As Mill notes in Utilitarianism, often the best indicator of the utility to an individual of some outcome is whether or not the individual would choose that outcome. This is intuitively consistent: each person is the leading authority on her own likes and dislikes, so provided that a person is well-informed about the nature of the options, her choice will result in greater utility for her than the choice of an equally well-informed third party.
For most real-world moral questions, the best that one can do is to give others choice. Take as an example charitable giving. Suppose that you could give a poor person in the developing world either a mosquito net that costs $10 dollars, or the $10 dollars: which is likely to maximize the utility of the person you mean to help through your charity? If the mosquito net is the utility-maximizing use of that money, then the recipient of the $10 will turn around and purchase it – the outcome will be the same. If, on the other hand, utility would be greater through some other use of the $10, say through investment in a small business or a tin roof, then the utility would be greater if the recipient were given the choice of how the $10 is spent. If those are the only outcomes (more on this below), then the expected utility of the act of giving cash is higher that that of giving a mosquito net.
Further, consider that a person will always be better informed about his own utility. That information can’t be faithfully passed on. As a result, an ideal world will be one in which individuals are able to act on the maximum of information about their choices: the most informed decision possible about what will maximize a specific individual’s happiness must be one made by that individual, because she alone has access to both the subjective and the objective data to inform the decision.
Of course, this is just the maximum. Above, I intentionally excluded the case where the person making the decision was less informed about the the options, limiting the choice to between two optimal outcomes. Certainly there are cases where a benefactor can be better informed than a recipient about the options available, so even though the benefactor’s knowledge of the recipient’s preferences is imperfect, that imperfection is outweighed by the difference in knowledge. Consider for instance a person choosing between medicine and alcohol who has no knowledge of modern medicine or the modern science of disease. This person may well choose the alcohol, knowing its effects on her utility, even though if they were informed they would choose the medicine.
But note that this is not really a choice between medicine and alcohol. For the individual concerned, it is a choice between alcohol and an unknown. No preference for drunkness over health has been expressed by the choice. And it is clear that adding real choice, here by informing the individual about what medicine can do for them, would increase utility, because the individual could combine her subjective knowledge and the knowledge of the objective choice.
A similar point could be made about addiction and other failures of rationality: is it a real choice is someone does something by habit, or without reflection, or to avoid short term pain in sacrifice of long-term pleasure? To some extent, these aren’t choices at all, and again, turning them in to choices, improving rationality, improving perspective, improving awareness – if a greater level of real choice is given to the individual, the resulting utility is expected to increase.
Choice, then, must be an integral part in utilitarianism. Individuals have better knowledge about what will maximize their own utility; introducing real choice should be expected to increase utility; and choice is the only way to access individuals subjective evaluations. Indeed, because of the limits on our ability to know others’ subjective experiences, choice must be inherently valuable in any applied utilitarian moral system: even though the aim is still utility, that utility can only ever be objectively determined by choice.