Classical utilitarians want to maximise the happiness in the world, conceiving of happiness as subjective feelings of pleasure. Really, they want to maximise the balance of pleasure over pain. Preference utilitarians want to maximise the sum of preference satisfaction.
These ideas, however, quickly lead us to apparent paradoxes, or at least disturbing thoughts. If we are morally required to maximimse utility (whether it is pleasure or preference satisfaction), then we'd better produce as much of it as possible. The more pleasure there is in the world, or the more satisfied preferences, the better. If this is our responsibility, we seem to be committed to what Peter Singer calls the total view, i.e. we are concerned with the overall total of utility in the world, and it does not matter how this is achieved. But here are some (not very original) thoughts about why the idea is so disturbing.
Think of the kind of case where a couple decide not to have any more children, or decide not to have children at all, even though the children would probably be happy. It seems as if this couple has done something morally wrong. I know that childless-by-choice couples are often thought of as morally suspect, though I strongly disagree, but for those who might be persuaded by this idea imagine a couple who already have four kids. They could have a fifth - perhaps this will be at some cost to their own happiness, but they are likely to raise the total sum of utility in the world. If they decide to have no more kids, should we say they have done a morally wrong thing? That seems to be a repugnant conclusion (to borrow a phrase from Derek Parfit).
The total view also seems to work badly in cases where the use of contraception is compared with infanticide. If one couple decide not to have children, so they use contraceptives, they are declining to add to the amount of happiness in the world. But even more paradoxically, the immorality of their action seems to be no greater than that of a couple who have a kid, then kill it in some painless way. Once again, they have declined to add to the sum of happiness in the world. Yet, surely there is a moral difference between the two cases!
More generally, the total view would seem to require that we have as many babies as we can afford to raise with a tolerably decent standard of living and prospect of being happy. It requires us not just to improve the lives of existing people, but to create new people. But surely this conflicts with most people's intuitions. Most of us probably think that if you do have a baby you (morally) should look after it, but we don't think you are morally obliged to have children — and certainly not to have as many as you can.
Philosophers sometimes try to avoid the problems of the total view by introducing a prior-existence restriction: leave out of your utility calculations those individuals who would come into existence only as a result of the very actions you are contemplating. Putting it another way, consider only those individuals who already exist (this can be expanded to a prior-existence + "existence anyway" restriction, enabling us to count those who will exist irrespective of your own actions, e.g. the future general public).
However, this way of rejecting the total view also has problems.
Consider this thought experiment (adapted from Parfit). Imagine a government-funded medical program which tests thousands of women before they become pregnant. If they are found to have a particular illness, they can delay pregnancy until they are cured. If they get pregnant while they have the disease, the resulting children will suffer congenital impairment (but will still lead lives with a positive sum of utility).
In this thought experiment, there are no relevant individuals - i.e. the babies we're thinking about - already in existence when the government decides whether to fund the program. Nor are we talking about individuals who will come into existence anyway, if that matters. Which children will, or will not, be born is actually contingent on whether the program goes ahead.
Also, think of a specific woman who has been diagnosed as having the disease. When she decides whether to try to get pregnant now, or to postpone it until she is cured, there are no actual, already-existing, children involved. Which of her ova will get fertilised by which sperm cell, and which child will end up coming into existence will depend on what decision she makes. Yet, we probably want her to delay her pregnancy and have a healthy child. We take a moral view on the issue, but the prior-existence restriction seems to say we can't.
The prior-existence restriction, then, operates on the fact that there is no already-existing child (or fetus or embryo) whose interests the government, or the woman who is contemplating having a kid, can take into account. But surely we want governments to sponsor such programs if the funds are available, and we want women to take opportunities to bring healthy babies into the world rather than unhealthy ones, where there is a choice. The prior-existence restriction would seem to prevent us from making moral claims that a government, or an individual woman, is doing the morally right thing in such cases. There is no relevant baby, or even an embryo, in existence when the decisions are made.
But allow me to harp on the point: all other things being equal, funding a program like this seems to be a morally good thing to do. Or if a woman has actually been diagnosed with the hypothetical disease, it seems that it would be a morally good thing for her to postpone getting pregnant until she is treated for the disease.
Such examples convince some philosophers that we cannot use the prior-existence restriction, at least not always; some kind of total-view thinking does have a role in morality. Or so they think.
In his article, "Research into Aging: Should it be Guided by the Interests of Present Individuals, Future Individuals, or the Species?", in the anthology Life Span Extension (ed. Frederick C. Ludwig), Singer ultimately relies on total-view thinking.
The way he sets up the thought experiment, the individuals in a society with life extension will have longer lives and more total utility than comparable individuals in a society without life extension. However, there will have to be less of them to avoid Malthusian type problems. The population at a point in time will be the same, but the number of people (think of each individual person as a four-dimensional space-time worm) in future space-time will be fewer. Also, their individual average utility will be less over their long lives than those of shorter-lived people (shorter space-time worms) over their shorter lives.
Thus, we needn't worry too much about the longer-lived people as individuals (they will be happy throughout their longer lives, though less happy as they get older, and they will think of their longer lives as a benefit). Yet the total amount of utility existing in space-time will be less if the future society has, rather than if it does not have, life extension technology.
If we accept something like the total view, we should (on the scenario developed by Singer) relinquish life extension technology. Singer wants us to adopt total-view thinking, so he comes out against life extension technology (at least if it has the kind of social impact he imagines it will have) (see page 144 of the book).
He offers a thought experiment of his own to support total-view thinking. Suppose we are confronted with the prospect of either all agreeing not to have children, thus ending the human species, or drastically cutting back on consumption of resources (at some cost to the utility of people who already exist). What should we do? Singer thinks it is obvious that we should cut back on consumption of resources and continue having children.
For utiltarians, it might be impossible, in the end, to avoid total-view thinking, despite its unpalatable aspects. However, that gives me no problem because I am not a utilitarian. I think that the justification for our moral norms is, roughly, that they help protect us from things that we fear and to preserve things that we value. Among the things that we value are having healthy children, laying the foundations for the continued existence of our society, reducing the misery in the world, but also having long, healthy lives for ourselves. There is no objectively right answer to what weight these things should be given - as individuals we give them different somewhat different weights, and to the extent that policy and law support these various values the weighting involves an element of social compromise. Reality leaves morality somewhat underdetermined.
Nonetheless, most of us do value the things I mentioned. We have perfectly good reasons to support moral norms that enshrine them as values, and to campaign for laws that will protect these values (among others).
This analysis accepts that there is no inescapably and categorically "right" answer to questions of morality and policy, a thought which may go against what we are socialised to believe and against much traditional common sense about morality. But is it really very surprising to us? I think that the mild kind of moral scepticism implicit in what I am saying should become the new common sense.
On that basis, we can worry about relieving suffering in the world, rather than maximising the sum total of utility. This needn't drive us to the crazy idea of killing all living things that can suffer (the reductio ad absurdum of negative utilitarianism) because the relief of suffering is not the only thing we actually value - e.g. we actually value the existence of complex, creative, flourishing human societies. We also value wilderness, health, freedom, our own continuing lives, and many other things.
When we look at what we actually value, there is no need to adopt any paradoxical theory such as the total view. Think of it like this. The future society with life extension technology, as depicted in Singer's scenario, will not contain people whom we should feel sorry for. Nor need it be a society that lacks complexity or creativity, even it is smaller in its space-time population than the alternative society without life extension technology. The people who live in this society will be glad to do so, and glad of the enhanced lives that life extension technology will enable them to have.
In short, no important value should lead us to try to avert such a society - all we need to do is abandon total-view utilitarianism, which gives a crude and unhelpful picture of what actually underlies our moral thinking.