Utility is the want behind value, the end that choice is meant to serve, and a quantity whose meaning changes with the work it is asked to do. The volume began with pleasure and passed through preference, risk, welfare, happiness and comparison. Each passage made a particular question more tractable; none made all the questions identical. A ranking can organize demand. A report can describe an experience. A judgment about well-being can ask whether the chosen and experienced life is going well. The last task is to determine what economics gains by keeping the name across these uses, and what it must disclose to prevent an answer in one register from silently settling another.
The pleasure account gives utility an object beyond the decision it explains. A person can choose something expected to be pleasant and then find it painful. If pleasure is the stated aim, that difference is evidence of a failed prediction. The account therefore permits a question that a bare record of choice leaves open: did the outcome deliver what was sought? Measuring the answer is difficult, because reports have scales, contexts and memories of their own. Yet difficulty in measurement differs from the absence of anything to investigate. Repeated reports of experience, their relation to circumstances, and their agreement or conflict with other evidence can make some claims about feeling better supported than others.
The preference account can do useful work with a thinner object. For a specified set of alternatives, it represents their ordering without requiring a common substance of satisfaction. Three scores can record that A comes before B and B before C even when the differences between the scores have no independent meaning. This economy of information is part of the achievement of the ordinal account. It lets a theory connect choices across budgets and menus without first constructing an instrument that reads pleasure. But what is gained in economy must be respected in the conclusion. The ranking gives an order under its assumptions; a claim about the intensity of satisfaction or the welfare of another person needs further information.
A stable ranking also exposes the model to failure when the same alternatives recur. Suppose, with circumstances fixed and each choice strict, a person chooses A from A and B, B from B and C, and C from C and A. A single utility ordering cannot represent that cycle: it would require A to score above B, B above C, and C above A. By comparison, choosing A over B and C, and B over C, is readily represented by scores three, two and one. The first pattern calls for a changed model or an explanation of what changed between the decisions. Calling every observed action utility maximization after assigning it a new ranking would remove precisely the stability that gave the account something to test.
The conditions of choice belong to the evidence from which a preference is inferred. Accepting a difficult job when the alternative is no income says something different from selecting it among several secure offers. A household that buys no treatment may lack money, information or access; the observation cannot identify a dislike of treatment by itself. These possibilities do not make behavior uninformative. They identify what else must be known before a particular inference is justified. The discipline of revealed choice is strongest when the menu and the constraint are part of the account. The freedom to choose among attractive alternatives is a further feature of the situation, one that the final purchase alone may conceal.
A coherent preference can express a purpose beyond the chooser's personal welfare. Sen distinguished sympathy, where another person's condition affects one's own well-being, from commitment, where a person can act for a reason even at a foreseen personal cost. Someone may keep a burdensome promise because it was made, rather than because keeping it supplies the greatest expected pleasure. Such a decision can be intelligible and consistent. Its consistency gives no warrant for redescribing the sacrifice as proof that personal welfare rose. Equally, the possibility of commitment establishes no presumption that ordinary consumption choices are always sacrifices. The point is to leave the person's reasons available for investigation instead of deciding their character through the definition of utility.
Well-being itself can require evidence beyond either desire or contentment. A person accustomed to severe deprivation may report modest expectations or learn to take satisfaction in what remains available. That adaptation is an important fact about the person's experience. It leaves a separate question about nourishment, health, education, mobility and the effective freedom to pursue a life. Sen's discussion of advantage makes the informational choice explicit: an account centered on mental states, one centered on resources, and one centered on what a person can actually do and be attend to different things. None can inherit the authority of the others merely by calling its result a utility score. The chosen account must explain why its information answers the evaluative question.
Resources also reach people through different circumstances. The same income can make different opportunities possible for people facing different needs or costs. An account of resources records what has been assigned; an account of capability asks what that assignment enables. That distinction does not make resources irrelevant or promise a single uncontested list of capabilities. It shows why a distributive argument can require information about conversion as well as possession. If a proposed improvement expands a person's effective opportunities without diminishing the features already valued, there may be reason to favor it even while harder comparisons remain unresolved. Evaluation can proceed by justified partial judgments without pretending that every life has already been placed on one complete scale.
The science retains discipline by stating which claim a utility function represents. If it represents a stable ordering, the relevant evidence includes choices across appropriately comparable situations. If it represents an assessment of experienced pleasure, the evidence includes reports and the assumptions that make their scale and aggregation meaningful. If it enters a social judgment, the account must identify how persons' advantages are compared and whose purposes or opportunities have been included. These inquiries can inform one another. They can also disagree for reasons that improve the analysis: a stable choice may rest on a mistaken forecast, a happy report may coexist with a deprivation, and a larger money total may leave a serious loss uncompensated.
The invisibility of utility therefore places a burden on interpretation rather than ending the inquiry. A quantity inferred from evidence can be useful when the evidence and the inference impose restrictions. The danger appears when the same number is first defended as a modest representation and then used as a complete measure of the good. Keeping the intended claim fixed allows the model to succeed at what it can show and to encounter its limits where further reasons are needed. The welfare question survives the success of the choice theory; the choice theory survives the discovery that welfare contains more than choice. Their connection is a subject for explanation.
The worth of a utility measure lies in the question it can answer.
Utility remains indispensable because value repeatedly asks how circumstances matter to someone. It can be felt, chosen, inferred and evaluated, with a different warrant for each claim. The volume's achievement is to make those warrants visible: an ordering, a scale, a report, a reason for counting one change as an improvement. Economics can keep a quantity it cannot directly see by making clear what observations support it and what conclusions those observations permit. What it cannot obtain from the name alone is the final meaning of a good life. That question travels onward into the accounts of wealth, risk and distribution, where every number must again answer for the human difference it is meant to record. The argument remains open.