On the Rationality of Sacrifice1Jean-Pierre Dupuy

Contagion: Journal of Violence, Mimesis, and Culture, Volume10, Spring 2003, pp. 23-39 (Article)

Published by Michigan State University Press

For additional information about this article

Access Provided by Universitaet Wien/Bibliotheks-Und Archivwesen at 11/22/11 11:35AM GMT

http://muse.jhu.edu/journals/ctn/summary/v010/10.dupuy.html

ON THE RATIONALITY OF SACRIFICE1

Jean-Pierre DupuyEcole polytechnique, Paris, and Stanford University

i;"came to be interested in John Rawls'sy4 Theory ofJusticean active.interest which led me to become the publisher of the French versionofthat bookin part for the following, apparently anecdotal reason:

1)On the one hand, as early as the first lines of his book, Rawls makesit clear that his major target is the critique of utilitarianism. Utilitarianismis the defendant, charged with vindicating sacrifice. As everyone knows,"justice does not allow that the sacrifices imposed on a few are outweighedby the larger sum ofadvantages enjoyed by many (...) it denies that the lossof freedom for some is made right by a greater good shared by others(Theory 3-4; my emphasis).

2)On the other hand, although the term "sacrifice" pops up again andagain throughout the book, you may look up the index: it doesn't figurethere. "Sacrifice" is not granted the dignity of a concept.

I found that shocking and my curiosity was aroused. My own researchis driven by a few anthropological convictions (see Dupuy, Le Sacrifice etl'envie). I am going to list them here, for the sake of clarity and honesty,without trying to vindicate them. The arguments I am going to present donot depend in the least on whether these convictions are valid or not; theyshould be assessed on their own merits.

a) AU non-modern social and cultural institutions are rooted in thesacred.

1 I should like to thank John Rawls and the late Robert Nozick for their remarks on earlierversions of this paper.

24Jean-Pierre Dupuy

b)Of the three dimensions of the sacred: myths, rituals, andprohibitions, the most fundamental is the ritual.

c)The most primitive and fundamental ritual form is sacrifice.d)Sacrifice is the reenactment by the social group ofa primordial event

which took place spontaneously: a process of collective victimage whichresulted in the murder of a member of the community. This elimination ofa victim reestablished peace and order. There lies the origin of the sacred.The victim is taken to be the cause or the active principle both of theviolent crisis and its violent resolution. It unites within itself oppositepredicates: it is at the same time infinitely good and infinitely evil. It canonly be of a divine nature.

e)Christ's death on the cross is just one more occurrence of theprimordial event. As far as facts are concerned, there is no differencebetween primitive religions and Christianity. The difference lies in inter-pretation. For the first time in the history of humankind, the story is toldfrom the victim's viewpoint, not the persecutors.' The story (the Gospel)takes side with the victim and proclaims its innocence. When, in ourmodern languages, we say that the victim was scapegoated, we just say asmuch.

f)Modern institutions embody a tension between two contradictorydrives. On the one hand, the drive to resort to more of the same: the driveto scapegoating. On the other, the anti-sacrificial drive set in motion by theChristian Revelation.

In totality or only partially, these points have been made by severalanthropological traditions, in particular the French sociological school,with the works of Fustel de Coulanges, Durkheim and Mauss; and theBritish anthropological school, with Frazer and Robertson-Smith. Freudand the Belgo-British anthropologist Hocart gave them a new momentumand, more recently, Ren Girard has produced an impressive synthesis.2 Asis well known, Nietzsche's philosophy relies on point e), for which theauthor of The Gay Science ("there are no facts, only interpretations") feltjustified in bringing a radical indictment against Christianity, taken to bethe morality of the slaves.

In this light, Kantianism appears to be a secularized form of Chris-tianity. The question I want to raise is: to what extent is this true ofRawlsianism?

2 See, in particular, his Violence and the Sacred, and also The Scapegoat.

On the Rationality ofSacrifice25

It might be retorted: these references to anthropology are irrelevant tomodern political philosophy, and, at any rate, the word "sacrifice" has nowacquired ameaning altogether disconnected from its alleged religious roots.I do not think the burden of the proof is on my side, all the more so since,as we are going to observe, most authors who participate in the debateabout the rationality of sacrifice choose their examples within a verylimited range, whose importance for anthropology is crucial: thescapegoating mechanism.

1. A few sacrificial case studies.Consider the fictional (and fictitious) situation imagined by William

Styron in his novel Sophie's Choice. It has the same structure as the storymade up by Bernard Williams in his indictment of utilitarianism ("Jim'schoice" in "A Critique"). The Nazi officer orders Sophie to choose whichof her two children will be sent to the gas chamber, the other one thenbeing saved. Should she refuse to choose, both of them would die. What isit rational and/or moral for Sophie to do, if that terminology can be used atall in such a nightmarish context? It seems that utilitarianism (and, moregenerally, consequentialism) would have it that Sophie abide by the Nazi'scommand and choose to sacrifice one ofher children: at least the other onewill live. In this light, utilitarian ethics would appear to be guilty ofjustifying a sacrificial choice which most of us find appalling.

Take now the fairly different structure imagined by Robert Nozick."Utilitarianism doesn't, it is said, properly take rights and their non-violation into account; it instead leaves them a derivative status. Many ofthe counterexample cases to utilitarianism fit under this objection; forexample, punishing an innocent man to save a neighborhood from avengeful rampage (...) A mob rampaging through a part oftown killing andburning will violate the rights of those living there. Therefore, someonemight try to justify his punishing another he knows to be innocent of acrime that enraged a mob, on the grounds that punishing this innocentperson would help to avoid even greater violations of rights by others, andso would lead to a minimum weighted score for rights violations in thesociety."3 It will be noted, in passing, that Nozick avoids carefully giving

3 R Nozick, Anarchy, State, and Utopia 28-29. Reminder- Rawl's two principles ofjusticeread as follows: "First Principle- "Each person is to have an equal right to the most extensivetotal system of equal basic liberties compatible with a similar system of liberty for all."Second Principle: "Social and economic inequalities are to be arranged so that they are both

26Jean-Pierre Dupuy

this structure its true, original name: Caiaphas's choiceprobably theparadigmatic case, ifone is to accept the anthropological tenets stated at thebeginning. Caiaphas, addressing the Chief Priests and the Pharisees, saysin effect: "You know nothing whatever; you do not use your judgment; itis more to your interest that one man should die for the people, than that thewhole nation should be destroyed" (John 11. 49-50).

It is my contention that it would be extremely unfair to chargeutilitarianism alone with that crimeif, indeed, vindicating sacrifice is acrime. Our whole conception of rationality is a party to it. Most analyticalmoral phi losophers, I guess, along with normative economists, consider thePareto principle, known also as the principle of efficiency, to be a self-evident axiom which any consistent moral doctrine should adopt. If atransformation, virtual or actual, makes some people better off withoutmaking the others worse off, how could anyone complain about it? In myfollowing remarks, I shall call the Pareto principle the principle ofunanimity. Returning to Sophie's and Caiaphas's choices, it is easy tounderstand that it is not only utilitarianism that concludes in favor of therationality of sacrifice in those cases, but more fundamentally the principleofunanimity and, as a consequence, all criteria compatible with it. And thisis so because ofa feature that these sacrificial situations share. Whether thevictim is sacrificed or not, its physical well-being remains the same: it iskilled or eliminated all the same. As far asfacts are concerned, the victim'sfate does not depend on whether it is sacrificed or not (the child Sophiechooses to be sent to the gas chamber would have died all the same had she

(a) to the greatest benefit of the least advantaged [that is to so-called difference principle],and (b) attached to offices and positions open to all under all conditions of fair equality ofopportunity" (Theory 302) What may be most original about these principles are the rulesspecifying their priority: the first takes precedence over the second, and within the second,the clause concerning the fairness of equality of opportunity has priority over the differenceprinciple. Moreover, these priorities take the form that logicians call the lexicographical orlexical orderthat is the ordering ofwords in a dictionary (also called leximin), maybe theanti-sacrificial too par excellence The words having the different first letters are listed in theorder of the first letters, no matter what the other letters in the words may be. The first letterof a word, then, is lexically first in relation to the others, in the sense that if we replace thisletter by another one that comes later in the alphabet, a choice ofother letters could possiblycompensate for the change. There is no possible substitution, and it is as if the order of thefirst letter had an infinite weight. Yet at the same time, the order of the other letters has apositive weight, for it two words begin with the same letter, then the following letters decidethe word's place in the dictionary. A lexical order manages to give all of the elements of atotality a positive role, without making them all substitutable for each other. Now, the verybasis of the sacrificial principles is the substitutabihty of the elements within the totality

On the Rationality ofSacrifice27

refused to choose; Jesus is part of the Jewish nation: were he not to besacrificed he would perish all the same). What makes the victim's fate asacrifice, then, is not its death: it is the meaning thereof. The sacrificialvictim dies in order for the others to live on.

To put it in more philosophical terms: the concept of sacrifice, here, isnot defined counterfactually . If the victim were not sacrificed, her level ofwell-being would remain the same. The sacrifice has to do with the actualcausal connection between her level of well-being (she is put to death, orexpelled, or her rights are denied) and the others'. When she is sacrificed,the others are saved, or safe. And someone, possibly the victim herself,intentionally activated that causal connection in order to achieve a higherend, such as the maximization of the welfare of the community (or thesatisfaction of some divinity's needs).

Let me precisely call a unanimity-sacrificial situation any socialcontext such as Sophie's or Caiaphas's choices in which the principle ofunanimity suffices to conclude in favor of the rationality of sacrifice.Inasmuch as Rawls's principles ofjustice are by their very form compatiblewith the principle of unanimitywhich Rawls holds to be the caseitseems appropriate to assert the following: If they could be abstracted orextruded from the theory of justice in which they are embedded, andapplied to a unanimity-sacrificial situation, Rawls's principles ofjustice, bytheir very form, would favor the sacrificial choice.

It is important to make two technical remarks at this stage:First, when I contend that Rawls's principles ofjustice are compatible

with the principle of unanimity (or efficiency), I am not referring only tothe difference principle. This should be made perfectly clear, since it isonly in reference to the latter that Rawls asserts explicitly this compatibility(Theory 79). IfI were actually referring to the difference principle in orderto support my contention that the principles ofjustice favor the sacrificialchoice in a unanimity-sacrificial situation such as Sophie's or Caiaphas'schoices, one might immediately object that I am missing a fundamentaltenet of the theory of justice, namely, that the difference principle ishierarchically and "lexicographically" ranked lower than the principles onequal liberties and equal opportunities; in other terms, one might object thatwelfare, even the welfare of the worst-off, cannot be paid for in terms ofunequal liberties. However, please note that in the sacrificial situations Ihave been considering, there is no such trade-off between basic libertiesand economic and social gains. All the values at stake in the choice

28Jean-Pierre Dupuy

situation belong to the same category: fundamental liberties, lives, etc.Those are the values governed by the first principle ofjustice.

Now, in its broader version, the first principle takes on the same formas the difference principle: that of a lexical ordering between the in-dividuals ranked according to the way they fare on the values in question.For instance, Theory states: "All social valuesliberty and opportunity,income and wealth, and the bases of self-respectare to be distributedequally unless an unequal distribution of any, or all, of these values is toeveryone's advantage. Injustice, then, is simply inequalities that are not tothe benefit of all" (62); and, later we read, "...liberty can be restricted onlyfor the sake of liberty.. .a less than equal liberty must be acceptable to thosecitizens with the lesser liberty" (250).

From these statements, one is justified in inferring that the firstprinciple has the form of a leximin ordering, like the difference principle,and therefore is compatible with the principle of unanimity.

Howeverand this is my second technical remarkit could be arguedthat it is still an open question whether or not Rawls's principles areactually compatible with the Pareto condition. It should be noted first thatfor my contention to be truenamely, that these principles favor thesacrificial choice in a unanimity-sacrificial situationthey have to becompatible with the Pareto condition in its stronger version. It is sufficientthat some people are made better off by a given transformation, the others'lots remaining the same, for this stronger principle to approve of thattransformation. Now a large number of authors have interpreted thedifference principle in a way that renders it incompatible with the strongPareto condition. Thus, Nozick: "With regard to envy, the differenceprinciple, applied to the choice between either A having 10 and B having5, or A having 8 and B having 5, would favor the latter. Thus, despiteRawls's view, the difference principle is inefficient in that it sometimes willfavor a status quo against a Pareto-better but more unequal distribution(Anarchy 229).

It is because these commentators interpret the difference principle inthis way that they feel justified in accusing it ofbeing a mere expression ofenvy. It is my contention that this interpretation is unwarranted and shouldbe dismissed. It contradicts the maxim Rawls hammers home all the time:injustice is inequalities that are not to everyone's advantageeveryone, thatis the better-off as well as the others. The difference principle, then, favorsa transformation that betters the condition of the better-off without

On the Rationality ofSacrifice29

betteringnor damagingthe condition ofthe worse-off. The same appliesto the first principle in its broader version.

Nozick might reply: "Rawls's maxim that injustice is inequalities thatare not to everyone's advantage, leaves open the possibility that "everyone"means everyone, and "advantage" does not mean merely "non-dis-advantage" but in fact means advantage, that is improvement. In this casethe difference principle would favor (8, 5) over (10, 5).4 However, thisresults from a confusion between a constraint on inequalities and aconstraint on a social transformation. IfRawls's maxim did apply to a socialtransformation, it would rule out any transformation that did not meet theweak Pareto criterion. It would then favor (10, 5) over (10, 6), which isclearly absurd. Nozick may well want to restrict his interpretation ofRawls's maxim to those transformations that increase the degree ofinequalities. But then, (8, 8, 5) would be preferred to, say, (16, 8, 6), whichis no less absurd.

If we now take "inequalities that are not to everyone's advantage" tomean inequalities that are not to everyone's advantage, it becomes clear thatLeximin is the form of Rawls's principles of justice. Suppose 5 is themaximum the worse-off can get, and the better-off can get as much as 10.It is appropriate to say, then, that the inequality corresponding to (8, 5) isless "to everyone's advantage" than the inequality corresponding to (10,5)since the worse-offare, in either state, treated more favorably than theywould in any other state, and the better-off are treated in the former lessfavorably than they could, given this constraint.

I deemed it necessary to make all of these points in order to uphold mycontention that theform of Rawls's principles ofjustice is favorable to thesacrificial choice in sacrificial situations, which may have sound somewhatprovocative, since the Theory ofJustice itself may be read as a powerfulanti-sacrificial scheme.

2. Rawls and utilitarianism on sacrifice.I want now to generalize the previous point and defend the following

thesis. The most important cases of social transformations which it islegitimate to dub "sacrificial" are such that both the utilitarian principle andRawls's principles vindicate them. This is so because the weakest normativeprinciple thatjustifies them is weaker than both the utilitarian principle andRawls's principles.

4 Robert Nozick, personal communication.

30Jean-Pierre Dupuy

If this is correct, and if we are opposed to sacrifice, it follows that byand large, utilitarianism doesn't fare worse than Rawls's principles ofjustice as to the vindication of sacrifice.

This is shocking. Isn't one of the major accusations leveled atutilitarianism the fact that it may allow for serious infractions of liberty forthe sake of greater benefits for others? When, on the first page ofhis book,Rawls contends that "each person possesses an inviolability founded onjustice that even the welfare of society as a whole cannot override," withwhat is he contrasting his conception of justice, if not with utilitarianethics? What ifwe showed that whenever utilitarianism justifies sacrifice,Rawls's principles are likely to do the same?

A remark, first, which cannot serve as a demonstration, to be sure, butwhich is puzzling nevertheless. Whenever an author belonging to theanalytical tradition has to exemplify the assertion that utilitarianism favorssacrifice, he or she seems compelled to resort to a single class of instances,a class that happens to be a sub-class of the class of unanimity-sacrificialsituations: the scapegoating mechanism. We saw an example of that inNozick and it would be tedious to quote from other philosophers, such asSandel, Larmore, Williams, etc. Normative economists are no exception tothe rule. Thus, E. J. Mishan: "There is much that might increase totalutility, or that might realize Pareto improvements, that is nonetheless quiteunacceptable to civilized societies and can, therefore, become no part oftheir agenda. However much the aggregate utility enjoyed by a hystericalmob in kicking a man into insensibility exceeds the disutility of the victim,society would feel justified in intervening" (971).

Is it inevitable for the critics ofutilitarianism to appeal almost uniquelyto the scapegoating mechanism? Logical analysis in itself does not imposesuch limitation. The fact that utilitarianism is likely to favor the sacrificeof some for a greater good shared by others results, we are told, from twomajor factors. First, it is a ideological doctrine that gives priority to theadvancement of some independently defined overall social good over theright. Secondly, since it defines that overall good as an aggregate in whichthe various individual conceptions of the good are melted, as it were, itdoes not take seriously the "plurality and distinctiveness of individuals"(Rawls); it does not sufficiently respect the fact that each one "is a separateperson, that his is the only life he has" (Nozick). From these twocharacteristics, it seems to follow that the ills caused to some may be maderight by a greater good enjoyed by others. However, the task of the criticsof utilitarianism is not yet complete. They still have to exhibit contexts in

On the Rationality ofSacrifice3 1

which someone's being seriously wronged brings about a greater good forothers. And this is where the weight ofour cultural heritage, I submit, over-rides the constraints of sheer logical analysis.

Of course you can appeal as much as you like to fantastic, exotic, oddor queer psychologies, for example that ofa sadist fiend. You can postulate,as the fancy takes you, that my sleeping on my stomach, or my eating frogsand snails, causes such intolerable pain to a great number ofpeople aroundthe world as to justify, from a utilitarian stance, the violation of my in-alienable rights to sleep and eat as I like. I must sadly confide that it is thefrequent recourse to examples of this kind that is too often responsible forthe fact that what we call "Anglo-Saxon" moral and political philosophy isnot taken as seriously as it should by "continental" philosophers. On theother hand, the latter may have too tragic and heavy a sense ofhistory, andlack a certain sense of humor.

Let me put it this way: we are more readily satisfied with sociological,historical or anthropological accounts than with psychological ones.

Let me then restate my question as follows. Can the critics ofutilitarianism come up with a plausible and non-trivial social context inwhich, from a utilitarian stance, a serious loss for some would be maderight by a greater good for othersapart from a unanimity-sacrificialsituation?

One might argue as follows. Let us consider a situation in which theinterests of different individuals are not in agreement and any choice ofaction will benefit some at the expense ofothers. Suppose that in the nameof a certain conception of the social good an action is taken that is held tofurther the general welfare. One might be tempted to say that the decisionto carry it out entails the sacrifice of those who are opposed to it. Accord-ing to this interpretation, the mere fact of positing an overall social goodwould amount to sacrificing victims on the altar of the general will.

However, this line of reasoning is hardly acceptable. At that rate, any-one could complain that she is sacrificed in any social state that maximizesa given social welfare function, be it utilitarian or not. She would just haveto point to the state that maximizes her own utility function or her owninterest, taking it as a benchmark. For instance, Rawls's difference principlecould be said to sacrifice the interests of all but the worst-off (a critiquethat has actually been leveled at Rawls by his right-wing or libertarianopponents). Any definition of sacrifice in counterfactual terms ("I, a richperson, would be better off if the income tax system were proportionalrather than progressive; therefore I am entitled to declare myself a sacrifi-

32Jean-Pierre Dupuy

cial victim") is bound to wash its meaning out. That is why it is soimportant to emphasize that no such counterfactual characterization ofsacrifice applies to the religious or anthropological roots of the notion.

However, Rawls himself provides an answer to our question. I amreferring to the case of religious (and, derivatively, philosophical, moral,political or racial) prejudice and persecution which proves to be of para-mount importance for a correct understanding ofA Theory ofJustice. Sincethe publication of his book, Rawls has made it very clear that a majorpurpose of his theory is "to spell out the implications of the principle of[religious] toleration" such as arose historically after the Reformation andthe subsequent wars of religion. From then on, citizens of democraticnations have been aware of the impossibility of organizing social co-operation as before, through a public agreement on a single and commondefinition of the good. The political problem has become that of achievingsocial unity "in a society marked by deep divisions between opposing andincommensurable conceptions of the good" (Rawls, "Justice as Fairness"22). It is out of the question for its solution to depend on a general moral,philosophical, let alone religious conception, for the latter would be but oneof the many coexisting in society.5

"Justice as Fairness," and the absolute priority it gives to the principleof equal liberty of conscience, is the solution, according to Rawls. Now,even if he does not state this explicitly, it does not seem untrue to histhought to say that the principle of utility is to a traditional, religious, in-tolerant society as justice as fairness is to the "public culture of a con-stitutional democracy." The former may vindicate intolerance whereas thelatter embodies the spirit oftoleration. For in a society regulated by a singleconception of the good, those who do not adhere to it may have their libertyrepressed for the majority's sake.

On what ground is utilitarianism supposed to make this right?Apparently we are not dealing with a unanimity-sacrificial situation as Idefined it. For the latter requires that the well-being of those who aresacrificed would have been the same had they not been sacrificed (anecessary condition for the strong Pareto condition to apply). Thiscondition obviously is not met in the present context, for in one case theminority is the victim of intolerance, and in the other it benefits from theliberty of conscience.

3 In his Political Liberalism, Rawls fleshes out the broad implications of pluralism andrespect for others' conception of the good for his theory of justice.

On the Rationality ofSacrifice33

Nothing can be added to this evidence as long as one is content with apsychological account. Thus, one could suppose that the majority'spsychology is such that they feel uncomfortable or upset with the others'not sharing their beliefs. However, as I said, we must go beyond this stage.

Anthropology teaches us that there is a strong connection betweenunanimity and the religious mind. You can think ofDurkheim's contentionthat the ideas of divinity, society and totality are one and the same: thesacred corresponds to the reification ofsociety in its entirety. You can thinkof the thesis advocated by the tradition of thinkers I mentioned at thebeginning, namely that sacrificial ritual is the original keystone of religioussociety: one of the main features of sacrifice as a ritual is that it gathers thewhole community around a center, the sacrificial altar. You can think of thesecularized versions of this: Rousseau's conception of democracy as re-quiring the direct participation of all, without any exception whatever; orthe Moscow and Prague trials whose craving for unanimity was so strongas to require the defendants' self-criticism.

In all of those social contexts, social order depends critically on theattainment of unanimity. Short of this, social chaos breaks out. Let us takethe standpoint of a utilitarian judge: in the sacrificial case, the rights of thesacrificial victims are violated, but order is maintained; in the non-sacrificial case, the unanimity condition is not met, chaos takes over, andthe rights of all become a dead letter. The condition for a unanimity-sacrificial situation to obtain is satisfied. By itself, the principle ofunanimity concludes in favor of the rationality of sacrifice.

It can be objected that this reasoning is contingent upon a false belief:namely that a breach ofunanimity causes the disruption ofthe social order.(In his discussion of the limitations imposed on liberty ofconscience by thecommon interest in public order and security, Rawls insists strongly on thenecessity of assessing correctly the likelihood of damage to public order:see Theory 213-16). But in a religious or quasi-religious setting, "false"beliefs may actually turn out to be true by the simple fact that when peopleact on them, they become true. The sacred is the realm of self-fulfillingprophecies. If men believe that the social order will collapse if they ceaseto feed their gods with victims, that will certainly be the case.

If the foregoing is correct, it seems that the major social contexts inwhich the principle of utility favors the sacrifice of the fundamental rightsof some for the sake of society as a whole are such that the principle ofunanimity alone permits to reach the same conclusion. Insofar as they arecompatible with the latter, Rawls's principles ofjustice do just the same.

34Jean-Pierre Dupuy

Before broaching the next and last step of my argument, I want to dojustice to an obvious objection. What if the sacrificial victim does notbelong to the social group we are considering? Then, if the potential victimwere not sacrificed, it would not be the case that her lot would remain thesame. From an anthropological standpoint, the possibility of this con-figuration should be held with some suspicion. It is precisely one of thedeluding effects of the scapegoating mechanism to make the victim appearexternal or alien to the group (after all, her sacralization through sacrificeaccomplishes just that). The persecutors' interpretation is that she waseliminated because she was different, whereas the truth is that herdifference stems from the fact that she was singled out for elimination.From a logical point ofview, however, that possibility cannot be ruled out.

Consider the following fancy story. N people are dying because one oftheir vital organs has turned dysfunctional. Each of them might be saved ifonly they could benefit from a transplant: a heart for one, a lung foranother, a liver for the last. The question is: should we put Alter to death,Alter being a young and healthy fellow, and give his heart, lungs, liver, etc.to the dying ones?6

Obviously, the sacrificial and non-sacrificial cases are no longerPareto-comparable. However, a concept introduced by Serge-ChristopheKoIm can be brought to bear on this situation: fundamental dominance(105). A state "fundamentally dominates" another state if and only if thereare permutations of the payoffs distributions that result in the first statePareto-dominating the second. One verifies immediately, in the situation athand, that although the sacrificial case does not Pareto-dominate the non-sacrificial case, the former fundamentally-dominates the latter.

Fundamental dominance does not have the same self-evidence as theprinciple ofunanimity since it is stronger than the latter. However, it is verymuch in keeping with Rawls's principles ofjustice, inasmuch as their formis that of Leximin. Indeed, fundamental dominance as a partial ordering iscompatible with Leximin as a total ordering: whenever the former sayssomething, the latter concurs. It is likewise compatible with the principleof utility.

Leximin and fundamental dominance are identity-neutral (as is theprinciple ofutility): they are indifferent to the identity ofpersons. The only

6 This example has appeared many times in the ethics literature, in writings of JudithThomson, Francis Kamm, John Ferejohn, etc. I should like to thank Robert Nozick forpointing that out to me.

On the Rationality ofSacrifice35

thing both principles care about is how payoffs are distributed acrosspopulations. It is true that Leximin sides with the potential victim.However, who is the victim? It is not a person with a name, as in: "You arePeter." It is an anonymous position in a structure. When it takes the formofLeximin, the secularization of the Christian drive to side with the victimis inevitably corrupted. This is ultimately, I submit, the reason why Rawls'sprinciples of justice end up by their very form justifying sacrifice in thewhole category of sacrificial situations I have examined.

The fact that most of the important cases of sacrificial choices areamenable to strong-Pareto improvements does not entail that every strong-Pareto improvement is necessarily sacrificial, and thereby runs counter toour deep-seated anti-sacrificial bias. A fire is destroying an apartmentbuilding. I, a fireman, can, given the limitation ofmy resources, rescue tenpeople out of the fifteen whose lives are threatened. Should I refrain to doso on the ground that the other five would then be "sacrificed" to the othersor to the whole community? Here the strong Pareto optimal choices areobviously the rational, efficient and just ones, and no "sacrificial" elementseems to be involved

However, a strong-Pareto improvement can easily become the locus ofa sacrificial choice. Recall that what makes a victim a sacrificial one is notits physical state, it is the meaning ascribed to this state. Is the victim to diein order for the others to live on? Suppose I am about to rescue this womanwhen she says to me: "Go and save my son, his life is more dear to me thanmy own." The meaning of her death will have dramatically changed. Ifsacrifice, rather than self-sacrifice, is inimical to us, we may be tempted totry and eliminate as far as possible all meaning from the decision-makingprocess. The "modern," logical solution, it seems, would be here therecourse to chanceuntil we realize that most primitive sacrificial rituals,as well as Christ's passion, had, as one of their key elements, the drawingof lots. The difference, though, is that in a religious setting, there is no"chance event" (etymologically speaking, the cast of a dice) withoutmeaning.

What I have been trying to do, so far, is to drive a wedge between thespirit of the theory of justice, which is anti-sacrificial, and the letter, orrather the form of its principles, which is such as to justify sacrifice in awhole class of sacrificial situations. However, this inner contradiction, asI see it, should not be blamed on Rawls's incoherence. The sacrificialelement in the principles of justice results from their meeting a basicconstraint of rationality. Insofar as the theory purports to be, not only

36Jean-Pierre Dupuyreasonable, but also rational, its compatibility with such a minimumprinciple of rationality as the principle of unanimity, or even fundamentaldominance, appears to be an indispensable requirement.

3. There is no reflective equilibrium about the rationality of sacrifice.On the one hand, justice as fairness purports to be a political

conception of justice suitable for a public culture shaped by the principleof toleration, which is the anti-sacrificial principle par excellence; on theother, the principles to which it leads favor, by their very form, sacrificeafter the fashion of utilitarianism in most of the important cases ofsacrificial situations. The apparent contradiction dissolves when onerealizes that Rawls's principles ofjustice do not apply, and are not meantto apply, to sacrificial situations.

Some may think it would have saved me much toil if I had started withthis proposition. But things are more intricate than they seem primafacie.

It might be said that Rawls's principles of justice are not meant to beprinciples which we appeal to in determining whether an action is just ornot, since they are meant only to apply to the basic social and politicalinstitutions of a society. Now sacrificial choices concern actions, not thedesign of institutional arrangements. However, this distinction betweenaction and institution becomes invalid if we take the anthropologicalstandpoint that has been ours throughout this essay, and take account of thewell-documented fact that at the origin ofmost, ifnot all social institutions,we find a sacrificial choice made in a sacrificial situation, such asCaiaphas's choice.

The true reason why Rawls's principles of justice do not apply tosacrificial situations is that they belong to the ideal conception ofjustice(or "ideal theory"). They are meant to regulate a well-ordered and evenperfectly just society "under favorable circumstances" (Theory 351).Apparently a sacrificial situation is altogether alien to this description.

The method of comparison between the principle ofutility and Rawls'sprinciples of justice may then seem extremely unfair. Utilitarianism isaccused of favoring sacrifice in contexts that are excluded from the scopeofJustice as Fairnessand in which, were this exclusion to be revoked, thelatter wouldn't fare differently from the former. All this on the ground that,contrary to Justice as Fairness, utilitarianism purports to be of universalapplication.

It is too easy for utilitarians to counter-attack by using the same kindof strategy as Justice as Fairness. A possibility is Harsanyi's decision to

On the Rationality ofSacrifice31

exclude all "antisocial preferences" from the utilitarian calculus (56).Another possibility is to restrict the scope ofutilitarianism to that permittedand defined by the original position and the features of a well-orderedsociety and/or the public culture of a democratic society. Anything re-sembling a sacrificial situation will be automatically ruled out.

My aim is not to criticize the idea of Justice as Fairness, but to under-stand and circumscribe its meaning. I submit the following.

A caricatured presentation of the foregoing would be to say: Rawls'sprinciples ofjustice are meant to apply to a society already governed by thevery same principles ofjustice. However, as Paul Ricoeur has shown ("Lecercle de la dmonstration), this circularity is not vicious: it is bothinevitable and productive. The aim is not one of foundation but ofdisclosure. The philosopher's task is to organize the basic ideas andprinciples already implicitly existing in our considered judgments andconvictions about justice and injustice into a coherent conception. Hencethe concept of "reflective equilibrium."

Seen in this light, the great merit of Rawls's Theory of Justice is toreveal that the ethos of "democratic societies" rests on an exclusion: theexclusion of those sacrificial situations which the Theory preciselyexcludes from its field of application. What the Theory excludes from itsfield of application is in fact constitutive of the Theory. The latter tells usat least as much by what it rejects as by what it affirms.

To put this another way: a fundamental result ofA Theory ofJustice isnot that Rawls's principles should be preferred to the principle of utility.The main effect brought about by the Theory, its staging of the originalposition, the veil of ignorance, etc., is of that it leads the parties to rank theprinciple oftoleration above the principle ofsacrifice. It is that it makes thelatter inconceivable, impossible. It excludes it altogether.

Let us take the example of the scapegoating mechanism, the para-digmatic case of all sacrificial situations. For it to function and beproductive, the following conditions must be met, among others:

The individuals must fall prey to violent and contagiouspassionsenvy, jealousy, hatred, spite, etc.so that they attribute all theevils of the community to one individual whose elimination restores peaceand order.

The individuals must be unaware ofwhat is actually going on, theymust believe in the guilt of the victim. If someone is to manipulate themechanism, he must see that which remains concealed from the others: theinnocence and arbitrariness of the victim.

38Jean-Pierre Dupuy

If the story of what happened is to be told, it must be from thevantage point of the persecutors, not from the victim's.

On each one of these three points, the setting of the original positionmakes it impossible for these conditions to be met:

The people know that they live in a well-ordered and just society,that they have a sense of justice, all of which results in the confining ofdisruptive passions to a harmless level.

The publicity condition deprives the scapegoating mechanism ofanyefficiency whatsoever.

The original position is entirely devised to give the worst-offi. e.,the potential victimsa privileged position.

When Rawls resorts to the concept of reflective equilibrium, there aretwo possibilities he does not envisage: multiplicity and non-existence. Asfar as sacrifice is concerned (the major stake in the Rawls versusutilitarianism debate), I submit that there is no fixed point. What we haveinstead, in our deep-seated convictions as well as in the Theory, is anunresolved tension between two opposite drives. One is sacrificial: it isreflected in the Theory by the form taken by the principles. The other isanti-sacrificial: it is made manifest in the spirit and the goals of the Theory.

Quite unwittingly, the Theory reflects our moral predicament, asexpressed by Thomas Nagel: "the world can present us with situations inwhich there is no honorable or moral course for a human being to take, nocourse free of guilt and responsibility for evil" ("War and Massacre). Thisis, I guess, what the Christians mean when they invoke the existence of anoriginal sin.

WORKS CITED

Dupuy, Jean-Pierre. 1992. Le Sacrifice et l'envie. Paris: Calmann-Lvy.Girard, Ren. 1979. Violence and the Sacred. Baltimore: The Johns Hopkins UP.------ . The Scapegoat. 1986. Baltimore: The Johns Hopkins UP.Harsanyi, John C. 1982. "Morality and the Theory of Rational Behavior. In

Utilitarianism and Beyond. Eds. A. Sen and B. Williams. Cambridge:Cambridge UP.

KoIm, S. -C. 1972. Justice et quit. Paris: CNRS.Mishan, E.J. 1972. "The Futility ofPareto-Efficient Distributions." TheAmerican

Economic Review 62:971-76.Nagel, Thomas. 1979. "War and Massacre." In Mortal Questions. Cambridge:

Cambridge UP.

On the Rationality ofSacrifice39

Nozick, Robert. 1974. Anarchy, State, and Utopia. New York: Basic Books.Rawls, John. 1971. A Theory ofJustice. Oxford: Oxford UP.------ . 1987. Thorie de la Justice. Paris: Seuil.------ . 1985. "Justice as Fairness: Political not Metaphysical." Philosophy and

Public Affairs 14.------ . 1993. Political Liberalism. New York: Columbia UP.Ricoeur, Paul. 1988. "Lecerclede la dmonstration." In Individu et Justice sociale.

Autour de John Rawls. Eds. C. Audard, J.-P. Dupuy, and R. Sve. Paris: Seuil.Styron. William. 1979. Sophie's Choice. New York: Random House.Williams, Bernard. 1973. "A Critique ofUtilitarianism." In Utilitarianism For and

Against. Eds. J. J. C. Smart and B. Williams. Cambridge: Cambridge UP.