+ All Categories
Home > Documents > Retrospective and Prospective Utilitarianism

Retrospective and Prospective Utilitarianism

Date post: 23-Jan-2017
Category:
Upload: brian-ellis
View: 213 times
Download: 0 times
Share this document with a friend
16
Retrospective and Prospective Utilitarianism Author(s): Brian Ellis Source: Noûs, Vol. 15, No. 3 (Sep., 1981), pp. 325-339 Published by: Wiley Stable URL: http://www.jstor.org/stable/2215436 . Accessed: 14/06/2014 00:53 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Wiley is collaborating with JSTOR to digitize, preserve and extend access to Noûs. http://www.jstor.org This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AM All use subject to JSTOR Terms and Conditions
Transcript
Page 1: Retrospective and Prospective Utilitarianism

Retrospective and Prospective UtilitarianismAuthor(s): Brian EllisSource: Noûs, Vol. 15, No. 3 (Sep., 1981), pp. 325-339Published by: WileyStable URL: http://www.jstor.org/stable/2215436 .

Accessed: 14/06/2014 00:53

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Wiley is collaborating with JSTOR to digitize, preserve and extend access to Noûs.

http://www.jstor.org

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 2: Retrospective and Prospective Utilitarianism

Retrospective and Prospective Utilitarianism

BRIAN ELLIS

LA TROBE UNIVERSY

According to G. E. Moore, 'the question whether an action is right or wrong always depends on its actual consequences'. ([ 15]: p. 101) He says that 'there seems no sufficient reason for holding either that it depends on-the intrinsic nature of the action, or that it depends upon the motive, or even that it depends on the probable consequences'. ([15]: p.101) Other utilitarians hold that the rightness or wrongness of an act depends, not on its actual consequences, but on the probabilities of its possible consequences, and their respective utilities.

Moore's position exemplifies what I call 'retrospective act- utilitarianism'. It is an act-utilitarian position, because the primary assessment is of individual acts rather than of groups or kinds of acts. It is retrospective, because, typically, the rightness or wrongness of an act can only be known after the event, i.e. once we have learned what its actual consequences are. Those who hold that the rightness or wrongness of an act depends in some way on the probabilities of its possible consequences are propsective utilitarians-prospective because, on such theories, it is possible to know in advance whether what one is planning to do is right or wrong, even though one does not know what the actual consequences of doing it will be.

There is little point in refuting ietrospective act-utilitarianism in the particular form in which it was held by Moore.' It may be of some interest, however, to show that no form of retrospective act- utilitarianism can survive criticism, because there is no such theory whichl can yield a satisfactory theory of obligation. The arguments against retrospective act-utilitarianism suggest that some form of prospective utilitarianism may be more satisfactory. However, we find that similar arguments can be mounted against any form of prospective act-utilitarianism. Hence it appears that no form of act-utilitarianism, retrospective or prospective, can be sustained.

The only theoretically defensible form of retrospective utilitar- ianism turns out to be form of group utilitarianism in which the

325

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 3: Retrospective and Prospective Utilitarianism

326 NOUS

primary objects of utilitarian assessment are not individual acts, but actions (not necessarily co-ordinated) of isolated groups of people, an isolated group being one the consequences of whose actions cannot be affected by the actions of others. However, there may not be any such groups short of the totality of all individuals, present and future. And if this is so, then the only defensible form of retrospective utilitarianism is a kind of utilitarianism in which the primary objects of utilitarian assessment are future realizable world histories.2 On this latter theory a particular act is categorically right only if it is necessary to bring about an optimally good world; it is categorically wrong only if it does not occur in an optimally good world that could be brought about. In any other case, it is only conditionally right or wrong, depending on what else is done, either now or in the future.

The difficulties with prospective act-utilitarianism lead to a similar conclusion. For ideally rational beings, with common belief and value systems, the only theoretically defensible form of prospective utilitar- ianism is one in which prospective utilities of total situations are primary.

But while such a comprehensive form of prospective utilitarianism may be theoretically defensible, it is not a practical ethic. It is an ethical system that only an omniscient, or near omniscient being could operate with. A practical ethical system must be adapted to the capacities of ordinary people. If it is not, then we will not be able to do what, theoretically, we ought to do. The principle that 'ought' implies 'can', in the sense of our being able to do what we ought to do if we want to, will be violated. To develop a practical utilitarian ethic, we must discuss practical strategies for maximizing utility-as much as it is humanly possible to do this. The concluding section of this paper is concerned with this practical question.

1. CRITIQUE OF RETROSPECTIVE ACT-UTILITARIANISM

An adequate theory of right and wrong, and of obligation, should be compatible with the following principle:

P1. X did wrong if and only if X ought to have acted otherwise.3

This is the axiom which links the theory of right and wrong to the theory of obligation in all classical utilitarian theories.4 I shall argue that there is no form of retrospective act-utilitarianism that is compatible with P1, and that no satisfactory utilitarian theory can be developed if P1 is rejected. I conclude that there is no adequate retrospective act-utilitarian theory.

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 4: Retrospective and Prospective Utilitarianism

RETROSPECTIVE AND PROSPECTIVE UTILITARIANISM 327

Suppose that X has a choice only between actions or courses of action, A1 and A2, and Y a choice only between actions or courses of action, B1 and B2. Suppose that the combinations A 1BI and A21B2 are both disastrous, and equally so, while the combinations A 1B2 and A2B1 are not disastrous, but equally good. Suppose thatX and Y are ignorant of each other's intentions and choose to do A1 and B1 respectively, with very bad consequences. Then, retrospectively we can say that X did wrong, and that Y did wrong. X did wrong, because had he chosen to do A 2 there would have been no disaster, and Y did wrong because had he chosen to do B2 all would have been well. Therefore, by P1, X ought to have done A2 and Y ought to have done B2. But had X done A2 and Y doneB2 the consequences would have beenjust as bad. Therefore, if PI is retained, no retrospective act-utilitarian theory determines a satisfac- tory theory of obligation. It should be noted that this refutation does not depend on how an action or course of action is defined, i.e. on what the alternatives open to an individual are considered to be.5 Therefore, the criticism applies to all retrospective act-utilitarian theories.6

To illustrate the argument, suppose that X and Y are driving in opposite directions along a road approaching the brow of a hill. Nei- ther can see, or is aware of the other's existence. X is driving on his left-hand side of the road and Y on his right-hand side. They collide at the top and both drivers are badly injured. Retrospectively, we can see thatX did wrong, because had he driven on the other side there would have been no accident, and we can say the same thing of Y. Therefore, by P1, X ought to have driven on the other side, and Y ought to have driven on the other side. But had they done so, there would still have been an accident.7

The retrospective act-utilitarian cannot appeal to the road laws. He cannot argue that the driver who was obeying the law did not do wrong just because he was obeying the law, because that would be to appeal to a rule, general observance of which has good consequences, to justify his claim that only the law-breaker did wrong. Such an appeal may be open to a rule-utilitarian, but not to a retrospective act- utilitarian. Nor can he argue, as a prospective utilitarian might, that since most people in the community obey at least this road law, the prospective utility of obeying the law is greater than the prospective utility of disobeying it. For this is to make the rightness or wrongness of the act depend not on its actual consequences, but on the probabilities of its possible consequences.

Two courses appear to be open to a retrospective utilitarian, i.e. one who insists on the retrospective assessment of rightness and wrongness. One is to abandon act-utilitarianism for a form of utilitar- ianism in which the actions of groups of people are the primary objects of evaluation. The other is to abandon the principle P 1, and try to develop

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 5: Retrospective and Prospective Utilitarianism

328 NOUS

an independent theory of moral obligation, e.g., by making what one ought to do depend on what is the most rational to do to maximize utility, rather than on what it is right to do. Let us consider these alternatives.

The first course involves saying that what it is right or wrong for an individual to do may depend on what others do. Thus, in the case we have been considering, we must say that whatX did was right or wrong depending on what Y did; or, if the actions of other individuals are relevant, then it depends on what they did too. X's action A was right categorically if every optimal set of actions which might be done in the circumstances includes X's doingA . His action was wrong categorically if it does not belong to any such set of actions. In any other case, X's action was only conditionally right or wrong, i.e. right or wrong depend- ing on what the others did. In the case we have considered,X was wrong to have done A1, given that Y did B1, and Y was wrong to have done B1, given that X did A 1. But we cannot say categorically that what X did was Wrong or that what Y did was wrong. Therefore, we cannot use P1 to draw the conclusion that X ought to have done A2 and Y to have done B1. We can use P1 to say whatX and Y ought to have done, viz. eitherA1 and B2 or A2 and B1, (provided that there is no third person whose actions were relevant) but we cannot use it to say what either X or Y ought to have done, since none of the alternative actions or courses of action open to them as individuals was categorically either right or wrong.

To take such a way out of the dilemma is to abandon act- utilitarianism for a form of retrospective-utilitarianism in which the actions of isolated groups of people are the primary objects of utilitarian evaluation. An individual action can be said to be right or wrong categorically only in rather special circumstances, and even then, its rightness or wrongness derives from that of the various sets of alternative actions or courses of action open to people at the time. It should also be noted that the move to this form of group-retrospective utilitarianism forces us to abandon at least one long-cherished princi- ple of deontic logic, viz.

P2 Every action is either right or wrong,

and hence also to abandon the equivalence of its not being wrong not to do A, and its being right to do A. We may have to take this way out, or something like it, but it must be seen as a radical departure from the traditional act-utilitarian framework. But first let us consider the other alternative.

If we reject P1, then we might, perhaps, be able to maintain a retrospective act-utilitarian theory of rightness or wrongness together

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 6: Retrospective and Prospective Utilitarianism

RETROSPECTIVE AND PROSPECTIVE UTILITARIANISM 329

with some independent theory of obligation. For example, we might be able to combine it with a prospective act-utilitarian theory of obligation. J. J. C. Smart, sometimes writes as though this were his position. His theory of rightness and wrongness is certainly retrospective, but his theory of obligation appears to-be p7rospective. On the one hand he says that he will 'reserve the word "right" as a term of commendation for the act which does infact produce the best results'. ([18]:47) On the other, he formulates his Act-Utilitarian Principle as a prospective utilitarian would-'Maximize probable benefit'. ([18]:12) If Smart's Act- Utilitarian Principle has the status of a categorical imperative, i.e. a principle on which people ought always to act, then Smart is certainly committed to rejecting P1, because he cannot reasonably claim that when people act on this principle they will always be acting rightly. Acting to maximize probable benefit will not always maximize actual benefit. If, on the other hand, the Act-Utilitarian Principle is the hypothetical imperative, 'If you want to maximize your chances of doing what you ought (categorically) to do, then you ought to maximize probable benefit,'8 then it is false. It is false because, if this is your aim, you should do what will probably maximize benefit, which is often different from what will maximize probable benefit.9

Quite apart from any objections there may be to a prospective act-utilitarian theory of obligation,10 there are serious objections to any such mixed retrospective-prospective theory of rightness and obligation in which the linking axiom P1 is rejected. First, a prospective act-utilitarian theory of obligation is based on the categorical principle that one ought always to act rationally to maximize benefit. Acting to maximize probable benefit is acting rationally to this end, according to defenders of this position. But then, why should one act rationally to maximize benefit? A mixed retrospective-prospective theorist cannot claim that it is always right to act rationally to maximize benefit. Hence he cannot give this as a reason why one ought to act in this way. He cannot put an independent value on acting rationally, and claim that the justification for acting rationally is the intrinsic value of doing so. A mixed theorist is required to develop a theory of obligation which is independent of his theory of rightness and wrongness. He is required to justify the categorical principle that one ought always to act rationally to maximize benefit, whatever the actual consequences of doing so may be. "

Secondly, if what one ought to do depends on what it is rational to do to maximize benefit, while the rightness or wrongness of acts is still dependent on actual consequences, what role does this leave for judgments of right and wrong? A right act is one which happens, perhaps miraculously, to have optimal consequences. A wrong act is one that does not have such consequences. One would have thought that to judge an act to be right, to think of it as such, is to view it with

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 7: Retrospective and Prospective Utilitarianism

330 NOUS

moral approval, as the kind of thing one ought to do or have done in such circumstances. But on a mixed theory this cannot be so. Given the circumstances never recur exactly, the fact that a given act turned out to be the right thing to do in the circumstances, carries no implications whatever concerning what anyone else, or the agent himself, should do in other circumstances, however similar to the first. And if the world is not deterministic, then even the same act, performed in identical circumstances, may not be right the second time around.

Thirdly, Moore made a distinction between what one is obliged or what it would be right to do on the one hand, and what it would be praiseworthy to do on the other. ([15]: 100) He held that it might be wrong to praise someone who in fact did what was right, or what he was morally obliged to do, or conversely. A similar distinction must also be made on any prospective act-utilitarian theory. For the prospective utility of praising someone for doingX might not be optimal even though the prospective utility of doingX is. On a mixed retrospective-prospective act-utilitarian theory, however, in which P1 is rejected, it is necessary to make both sets of distinctions. There is the question of whether the act X was right or wrong, and the distinct question of whether X was obligatory or not. There is the question of whether the actP of praising the agent for doing X was right or wrong, and the further question of whether the agent ought or ought not to be praised. And to these four questions there are 16 possible consistent sets of answers. If two people praise the agent for doingX, then there are six independent questions we can ask about the rightness or wrongness, obligatoriness or otherwise of the three acts in question, to which there are 64 consistent sets of answers. The conceptual cost of rejecting P1 is therefore very great, and one may wonder what point there could be in so separating the theory of right and wrong from the theory of moral obligation.

I conclude that any attempt to save retrospective act-utilitarianism by rejecting P1, and adopting an independent theory of obligation, leads to an unsatisfactory mixed theory, the normative content of which resides wholly in the theory of obligation. Act-utilitarian judgements of rightness and wrongness cease to have any direct moral content; they do not imply anything about what one ought or ought not to do. To retain a retrospective act-utilitarian theory of rightness and wrongness by adopting a mixed theory seems to be a pointless exercise which does nothing more than complicate the language of moral discourse. If we are going to adopt, say, a prospective act-utilitarian theory of obligation, then we should retain P1, and adopt a correlative prospective act-utilitarian theory of rightness and wrongness.

If, nevertheless, retrospectivity is to be preserved, and P1 retained, then act-utilitarianism must be abandoned for a form of utilitarianism in which the actions of isolated groups of people become the primary

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 8: Retrospective and Prospective Utilitarianism

RETROSPECTIVE AND PROSPECTIVE UTILITARIANISM 331

objects of utilitarian evaluation. By an isolated group of people, I mean a group of people the consequences of whose actions cannot be affected by the actions of other people outside the group. Whether there are such groups, I do not know. If not, then we must consider the actions of all people; and since the consequences of our actions today will presumably be affected by what is done in future generations, we must consider their possibilities for action too. So ultimately, if retrospective assessment of rightness and wrongness is to be preserved, along with the principle P1, we might have to adopt a form of utilitarianism in which the primary objects of assessment are realizable futures. On this view, a categorically right act will be one that is necessary to an optimally good future. A categorically wrong one will be one that does not occur in any optimally good world that might be realized. Anything else will be only conditionally right or wrong, depending on what else is done, now or in the future.

- Utilitarianism has usually been seen as a theory which provides a basis for individual morality. Neither the global form of utilitarianism, not any form of isolated group utilitarianism does this, except in special cases. But these are the only kinds of retrospective utilitarian theories, which retain P1 which are theoretically defensible. We have argued that there is no satisfactory mixed utilitarian theory. Therefore, if a form of utilitarianism is required that will provide an ordinary person in society with a satisfactory theory of obligation it cannot be any kind of retrospective utilitarianism.

The example of the crash on the hill suggests that some form of prospective utilitarianism may be better. Let us consider this possibility.

2. CRITIQUE OF PROSPECTIVE ACT-UTILITARIANISM

I assume that the prospective utility of an act is to be gauged by the probability-weighted average of the utilities of its possible conse- quences.'2 I assume also that a prospective act-utilitarian holds that an act is right if and only if it has maximal prospective utility, so defined, and that he derives his theory of obligation from this theory of right and wrong by using the principles P1 and P2.

My argument against prospective act-utilitarianism is somewhat similar to my argument against retrospective act-utilitarianism. Briefly, I show that there are cases in which, according to the theory, X ought to do A and Y to do B, even though, on the theory, and with the information available to all to do the necessary calculations, the prospective utility of the combination of actions consisting of X doing A and Y doing B is minimal. Hence the theory would make it right for X and Y, as individuals, to do what they as a pair ought not, and know they ought not, to do.'3

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 9: Retrospective and Prospective Utilitarianism

332 NOUS

Let C and C be the mutually exclusive possible outcomes of doing or not doing the actionsA andB. Suppose that the value of C, V(C) = 30 and the V(C) = 0. Then the utility of A, U(A) = 30 P(C/A), U(B) = 30 P(C/B), U(AB) = 30 P(C/AB), and so on. There are, however, cases in which

(i) P(C/A) > P(C/A) (ii) P(C/B) > P(C/B)

(iii) P(C/AB) > P(C/AB)

For example, let P(C/A) = P(C/B)= 1/2, P(C/A) = P(C/B) = 7/15, P(C/AB) = P(C/AB) = 2/3, P(C/AB) = 0 and P(C/AB) = 4/15. For these probability assignments, which are demonstrably consistent, the above valuations yield:

U(A) = 15 U(A) = 14 U(AB) = 0 U(B) = 15 U(B) = 14 U(AB) = 8 U(AB) = 20 U(AB) = 20

In this example, the combination of actions, AB, has the lowest prospec- tive utility, even though the actions A and B considered individually have the highest prospective utility. The combinations of actions, AB and AB, have the highest prospective utility of all, when the actions are considered in combination.

Suppose that A is an action to be done or not done by X, and B an action to be done or not done by Y. Given these conditional prob- abilities and evaluations, and the usual method of calculating prospec- tive utilities, it follows, by P1, thatX ought to doA and Y to do B. Hence if each person does what it seems he ought, the prospective utility of their actions considered together will be minimal. The prospective utility of neither person's doing what he ought, or of a single person's doing what he ought to do, is greater than the prospective utility of both persons' doing what they ought. Why then should either X or Y, given that they can make all of the relevant calculations, do what, according to the prospective act-utilitarian theory we are considering, do what he ought to do? Wouldn't it be just irrational for him to do so?

There are two possible responses that a prospective utilitarian can Aiake to this dilemma. One is to abandon the principle P1, and look for a theory of obligation which is independent of the prospective act- utilitarian theory of right and wrong. But the prospects of developing such a theory which would satisfy a utilitarian, and would not suffer from the defects of the mixed theory that we have already considered, seem remote. The other is to abandon prospective act-utilitarianism for a form of prospective utilitarianism in which the actions of isolated groups of people are the primary objects of prospective utilitarian evaluation. We shall not consider the first alternative. It the second

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 10: Retrospective and Prospective Utilitarianism

RETROSPECTIVE AND PROSPECTIVE UTILITARIANISM 333

alternative is adopted, then the principle P2 must be abandoned. For there will be some acts which are categorically neither right nor wrong. In the above case, for example, X's action in doingA will be right, given that Y does not doB, and Y's action in doingB will be right, given thatX does not do A. But we cannot say of either act that it is right categorically; nor can we say that either act is categorically wrong. Therefore, assuming that no satisfactory mixed theory can be developed, prospective act-utilitarianism suffers from the same kinds of defects as retrospective act-utilitarianism. Therefore, no form of act-utilitarianism, whether prospective, retrospective or mixed yields a satisfactory theory of obligation. Therefore, act-utilitarianism must be rejected.

I conclude that the only theoretically defensible form of prospec- tive utilitarianism is an isolated group or comprehensive prospective utilitarianism, in which the primary objects of prospective utilitarian assessment are the actions of groups of people, the probabilities of the consequences of whose actions are independent of what other people outside the group may do. either then or later. On such a theory, each person must consider what total realizable situations have maximum prospective utility, and do their part in bringing them about. If everyone did this, utility would be maximized as much as is theoretically possible.

3. TOWARDS A PRACTICAL UTILITARIAN ETHICS

If all people could agree on the prospective utilities of all of the various courses of action open to them all, and one could be reasonably sure that everyone would do what he had to to bring about a total situation of maximum prospective utility, then this comprehensive form of prospective utilitarianism would define a rational strategy for maximiz- ing utility. But people will disagree about what courses of action are open to them, what the possible consequences of the various actions are, what values should be placed on these consequences and the conditional probabilities of their realization, and on whether prospec- tive utility ought in any case to be maximized. Hence, while the comprehensive form of prospective utilitarianism may be a satisfactory theoretical ideal, it is not a strategy that is well-adapted to the aspirations and capacities of ordinary people. It defines an ideal of behaviour for ideally rational good-willed beings, who share the same beliefs, and have a common system of values.

These are perhaps cases in which I can be sure that a com- templated action would be what I should do to realize a total situation of maximum prospective utility. If the act that I am comtemplating would almost certainly do a great deal of good, whatever anyone else may do, and there seems to be nothing better that I could do in the circumstances, then we may have a case of this kind. But, in general,

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 11: Retrospective and Prospective Utilitarianism

334 NOUS

I cannot know what I ought to do if I am to act on the principle of comprehensive prospective utility. In some simple cases I may know what I ought to do. In other cases I may accidentally succeed in maximizing prospective utility. But in general the principle is beyond my capacity to follow. I just cannot always act in the required way to maximize prospective utility.14 Therefore, I cannot be obliged so to act. One could, of course, escape this conclusion by denying the principle that 'ought' implies 'can', i.e.

P3. If X ought to do A, then X can do A.

But I have no wish to deny it, because it seems to me pointless to insist that anyone has an obligation that he is incapable of fulfilling. The principle that one ought always to act to bring about a total situation of miximum prospective utility entails that one can always so act. Since one cannot, there cannot be any such obligation.

Of course, one can always try to act on the principle of comprehen- sive prospective utility. Therefore, one might have an obligation always to try to do so. But the prospective utility of everyone's trying to act on this principle may actually be quite low compared with some alterna- tives. The prospective utility of everyone's trying to follow some accepted or proposed moral code of a more specific kind may actually be much greater than that of everyone's trying to act on any of the utilitarian principles I have so far discussed. It will not do to object that if this is so, then the utilitarian principles (any of them) require that we should act on this moral code. Some forms of rule-utilitarianism might demand that we do so, but not these theories. Even comprehensive prospective utilitarianism does not require that we should act or try to act on the best practical strategy for maximizing utility, because the actions which would be produced by this policy would be very unlikely to coincide with the actions of ideally rational, good-willed beings with common beliefs and a common value system. And if they do not do so, then, according to comprehensive prospective utilitarian theory, it would be wrong to do them.

My main criticism of comprehensive prospective utilitarianism is that it is utopian. In principle, it requires us to consider and evaluate the possible consequences of all possible combinations of actions, whether done by us or by others, decide which set of actions has the greatest prospective utility, and to act accordingly. And the success of this strategy presumably depends on everyone else's making the same evaluations of, and, probably, assignments to, the consequences of the various possible actions, drawing the same conclusion about what set of acts in prospectively optimific, and being prepared to do whatever he or she must to see that these acts are done; ordinary human beings

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 12: Retrospective and Prospective Utilitarianism

RETROSPECTIVE AND PROSPECTIVE UTILITARIANISM 335

cannot do these things, therefore, it cannot be the case that they ought to do them.

We have seen that retrospective utilitarianism is unsatisfactory because it fails to determine a satisfactory theory of obligation, and that prospective act-utilitarianism is not a rational strategy for maximizing utility, even for ideally rational and good-willed beings. The com- prehensive form of prospective utilitarianism is a rational strategy for such beings, but it is ill-adapted to the capacities of ordinary people. It seems to me, therefore, that rational utilitarians should start to think more practically about maximizing utility. They should take into account human differences, limitations and frailties in devising strate- gies for doing so. They should not aim at maximizing utility as much as is theoretically possible, disregarding human limitations, but to do so as much as possible, given human limitations. For it is only by doing so that they can arrive at practical strategies and codes of behavior adapted to human capacities. Suppose our aim is to beat Karpov at chess. It does not help much to be told that we should play in such a way as to checkmate him. It is no better if we are told that we should play so as to maximize our chances of winning. Nor does it help if we are told that we should play to achieve an advantageous position. Yet the kind of advice utilitarians offer us generally seems to be like this. It is really no advice at all, and we are no better off for having it. The important questions have not even begun to be answered.

To maximize utility, as much as humanly possible, I assume that the main problem is to set objectives and standards of human behaviour, which are capable of acting as guides to life, which do not demand of ordinary human beings more than they are capable of doing and achieving, which ordinary human beings are able to live by throughout their lives, which are specific enought to determine what courses of action are to be considered obligatory in the specific situations in which people are likely to find themselves, and which are as well-suited as possible, given these requirements, to realizing total situations of maximal prospective utility. In short, we need some strategic objectives and standards.

If one's aim is to beat Karpov at chess, it is necessary to examine his play carefully, to try to discover his weaknesses, if any, and to capitalize upon them by adopting a strategy which will exploit them. Suppose he has a weakness in defending himself against a king-side attack. In that case, perhaps the best strategy is to prepare to mount such an attack. Likewise, if one's aim is to realize a total situation of maximal prospective utility, one must examine the society in which one lives very carefully, try to discover how and why people are suffering, or are not as happy as they could be, look for the practical strategy which seems best suited to dealing with these problems, and, presumably, advocate

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 13: Retrospective and Prospective Utilitarianism

336 NOUS

its adoption by the community. The strategy might involve setting sub-goals such as removing racial or sexual discrimination, or achiev- ing a state in which people may live without fear of starvation, deprivation, exploitation or domination. Or, it might involve defining codes of behavior by which ordinary human beings could reasonably be expected to live, which offer hope of bringing about a total situation of maximal achievable prospective utility.

A practical and rational utilitarian will reject the utopianism of comprehensive prospective utilitarianism. He will not say, or even think, that ordinary mortals ought to act in such ways as to play their part in bringing about total situations of maximum theoretically possible prospective utility. He might think that omniscient beings ought to act in these ways; but who cares how omniscient beings should act? What is important is what ordinary mortals should do. And this depends on their capacities. If striving to achieve various more limited objectives, or trying to live by codes of behaviour, which are as well-suited as any other practical codes to utilitarian goals, is the most human beings are capable of, then indeed they ought to strive to achieve these objectives, and to live by these codes.

In different societies, or at different stages in a given society, different problems are paramount, and different strategies are re- quired to cope with these problems. Therefore, we should not expect to be able to formulate a practical ethical system applicable to all societies at all times. To expect to find such a system is like expecting to find a universal chess strategy which could be used against any player whatever his strengths or weaknesses. I see no reason to believe that there is such a system.

To evaluate a proposed strategy for maximizing utility, we must consider the prospective utility of adopting it. To do this is no easy task, and it may be that no rational evaluation of strategies can be made. Nevertheless, the task is not completely hopeless, for different societies at different times, have evolved various strategies to deal with their problems, and we have at least this experience to call upon. If the strategy that is proposed is a minor variation of an existing moral code, which is operating effectively within the society, then at least we have a base from which to consider the effects of the variation. How well does it address itself to the identified ills of society?

If our aim is to maximize utility, then any strategy which may be devised to achieve this goal ought not to be too rigid. If one could see an obvious 'mate in two', it would be madness to go on developing one's king-side attack. Similarly, if one has the opportunity of doing a great deal of good by acting in ways contrary to the rules contained in the strategy that one has adopted, and there is little likelihood that it will lead to any general abandonment of these rules, then one should take

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 14: Retrospective and Prospective Utilitarianism

RETROSPECTIVE AND PROSPECTIVE UTILITARIANISM 337

that opportunity. Strategic rules and objectives adopted for the purpose of maximizing utility should not be followed or pursued blindly, without thought of the purpose for which they were adopted. I would not advocate 'rule worship'.

Nevertheless, the kind of strategic utilitarianism I think plausible obviously has a great deal in common with some forms of rule- utilitarianism. A strategy consists of a set of rules and sub-objectives geared to the achievement of some overall goal. I am convinced that we need strategies to maximize utility. A good strategy has to be practical, i.e. one that ordinary people can adhere to. The best utilitarian strategy, for a given society at a given time, would be the practical strategy that has the greatest prospective utility. My belief is that the best strategy would normally be a flexible one, i.e. one that allows individuals to break the rules where they see significant utilitarian advantage in doing so. The position of strategic utilitarianism is therefore not a standard rule-utilitarian one. First, the moral rules would be considered to be strategic rules to be acted upon unless there is a decisive utilitarian reason for not doing so. Secondly, the moral rules could not be assessed individually for their utility. Only the overall strategy within which they are incorporated could be so assessed. The moral rules upon which we ought normally to act would be those that belong to the best overall strategy. On the other hand, the position is not an act-utilitarian one. First, the moral rules that belong to the best overall strategy for maximizing utility are not dispensable, and are therefore not, in the ordinary sense, rules of thumb. They are not replaceable by better rules, as rules of thumb normally are, because if they were, they would not belong to the best overall strategy for maximizing utility. On the contrary, they would have an essential role in any such strategy. Secondly, the main focus of attention would be on strategies and the elements of which they are comprised, viz. practical ethical codes and sub-objectives. It is overall strategies for maximizing utility that would have to be evaluated to determine how people ought to act.

However, although a strategic utilitarianism of the kind outlined here may be theoretically defensible, it is not obvious that it would determine a satisfactory moral theory, because moral theories are essentially about how individuals ought to live. Strategic utilitarianism might provide a foundation for a political theory which would have implications concerning individual lives and behaviour; but any such theory is likely to leave wide areas for individual choice, and to permit a wide range of possible life-styles. Hence, the basic question of how one ought to conduct one's life may not be answerable on any utilitarian theory.

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 15: Retrospective and Prospective Utilitarianism

338 NOOS

REFERENCES

[1] Aqvist, L., "Improved Formulations of Act-Utilitarianism," Noubs 3(1969): 299- 323.

[2] Barnes, G., "Utilitarianisms," Ethics 82(1971): 56-64. [3] Bergstr6m, L., The Alternatives and Consequences of Actions (Skockholm, 1966). [4] , "Reply to Bronaugh," Theoria 30(1972): 148-49. [5] ,"On the Formualation and Application of Utilitarianism," Nofts 10(1976):

121-44. [6] 9"Utilitarianism and Future Mistakes," Theoria 43(1977): 84-102. [7] Bronaugh, R.N., "Bergstrom's Utilitarian Objection to T,," Theoria 38(1972):

145-47. [8] Castafieda, H-N., "A Problem for Utilitarianism," Analysis 28(1968): 141-2. [9] , "Ought, Value, and Utilitarianism," American Philosophical Quarterly

6(1969): 257-75. [10] _ "On the Problem of Formulating a Coherent Act-Utilitarianism," Analysis

32(1972): 118-24. [11] ,"Open Action, Utility and Utilitarianism" in Leach et. al. (eds.), Science,

Decision and Value, (Rediel, 1972): 128-47. [12] Feldman, F., "World Utilitarianism," in K. Lehrer (ed.), Analysis and Metaphysics

(Dordrecht, Holland, 1974). [13] Gauthier, D.P., "Coordination," Dialogue 14(1975). [14] Lewis, D.K., Convention (Cambridge, MA: Harvard University Press, 1969). [15] Moore, G.E.,Ethics (London: Oxford University Press, 1966), originally published

in the Home University Library, 1912. [16] Narveson, J., "Utilitarianism, Group Action, and Coordination," Nouls 10(1976):

173-94. [17] Pritchard, H.A., Moral Obligation, and Duty and Interest (London: Oxford University

Press, 1968). The essay Moral Obligation was first published by the Clarendon Press in 1949.

[18] Smart, J.J.C. and Williams, B., Utilitarianism: For and Against (Cambridge Univer- sity Press, 1973).

[19] Sobel, J.H., "Utilitarianism and Past and Future Mistakes," Nouls 10(1976): 195-219.

[20] Westphal, F., "Utilitarianism and Conjunctive Acts," Analysis 32(1972): 82-5. [21] Zellner, H., "Utilitarianism and Derived Obligations," Analysis 32(1972): 124-25.

NOTES

'The deficiencies of Moore's relatively crude form of retrospective act- utilitarianism are well known. It has effectively been refuted by H-N. Castafieda in [9] and by L. Bergstrom in [3].

2This position is to be distinguished from F. Feldman's "World Utilitarianism" [ 12] in which the alternatives open to an individual are future life histories. Here, what is up for assessment are future realizable world histories.

3I do not, of course, maintain, as R. Price evidently did (see [ 17]: 151) that rightness entails obligatoriness. Throughout this paper I use the words 'optimal' and 'maximal' in the senses of 'not worse' and 'not less' than any others, and 'optimific' and 'maximum' in the senses of 'better' and 'greater' than any alternative.

4The principle P1 has not often been questioned in the philosophical literature. I know of no good evidence that it was challenged by any of the classical utilitarians- Jeremy Bentham, James Mill, Henry Sidgwick or G.E. Moore. J.J.C. Smart seems, by implication, not to accept P1. I shall discuss his views presently.

5Several attempts have been made in recent years to refine the notion of 'an alternative open to an individual' in the attempt to formulate a consistent form of act-utilitarianism, and whether, and if so what proposals have been successful, has been a matter of considerable philosophical debate. Some of the main contributors to this debate have been L. Bergstr6m, H-N. Castafieda, J.H. Sobel, L. Aqvist, R.N. Bronaugh, F. Westphal and H. Zellner. See bibliography for details.

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions

Page 16: Retrospective and Prospective Utilitarianism

RETROSPECTIVE AND PROSPECTIVE UTILITARIANISM 339

6It should also be noted that the argument is independent of any solution which may be found for utilitarian co-ordination problems, since it concerns only retrospective act-utilitarian assessment. We may suppose that neitherX nor Y knew that he had such a problem to solve. Co-ordination problems of the prisoner's dilemma kind, and of the kind discussed by G. Barnes in [2], have been widely discussed elsewhere. (See references [13], [14] and [16].) Many writers seem to think that these problems can be resolved within aprospective act-utilitarian framework. We shall see that this is not so. (See Section 2 below.) But in any case, whether they have a solution within such a framework is irrelevant to the retrospectively assessed rightness or wrongness of what X or Y did.

71t is a familiar point in the logic of counterfactuals that 'A -B' and 'CO-B' do not together entail 'ACO-*B'. The argument here trades on this point.

8The suggestion that the Act-Utilitarian Principle, as he formulated it, should be regarded as a hypothetical imperative in which a categorical 'ought' is buried in this way is Smart's. However, since the suggestion was made as an 'initial response' to an earlier version of this paper, I do not know whether he wishes to defend it.

9The distinction is an elementary one. The probable benefit of an action is maximal if the probability weighted average of the benefits of its possible consequences is maximal.

?0See section II below for a critique of prospective act-utilitarianism. "1What the prospective act-utilitarian has to justify is acting on a principle

observance of which would in a sufficiently long run have better consequences than acting on any other principle or rule. This may seem to make him into a kind of rule-utilitarian. But a prospective act-utilitarian can reply that his principle subsumes all others. For if any other principle would in the long run have better consequences it would be rational to act on it.

121 disregard other strategies, such as mimimax and miximin, which are either safety strategies, or strategies that are not of purely utilitarian design (because they are also concerned with questions of distributivejustice). I do so, not because I think that no good utilitarian case can be made out for adopting some safety strategies, but because it does not affect my argument much how the prospective utility of an action is defined, so long as it is defined as a function of the values and probabilities of its possible consequences. For a discussion of such strategies, and a new proposal, see [11].

13J.H. Sobel in [19] draws attention to some other difficulties with this particular form of prospective utilitarianism. He suggests two ways of avoiding these difficulties. One involves restricting the prospective act-utilitarian principle to what he calls 'basic options'. The other is to make 'relevant' not the instrumental values (i.e. prospective utilities) of the actions open to choice, 'but instead the instrumental values of the choices of these actions'. (383) Thus he obtains 'For any action a and agentx, if a is open tox and, for each action d' that is an open alternative to a, the instrumental value of the choice of a is greater than that of the choice of d', then a is right.' (p. 383) My argument against prospective act-utilitarianism also applies to Sobel's modified prospective utilitarian principle.

141 cannot do it in the sense that I cannot beat Karpov at chess. I may have the physical capacity to do the required actions, if only I knew what they were, just as I have the physical capacity to move the chess pieces into a checkmating position. But what I can or cannot do, in the sense in which the term is being used here, is not determined by my physical capacities.. It also depends on my mental capacities and capacities for knowledge. I believe that 'ought' implies 'can' in this strong sense of 'can'. Castan-eda makes a similar point in [9], p. 263.

This content downloaded from 185.2.32.89 on Sat, 14 Jun 2014 00:53:52 AMAll use subject to JSTOR Terms and Conditions


Recommended