ... [the Difference Principle] and (b) attached to offices and positions open to all under conditions of fair equality of opportunity [the Principle of Fair Equality of Opportunity].

Suppose D1 is the maximum equal distribution.  The Difference Principle would condone (indeed, require) D2.  Utilitarianism would require D3, which both Equality and the Difference Principle would condemn.

Rawls takes utilitarianism and, to a lesser extent, libertarianism as the chief rivals to his more egalitarian conception of justice.  Some of Rawls's arguments are arguments of overall comparative plausibility, independent of the contract argument.  The libertarian conception gives moral significance, as justice as fairness does not, to the morally arbitrary effects of the social and natural lotteries (TJ 65-83).  Laissez-faire must give way to a scheme which recognizes the moral arbitrariness of the social lottery and, ultimately, to one which recognizes the moral arbitrariness of the natural lottery as well. In contrast with utilitarianism, justice as fairness places a lower limit on the sacrifices that some might have to make in order to improve the position of others (28, 175-83, 500).

RAWLS'S SOCIAL CONTRACT ARGUMENT
But Rawls's most distinctive argument for his conception of social justice is his social contract argument.  Rawls's social contract is hypothetical (12, 121), appealing to an original position that excludes factors arbitrary from the moral point of view (120).  Parties in the original position (OPs) are placed behind a veil of ignorance, which deprives them of information about their identities and attributes (137), including

identity
gender
race
class position
natural assets
aversion to risk
conception of the good
circumstances of society
generation

are mutually disinterested, but concerned to advance their own prospects (143)
use primary goods (maximally flexible social assets) to measure their prospects (62, 92)
evaluate representative social positions (64)
display instrumental rationality (143)
have general social knowledge (137)

Within the contractual argument, Rawls takes utilitarianism to be the main rival to justice as fairness.  We must distinguish between classical utilitarianism, which tells agents to maximize total happiness, and average utilitarianism, which tells agents to maximize average or per capita happiness.
 

	A	B	C	D
D1	5	5	5
D2	4	4	4	4

Whereas classical utilitarianism endorses D2, average utilitarianism endorses D1.  OPs would prefer average to classical utilitarianism.

Rawls claims that OPs would prefer his two principles to (average) utilitarianism.  Rawls contrasts the maximin decision rule with the rule of maximizing expected (average) utility.  Rawls says there are three conditions that favor maximin (154-55).

1. Ignorance of the probabilities of the various outcomes.
2. A conception of the good such that one cares little, if anything, for what one might gain above a certain minimum.
3. Alternative decision rules have possible outcomes that would be unacceptable.

Consideration (3) seems to dominate.  Utilitarianism would allow some to suffer so that others might benefit in ways that one could not tolerate if one was the worst-off, and this also threatens the social stability of utilitarianism.

THE RAWLSIAN INTERPRETATION OF KANT AND THE KANTIAN INTERPRETATION OF RAWLS
There are at least two possible connections between Kant and Rawls.

(1) We might compare the content of justice as fairness with the content of the Categorical Imperative by comparing Rawls's two principles with the Humanity formula (F2).  Here, Rawls's thought seems to be that his two principles provide a compelling interpretation of F2's implications for the justice of the basic structure of society.  Because utilitarianism can in some circumstances require that some individuals make huge sacrifices so that others may gain, Rawls thinks that it fails to ensure that each is treated as an end and none as a means.  By contrast, because his principles, especially the difference principle, ensure that the only deviations from equality actually improve the prospects of the worst-off, Rawls thinks that justice as fairness does treat each as an end and none as a means.

(2) We might also compare the ways that Rawls and Kant justify their principles.  In particular, we might compare Rawls's contract argument with our preferred interpretation of the Universality formula (F1).  We might see the choice of OPs as a way of modeling the will of someone insofar as she is rational.  Kant thinks that we must ask what a rational being as such, independently of her contingent interests and inclinations, would will.  We might model this as the problem of what terms of conduct one would choose -- insofar as one was rational and valued rational agency -- to govern a world of rational beings who have different, sometimes conflicting, contingent interests and desires in which resources are scarce.  Call these conditions the circumstances of humanity.  To make sure the agent's choice does not reflect her contingent interests and aims, we need to abstract from her contingent and variable aims and interests.  One way to do this is to insist that choice is as one that must be made by someone subject to important motivational and informational constraints.  We might appeal to a hypothetical social contract employing a veil of ignorance, such as Rawls does.

Does appeal to the circumstances of humanity reintroduce empirical considerations of need that we spent so much time weeding out of F1?  Not if we distinguish the conditions under which a choice is made and the conditions that a choice is for.  Empirical considerations can enter into the second issue, provided they are kept out of the first.