Utilitarianism: The Classical Principle and the Average Principle

I

Act Utilitarianism has traditionally been regarded as the view that you should always perform the action that will bring about the greatest possible excess of happiness over unhappiness or, if there is no such alternative, the least possible excess of unhappiness over happiness [I am indebted in this paper, as is often the case, to D.C. Brown. The paper has also benefited by comments and criticism from William Anglin]. Following Rawls, I shall call this the classical principle. An alternative which Rawls calls the average principle is the view that you should always do the thing that will bring about the highest possible average happiness level. Rawls, Rescher and Broad [John Rawls, A Theory of Justice (Harvard University Press, 1971), p. 162ff; Nicholas Rescher, Distributive Justice (The Bobbs-Merrill Company Inc., 1966), p. 27ff; C.D. Broad, Five Types of Ethical Theory (Routledge and Kegan Paul, 1930), p. 250] regard the average principle as superior to the classical principle, and there are considerable grounds for supposing that Mill accepted the average principle [Cf. Gunnar Myrdal, The Political Element in the Development of Economic Theory, trans. Paul Streeten (London: Routledge and Kegan Paul, Ltd., 1953), p. 38]. Smart favors the classical position but confesses that if someone doesn’t feel the same way, he doesn’t know how to argue with him [J.J.C. Smart and Bernard Williams, Utilitarianism for and Against (Cambridge University Press, 1973), p. 28]. It is not immediately apparent that an action can be right in terms of one of the two principles but wrong in terms of the other. They do not, indeed, conflict in cases where the population will not be altered by one’s action, because in such cases the average level of happiness is a function of the total amount of happiness and of nothing else. But in cases where the population will be altered by the action, the two principles would frequently dictate different actions. For example, an action that led to an increase in the population might increase the excess of happiness over unhappiness but decrease the average level of happiness.

It might seem that the issue between advocates of the two principles is of concern only to utilitarians, but this is not so. Almost all deontologists are concerned with the consequences of actions. They differ from utilitarians, of course, by holding that other factors should also be considered in determining what we ought to do. But insofar as they are concerned with consequences, they would need to decide between less extensive counterparts of the classical position, the average position, and several other alternative positions I shall discuss later.

The stand one takes on this issue will have implications in matters of great practical concern. I shall discuss three of these.

  1. Consider the possibility of the destruction of the world. An advocate of the average position might look upon this possibility with considerably less concern than an advocate of the classical position. Proponents of both positions would be concerned with the suffering of the people who would exist at the time the world was destroyed and the loss of their lives; but an advocate of the average position might have grounds for being far less concerned than an advocate of the classical position with the loss of future generations. If the future generations would be happier than the people who have already existed, he would have to be concerned about them, but he would not need to be concerned if he expected them to be less happy, because they would only pull down the average. An advocate of the classical position, on the other hand, would have to be concerned about their existence so long as he felt that their lives would on the average be worth living. One might suppose that it wouldn’t matter very much whether or not one had the additional concern for the continuation of life on our planet that an advocate of the classical position would probably have, because advocates of both positions would be firmly committed to avoiding the same catastrophe. But important differences could arise between them. The advocate of the classical position might feel that far greater sacrifices on the part of people now living would be justifiable for even a small chance of avoiding the catastrophe than the advocate of the average principle would be prepared to sanction. And an advocate of the average position might be prepared in order to improve the lot of people now alive to take risks of a nuclear war that would seem intolerable to an advocate of the classical position.
  2. Consider, secondly, the issue as it arises in our treatment of animals. Peter Singer has provided forceful arguments to the effect that we should be concerned with the happiness and suffering of animals as well as of human beings [“And Now It's the Animal Lib Movement,” The National Observer, April 28, 1973]. I agree with him that our treatment of animals is often shockingly immoral, but it seems to me that there is room for disagreement on the question of whether animals should be raised to be used as food. The problem arises because these animals would not exist if the people who raise them did not intend to sell them for food. An advocate of the classical position could argue that if the animals were given lives that were worth living — something that does not usually happen now — it might be right to allow people to raise them for food on the grounds that there would be a greater total amount of happiness if this were allowed than if it were not. An advocate of the average principle, on the other hand, would have to hold that if the animals would have a life that was less happy than that of the average sentient being, they should not be bred, even if they would have lives that were worth living.
  3. Advocates of the two principles could, of course, disagree on how much birth control should be practiced. However, in the present overcrowded state of the world, they would be likely to end up in agreement more often than not, because adding extra people is likely to pull down the total happiness excess as well as the happiness average. I should add that when there was disagreement it is by no means obvious that an advocate of the classical position would end up favoring less birth control than an advocate of the average position. Too much population may lead to our extinction, and we have seen earlier, an advocate of the classical position would have more incentive than an advocate of the average position for trying to avoid this catastrophe.
II

Now for a discussion of the principles. My own view is that when they conflict, we should sometimes do the thing that will bring about the highest average level of happiness; sometimes the thing that will bring about the greatest excess of happiness; and sometimes accept a trade-off between the two. I shall call this the trade-off position.

Consider the following case. A couple is deciding whether to have two children or six. Whether there are two children or six, they may be expected to have much more happiness than unhappiness in their lives. But if there were only two, let us assume that their average level of happiness would be slightly higher than it would be if there were six children. If these were the only two alternatives, and the question was to be decided in terms of the welfare of the children, I would suppose that the parents should have six children. But if additional alternatives were possible, I might hold that a trade-off would be in order. It might be the case that they should do something that would neither maximize the average level of happiness of the children, nor bring about the greatest excess of happiness over unhappiness — that neither the classical nor the average principle should be followed.

Consider next a case like the last except that this time if there are six children and we add together the excess happiness of all of them, it will only slightly exceed the amount of excess happiness there would be if there were only two children. It follows that in this case if there are six children, the happiness excess of each child will be only about a third as great as it would be if there were two children. In the case in question, I would suppose that the parents should follow the average principle and only have two children.

III

So much for my position. Now for some ways in which it could be attacked. The following argument could be used in defense of the average principle and against the classical principle or the trade-off principle.
  1. Whenever we have an obligation, it is to some person or persons.
  2. There cannot be an obligation to a possible person, only to an actual one, or to a person whose existence is from the point of view of the agent inevitable.
  3. Whenever the classical principle conflicts with the average principle, the only thing that could recommend adherence to the classical principle would be the fact that if you followed it you would bring about the birth of people who would have more happiness than unhappiness and who would not be born if you followed the average principle.
  4. If you had an obligation to bring about the birth of the people in question, it would have to be an obligation to them, that is, to certain possible persons.
  5. But, (by 2.), we could never have such an obligation.
  6. Therefore, we never have an obligation to follow the classical principle.
The second premise, the one about obligations to possible persons, seems to me to be open to question. Let us suppose that a woman has found out that if she has a child he will be so seriously defective that he will almost certainly have such an unhappy life that he will wish that he had never been born. Despite this knowledge, the woman would like to have the child. She has had severe emotional problems and she feels that she would be happier caring for a defective child than not having a child at all, and she is unable to adopt a child. Let us assume that she would in fact be somewhat happier if she had the child. From the point of view of society as a whole, the advantages and disadvantages of her having a child would balance about equally. The disadvantages are obvious; the advantage is that the woman would be less of a problem for society if her emotional problems were decreased. Most people (though by no means everyone) would suppose that it would be wrong for the woman to have a child —that she would have an obligation not to have it But her obligation not to have it could not be an obligation to society because society wouldn’t lose, and it couldn’t be an obligation to herself because she would gain by having the child. It would seem to be an obligation to the unborn child. Of course, one could drop the first premise of the argument, the claim that you can only have an obligation if there is some person to whom you have it. But there is a good reason for dropping instead the premise that you can’t have an obligation to a possible person. Because in this case it seems that you do, since if the child were born and found his life miserable, it would surely be in order for him to blame his mother.

This is enough to refute the argument we have been considering against the classical position and my mixed position. But a new argument could be constructed. It could be claimed that there is the following asymmetry between the case in which you are responsible for bringing an unhappy being into the world and the case in which you refrain from bringing a happy one into it. In the former case there will be someone to complain; whereas in the latter there will not. Because of this, one could substitute for (2), the claim that it is most possible to have an obligation to a possible person, the claim that it is possible to have such an obligation but only if the possible person may at some time be in a position to claim that he has been injured. Clearly, if you are responsible for the birth of an unhappy person, he will be on hand to complain, but if you are responsible for preventing the birth a happy person he will not be on hand to complain.

With the aid of the new second principle we can construct the following revision of our earlier argument against the classical position:
  1. Whenever you have an obligation it is to some person or persons.
  2. You cannot have an obligation to a possible person unless the possible person may at some time be in a position to claim that he has been injured.
  3. Whenever the classical principle conflicts with the average principle, the only thing that could recommend adherence to the classical principle would be the fact that if you followed the classical principle, you would provide a necessary condition for the birth of people who would have more happiness than unhappiness and who would not be born if you followed the average principle.
  4. If you had an obligation to provide a necessary condition for the birth of the people in question, it would be an obligation to them.
  5. But they would be possible people who would never be in a position to claim that they have been injured. (If they aren’t born, they will not be in a position to claim that they have been injured because they won’t exist. If they are born, they won't be in a position to claim that they have been injured because they won’t have been injured.)
  6. By (2) you cannot have an obligation to possible people who will never be in a position to claim that they have been injured.
  7. Therefore we never have an obligation to follow the classical position.
In order to answer the new argument I need another example. A has instituted a birth control program in a world that is not threatened with over-population; B could use his influence to stop the program. If the program is stopped, a great many additional persons will be born. Since we are dealing with large numbers of people, it is possible to determine statistically that it is virtually certain that some of them would have very unhappy lives, but it is also almost certain that most of the people in question would have happy lives. In such a case we would clearly regard it as morally permissible for B to refrain from using his influence to stop the birth control program. Let us compare this with the case of the mother who would have had an obligation to take measures to avoid having a child. In both cases we have good reasons to suppose that someone would be born who would have a very unhappy life. What is it then about the second case that makes it morally acceptable to refrain from preventing birth while it is not morally acceptable to do so in the first case? The obvious answer is that it is the happy people included in the second case but not in the first.

But why is this the case?

I can think of only two answers that seem plausible. Both of them are consistent with either the classical position or the trade-off view. (1). It could be claimed that there would be a prima facie obligation to the possible happy beings which outweighed the prima facie obligation to the possible unhappy ones. Or it could be claimed (2). that it would be desirable for the happy people to be born and that the obligation to the possible unhappy persons is outweighed (or balanced) by the obligation to bring about the desirable birth of happy persons. This view could be based on the supposition that ethical judgements reflect the things that a speaker wants and that one can want to bring about the existence of happy persons. I don’t feel that (1.) is indefensible, but if I did, I would regard (2.) as an acceptable alternative.

I can think of only one move for a philosopher who found both of these ways of accounting for our response to the birth control example unacceptable and who wanted to reconcile that response with an attack against the classical and the trade-off positions. He claims that we have no obligation to bring happy people into the world even when there are no disadvantages to the people already there in doing so, or when the disadvantages and advantages cancel each other out. He must give an explanation then of the fact that the happy lives make a crucial moral difference between the second case and the first, without admitting that, other things being equal, we should bring happy people into the world whenever there is an opportunity to do so. There is one claim that he could make. He could hold that we never have an obligation to bring happy people into the world,but that when we are confronted with a package that contains happy people and unhappy ones, it is morally acceptable to take the whole package so long as the happy lives sufficiently outweigh the unhappy ones. This principle strikes me as very strange. According to it, you could justify bringing about the existence of people who would wish that they had never been born in order to bring about the existence of happy people, but you wouldn’t have an obligation to bring about the existence of happy people, no matter how happy they would be, even if you could do so with no cost to anyone including yourself.

It has been suggested to me that there is in fact no prima facie obligation to prevent the birth of the very unhappy people, and therefore, that there is no need to hold that there is (in order to outweigh such an obligation) a prima facie obligation to bring about the birth of the people who would probably be happy. The defense is that we do not know who the unhappy people would be. But it seems to me to be a mistake to attribute this much significance to our lack of this information. Let us suppose that you knew that if a certain chemical were put in the water supply of a city the birth of a number of very unhappy people would be prevented and that no other significant consequence would result from use of the chemical. It seems obvious that in those circumstances you would have a prima facie obligation to put the chemical in the water. It is arguable though I think wrong that you would have a greater obligation if you knew which people would have the very unhappy offspring, but it is surely obvious that you would have some obligation even if you lacked that information.

IV

Although my attacks have all been directed against the average position, an advocate of that position would probably try out various other positions in response to my criticisms. Strangely enough, any argument in favor of the average position that relies on the supposition that we have no obligations to possible persons is only suited to the defense of a somewhat different view from the average position. The reason for this is that the average position would seem, paradoxically, to leave us with obligations to possible persons. For example, if it were possible to bring about the existence of a happier-than-average person whose existence would neither increase nor decrease the average happiness of everyone else, the average principle would dictate our doing so — there would be an obligation to this possible person, the obligation to bring about his existence. In fact, there could be an even more extreme case. We would have an obligation to bring about the existence of a person or persons, if their existence made the overall average higher, even though their existence decreased the average happiness of all those now existing. Thus it would seem that what a philosopher who used the premise that we can’t have obligations to possible persons would actually want would be a principle requiring us to maximize the average only for persons who exist and persons whose existence is, from the point of view of the agent, inevitable.

Peter Singer considers this position [In “A Utilitarian Population Policy” to appear in the forthcoming Ethics & Population, edited by Michael Bayles and published by Shenkman. The quotation that follows is from this paper, of which he has kindly sent me a copy], but he sees that it requires an alteration. In cases of governmental policy, attempts to regulate population that are at all successful will affect not merely the total number of people who will be born but also what particular persons will be born. A given person will be born only if a certain sperm is joined with a certain egg: a difference of a fraction of a second as to the time of copulation would lead to the birth of a different person. It is obvious then that minute changes in the lives of the parents before the mother was impregnated would have an effect on who it is that is born. But an effective governmental population policy would be bound, either directly or indirectly, to have at least minute effects on the lives of almost everyone. Accordingly, such a policy would have as a side effect the births of an almost completely different batch of people from those who would otherwise have been born. It won’t do then to say that a government devising a plan in regard to population should be concerned only with persons who actually exist or whose birth is inevitable because from the point of view of the government there is almost no one whose birth is inevitable. Still, Singer suggests, it is inevitable, no matter what population policy is adopted, that some minimum number of people will exist in a given country at a given time in the future (for example, it might be inevitable on any feasible population policy that there will be at least 25 million people in Canada by 2000 A.D.). In response to this sort of consideration. Singer switches to the following position:
If a possible future state of affairs is a world of P people at an average level of happiness A, it is wrong to bring into existence any greater number of people, P + N, such that no sub-group of P + N contains P people at an average level of happiness equal to or higher than A.
He thinks that this gives him what he wants. In fact, it doesn’t because (as he makes clear earlier in his paper) he wants a principle that will assure the highest possible average happiness for the group or subgroup of P persons, but his principle doesn’t meet this requirement. It would allow population planners to limit the population to P persons even if adding additional persons would make it possible to have a sub-group of P persons whose happiness level was higher than the happiness level of P persons would be if there were only P persons. To avoid this, his principle should be modified as follows:
If a possible future state of affairs is a world of P people at an average level of happiness A, it is wrong to bring into existence any greater number of people, P + N, such that no sub-group of P+ N contains P people at an average level of happiness equal to or higher than A, and it is wrong to fail to bring into existence additional persons if they would increase the happiness of a subgroup of P persons.
(My change in his principle is in italics.)

This principle is equivalent to the following formulation which is shorter and possibly easier to grasp:

If the smallest possible population at some time in the future is P, we should aim at a population such that the happiness level of a group of P members of it would be as high as possible.

This is, I believe, the principle that Singer wants, but the following counter-intuitive consequences would seem to follow from it: it would be wrong to prevent the existence of anyone whose existence would increase — no matter how little — the average happiness of the crucial sub-group, even if the person in question would be so wretched that he would wish that he had never been born. Singer is well aware of the difficulty but hopes to take care of it in the following way: he notes that if a utilitarian knew that a newly born child would have a miserable life he would have an obligation to perform an act of euthanasia on the unfortunate infant because it would exist and would thus have the rights of an actual person. This would provide (albeit in an unusual way) for the welfare of the child, but Singer observes that the parents would probably be distressed at having this done; and Singer finds in this distress an indirect justification for preventing the existence of wretched persons. But what if (as would be the case in many societies) a standard policy of performing euthanasia on newborn infants with wretched prospects was for one reason or another simply not on the cards? They couldn’t then be destroyed; they would have to live their wretched lives. In such a case the fact that the creatures in question would lead unhappy lives if they were born would clearly provide a good reason for preventing their conception, but this is a reason that Singer’s theory would be unable to accommodate. To put his problem in more general terms: to avoid the counter-intuitive consequence he must show that whenever we know that if a person is born his life will be wretched there will be adequate utilitarian reasons for preventing his birth that do not include considerations of his own welfare. If an argument for preventing the birth of such a person hinged on the future wretchedness of that person, Singer would be forced to reject it.

Singer has suggested to me that instead of having an obligation to prevent the birth of wretched persons you would have an obligation to try to alter the attitudes of people towards euthanasia for infants likely to have wretched lives. But there would surely still be cases where you would know that you couldn't lead the relevant people to practice euthanasia. Also even if this were not the case — even if there were no cases where euthanasia was in order and you couldn’t arrange that it be practiced — a satisfactory ethical system should give acceptable answers for hypothetical cases as well as actual ones.

Singer has argued quite correctly [In “Sidgwick and Reflective Equilibrium,” The Monist, July 1974] that an ethical principle can sometimes withstand counter-intuitive consequences: we may decide that our initial intuitions were mistaken. But in the present case I can’t see that this is at all likely to happen, and I expect that Singer would agree with me in this. He could argue that the principle in question wouldn't require us to bring very many wretched persons into the world for the sake of small gains to others, but I feel sure that he would agree that it shouldn't require us to do that sort of thing at all.

Could Singer still save a part of his position by shifting to a view that allowed for obligations to wretched extras (persons outside the crucial sub-group) but precluded (except as a tie-breaker) obligations to happy extras? I think not, because I could confront him with an analogue to my birth-control program argument in section II because it would be permissible to refrain from preventing the birth of a few wretched extras if they came as a part of a package that would contain a great many happy extras.


Utilitarianism: The Classical Principle and the Average Principle
Author(s): R. I. Sikora
Source: Canadian Journal of Philosophy, Vol. 5, No. 3 (Nov., 1975), pp. 409-419
Published by: Canadian Journal of Philosophy
Stable URL: http://www.jstor.org/stable/40230580