CHAPTER 1: THE UTILITY FRAMEWORK

Progress in any intellectual endeavor only begins when the investigator decides which elements of a situation are germane and which are essentially irrelevant, and then groups the necessary elements in recognized categories within a framework out of which sense can be drawn. Logically, this intellectual triage should be prior to any actual analysis and conclusion drawing, although in practice categories and analysis evolve simultaneously.

The most fundamental social science question concerns the basis of human action, for once the key drivers of such action are understood, other elements of social science fall into place. Not surprisingly then, one of the intellectual terrains over which fierce arguments and counter-arguments rage is that concerning this question.

In this thesis I am going to examine one general framework1* that seeks to illuminate this issue. This framework, which has gained prominence in the social sciences and particularly in economics, begins with three basic postulates:

(I) An action (or decision) is best viewed as the outcome of an interaction between individual preferences and an external environment that limits attainment of these desires. The process is one in which the individual examines his preferences and the external environment, performs the relevant calculations, and carries out the action. This resulting action bears a relationship (which will be addressed in the following pages) to the ideal or optimal action under the circumstances. This is the Principle of Optimality.

(II) Individual preferences are such that the potential consequences of actions can be directly compared. The preference ranking is complete and, hence, heterogeneous outcomes can always be expressed in a single, commensurable currency, be it utils, money, prestige, or face. An equivalent way of expressing the same idea is to say that the relationship between competing preferences is fixed, and is not under the decision-maker’s control. This is the Principle of Commensurability.

( III ) The context in which an individual makes decisions is completely determined by giving a description of the external world and the decision-maker’s fixed preferences. In this scheme, the individual is exhaustively described by his preferences; beyond these preferences there are no other relevant data for his decision. This is the Principle of Sufficiency of Preferences [and Events].

This approach2 to individual decision making and action3 can be called the utility framework4 which, together with its underlying principles, forms the foundation for least three strains of intellectual thought:5

1. Microeconomic Theory. Though there are competing versions here, the core of microeconomics sees individual action as motivated by the desire to choose among contending inclinations to maximize some notional function, such as a utility function.

2. Ethical Utilitarianism. This second strain of thought seeks to base moral rules on the principle of maximizing an aggregate of individual happiness.

3. Decision Theory. Finally, this line of thought attempts to sharpen decision-making ability in complex situations. Insights regarding appropriate decision making under such circumstances are developed with the utility framework acting as a starting point for analysis.6 This is not to say that every investigator interested in each of these three subject areas agrees completely with the basic principles. Like any complex topic, one has the option to discard part of the foundation if one can find a plausible substitute. Richard Hare [26], for example, has developed an elegant defense of ethical utilitarianism based on the notion that only certain preferences are admissible. Similarly, George Ainslie [1] has been in the forefront of introducing evidence from experimental psychology in order to modify the Principle of Commensurability. In Ainslie*s schema, preferences are malleable and partially under control of the decision-maker. Of course, if one strays too far the resulting theory would no longer be in the tradition of the utility framework at all. In broad outline, however, I think it is fair to say that thinkers concerned with microeconomics, utilitarian ethics, or decision theory have generally relied upon a view of choice that accepts these three principles.

In the following sections, I review the link between these three bodies of thought and the utility framework. Microeconomic theory receives the greatest attention both because it is undoubtedly the more familiar to economists and because it is drawn upon most heavily to illustrate difficulties with the utility framework.

1.1 Microeconomic Theory The economics profession’s analysis of the basic decision unit, as embodied by microeconomic theory, relies exclusively upon the utility framework. The economist is searching for a positive description of action and the model of a fixed set of preferences confronting external constraints mirrors perfectly the need to understand the allocation of scarce resources to competing ends.

Before the relationship between the utility framework and microeconomic theory can be demonstrated, the model of individual decision making upon which microeconomics draws must be described. At least three competing versions of this framework, as applied in microeconomics, can be distinguished:      

(I) Strong Rationality -decision-makers deliberately choose the most preferred element of their feasible set, A decision-maker examines his utility function and prices (yielding the feasible set), and chooses the consumption bundle that maximizes utility. In the uncertain case, expected utility is maximized. Elster [18:1:2] captures the essence of this approach:

Rational choice theories assert that [the mechanism for decision] is the deliberate and intentional choice for the purpose of maximizing some objective function, be it a real one (like profit) or a purely notional one (like the utility function representing preferences).

This thesis has a strong intuitive appeal. The decision-maker, presumably capable of the reason which is our hallmark, examines the external situation, weighs the options carefully, and makes the decision that is most preferred. If the preferences can be represented by a single ordinal scale, which they can be under certain relatively mild technical conditions (providing, of course, that the three underlying principles are satisfied), then this decision can be thought of as choosing in order to maximize utility. To reject this thesis is to claim that an individual is somehow choosing poorly or not in accordance with his desires -something that is prima facie incompatible with the reason which we attribute to ourselves. Casting this utility theory version in the language of rationality, it could be said that individuals are assumed to behave rationally in the strongest sense: action is based upon beliefs and preferences that are “true,” An individual who fails to choose the maximal element fails to make use of this capacity for rationality.

Many economists, however, do not accept utility theory in this, the strongest, version. The difficulties of compulation have been admirably highlighted by Simon [80:99] in his attempts to steer inquiry on an alternative course:

Broadly stated, the task is to replace the global rationality of economic man with a kind of rational behavior that is compatible with the access to information and the computational capacities that are actually possessed by organisms, including man, in the types of environments in which such organisms exist. Simon’s concept of bounded rationality and other work on the issue of satisficing has pushed defenders of utility theory to interpretations with caveats incompatible with this strong version.

     (II) Revealed Preference -it is assumed that individuals, in making decisions, always choose the most preferred, or maximal element from their feasible set. In this way, preferences are revealed by actual choices. A counterfactual argument is invoked here: if the most preferred element were not chosen, the individual would have chosen differently. The choice here may or may not be intentional, but the result is the same. De facto, one maximizes utility (or, in an uncertain world, expected utility). This is the idea behind the theory of revealed preference. According to Samuelson [65]:

The individual guinea-pig by his market behavior, reveals his preference pattern -if there is such a consistent pattern.

The observed demand functions are rationalized by finding the utility functions that would give rise to these demand functions. As Varian [88:101] states the issue in his standard treatment of consumer behavior:

. . .[up to this point we took] preferences as the primitive concept and [derived] restrictions on demand functions that come from preference-maximizing behavior. An alternative approach is to take the demand functions as the primitive concept and derive preferences that are consistent with such functions.

At first blush, the idea here may seem completely vacuous: an individual chooses what he wants, otherwise he would have chosen differently. A little reflection, however, should result in the conviction that there is a substantial insight that stems from this observation: choices are made for some reason. An individual’s action is the result of an act of choosing and, presumably, this act of choosing could have been different had a different set of circumstances been obtained. Choices are, in this sense, justified by the presumption of rationality: “Of course that is what I desired to do. If I didn’t want to do it, I wouldn’t have.” Thus, the theory of revealed preference attributes capacities of volition and self-consciousness to the individual. An individual, when making a self-conscious choice, must be able to point to something that rationalizes the choice for him; the action must he intelligible from the vantage point of the decision-maker.

For these reasons, accepting the theory of revealed preference entails much more than is at first apparent. To accept that the theory of revealed preference tells us something interesting is also to accept a set of propositions regarding action and consciousness.7

The point here regarding volition is related, I think, to Sen’s argument concerning the completeness of individual preference orderings implicit in revealed preference theory. What gives revealed preference theory its bite is the assumption that people really do reveal their preferences via actual choices. This requires strong assumptions concerning the set of comparisons that an individual must undertake -comparisons with which Sen [76:66-7] is not entirely comfortable:

If a person chooses x when y was available, it would seem reasonable to argue that he did not really regard y to be better than x. There is, of course, the problem that a person’s choices may not be made after much thinking or after systematic comparisons of alternatives . . . Even some important decisions in life seem to be taken on the basis of incomplete thinking . . .

Revealed preference theory takes the completeness of these comparisons as a starting point, rather than investigating empirically whether completeness does, in fact, attain.      

(III) Weak Rationality -uncertainty and complexity preclude the individual from always being able to choose the most preferred consumption bundle, but some unspecified type of evolutionary mechanism helps to ensure that the maximum is constantly being approached. The idea here is a sort of “maximizing through satisficing” -the individual does as well as he can and continues to learn about his environment in order to do better next time. In cases where the individual continues to simply satisfice, this merely indicates that the cost of changing behavior (either search cost or some disutility caused by the breaking of routines) is greater than the expected gain.

This is an important extension to Version (I), Strong Rationality, brought about by the criticisms of those advocates of satisficing that focus on the impossible informational demands of the first interpretation of utility theory. The costs of gathering the requisite information and the inability of the human mind to process such information are taken into account, and the decision-maker is assumed to act sub-optimally -but sub-optimally in some sort of optimal manner. Thus, satisficing is just a form of maximizing behavior, where the costs of maximizing are included explicitly.

This is also an important insight into the decision-making process. When information is costly or impossible to obtain, or where one cannot compute the optimal action even given all the information, an estimate is formed or a guess made. Since presumably one can vary to some extent how much information is gathered, this also becomes a control variable and one cannot be doing as well as one can until the amount of information gathered is chosen as well.

The difficulty that this doctrine suffers from is its a priori character. It is intuitively plausible to say that some sort of unspecified evolutionary mechanism will produce decision rules that cope with the problems of information processing. The claim, however, that such an evolutionary mechanism will eventually bring about the optimal sub-optimal rules does not follow. Part of the difficulty stems from a misunderstanding as to precisely what evolution can and cannot do. Elster [17:6-7] is clear here in distinguishing the global maxima that non-evolutionary processes (i.e., computation) can obtain from the local maxima that evolutionary processes (i.e., natural selection) tend toward:

. . . [my] main contention . . . is that there is a basic difference between the local optimization through small improvements, and the global maximization that permits steps of any size . . .

Evolutionary processes (those which proceed in small steps) must always climb the utility gradient, and thus can only ensure that local optima are reached. Global, non-evolutionary maximization, however, can attain the global maximizing with strategies that permit utility to decline before increasing. Thus, local optimization suffers from the “you can’t get there from here” problem. If utility can be ultimately increased only by it experiencing a diminution for a period, then evolutionary processes are not powerful enough to do what proponents of the “satisficing as maximizing” argument claim. An example here might be the football coach trying to choose the most effective defensive system. If we assume that the coach can envisage only small deviations from the present defense, and not all configurations simultaneously, then there is no guarantee that the evolution of the defense will be optimal. Another example might be the design of a welfare system. If only piecemeal changes are permitted, as is effectively the case within the constraints of the American political system, experience seems to indicate that the result will be sub-optimal. Only a decision-maker who could survey a much wider range of policy options (a possibility that ex hypothesis bounded rationality has ruled out) could hope to choose the global maximum.8

The upshot of this discussion is that satisficing is, and will necessarily remain, satisficing until empirical research establishes the actual mechanism by which satisficing leads to optimal rules.9

Despite fundamental differences, all three of these approaches to microeconomics share the principles of the utility framework. Decisions are the outcome of individual preferences and external events, with these preferences being the only necessary information to completely describe the individual. This corresponds to the Principle of Sufficiency of Preferences (and Events).

Further, all possible outcomes are reducible to an index of satisfaction, implying the ultimate commensurability of competing satisfactions. The existence of a utility function ensures this common currency exists, since the units (utils, presumably) of the utility function are, ex hypothesis, comparable. This corresponds to the Principle of Commensurability.10

Finally, optimality plays a key role as the decision-maker is assumed to choose in a fashion that bears one of three relationships -described by Strong Rationality, Revealed Preference, or Weak Rationality -to the true optimal decision.

This section has demonstrated the link between the utility framework and microeconomic theory, noting that all variants of the latter must rely upon the principles given by the former. The link between ethical utilitarianism and the utility framework is reviewed in the next section.

1.2 Ethical Utilitarianism Ethical utilitarianism, particularly in its modern guise, is a complex doctrine with many variants. At the risk of oversimplifying these nuances,11 three key elements can be distilled:

(I) States of affairs are to be ethically ranked according to the criterion of satisfaction. Individual preferences12 are the data necessary to make this ranking.

(II) The relevant yardstick of the ethical desirability of an action is always consequential. Thus, when evaluating the morality of a particular decision, all information that cannot be expressed as a consequence can be discarded.

(III) All preferences must be reducible to a single scale or index, with the ultimate decision based upon maximizing this index. This scale is constructed not only from the preference of one individual, but also from the preferences of all individuals making up the group in question. In the end, decisions are to be rendered by comparing a single number expressing the desirability of an action with a single number expressing the desirability of an alternative action.

This view of morality corresponds precisely to the principles underlying the utility framework. The individual is important only insofar as he is a locus of preferences that can be satisfied. The decision-maker (be it the individual or some social decision mechanism) requires only this information, and can discard the remainder.13 Once individual preferences are recorded, no further facts about the individual are relevant.

In this way, the Principle of Sufficiency of Preferences is used to simplify the decision problem. By ruling that only certain sorts of information are legitimate,14 ethical utilitarianism seeks to make the ethical dilemma tractable. This discussion will be continued in Section 1.4.

The Principle of Commensurability operates here at the level of the society. The claim is not simply that individual preferences are reducible to a single ordinal scale, but that the preferences for the relevant group are reducible and thus available for ethical decision making. The claim that utility is commensurable across individuals is much stronger than the claim that individual utility functions exist.

Finally, the Principle of Optimality is implicit within the ethical dictate to take the action that maximizes overall welfare. All three principles, thus, are essential to underpin ethical utilitarianism.

In the next section, decision theory, the third strain of thought arising from the utility framework, is reviewed.

1.3 Decision Theory Decision theory15 is a hybrid of economics, statistics, and operations research that takes the model of decision making given in economics and applies sophisticated techniques to identify the most preferred action in complex circumstances.16 Unlike ethical utilitarianism, there is no moral component. The purpose of decision theory, then, is to improve decision-making ability by both heightening awareness of the factors to consider and providing concrete tools for decision analysis.17 Examples here include decision making under risk, bayesian analysis, loss functions, and linear programming. In many practical applications, decision theory has managed to deliver powerful tools for coping with complexities inherent in real-life situations.

It is worth adding that although decision theory has no moral component, lurking behind such exercises is the notion that an individual should choose his preferred element, and that choosing this element will be best overall for his life.18 Thus, while people may not intentionally maximize, paying more attention to individual utility will go a long way toward increasing overall individual happiness.

This tack should be distinguished from the philosophy underlying microeconomics. Decision theorists openly admit that individuals do not always maximize; sometimes gross errors are made in simple cases. This strain of thought is illustrated by the typical responses to the Allais and other similar paradoxes: “Certainly there are people out there who don’t choose well, but if we explained the axioms of choice to these individuals, we could convince them to make the rational choice and thereby they would be better off.” Kahneman, Slovic and Tversky [34] and the other authors contributing to this edited volume provide numerous examples of individuals who choose sub-optimally in very simple cases; no doubt, overall utility could be easily increased if just a little more thought were applied. Typical errors include overgeneralizing from small samples or an “anchoring” effect when individuals place too much weight on the status quo. Zeckhauser [101] gives the example of his Harvard faculty colleagues who have spent less than one hour thinking about retirement planning. Presumably, this is a decision that should warrant some consideration.

The thrust of this view, then, is to see the expected utility theory (and utility theory if there is no uncertainty but great complexity) as a tool people can use to increase utility.19 Decision theory does this by relying upon the same utility framework underpinning microeconomics, with all three principles of this framework applying to decision theory.

The Principle of Sufficiency of Preferences is satisfied because an individual’s preferences are accepted as the only data pertinent when considering the decision at hand.20 Were this principle not satisfied, the additional information beyond preferences would overwhelm decision theory’s capacity to provide concrete guidelines.

Without the Principle of Commensurability, no ultimate advice could be proffered since outcomes could not always be compared. More generally, a framework for decision making is only possible if a common currency is available to directly compare outcomes.

Finally, the Principle of Optimality stands behind all the techniques of decision theory. The ultimate objective is always to attain optimal decisions, and the value of a particular decision theory technique is least partially measured by how close its solutions come to the optimum.

Having briefly related these three strains of thought -microeconomics, ethical utilitarianism, and decision theory -to the principles underlying the utility framework, I now wish to focus on the question of why the utility framework is so appealing as a starting point for an analysis of decision making.

1.4 The Utility Framework as a Solution The three strains of thought arising from the utility paradigm -microeconomics, ethical utilitarianism, and decision theory -have gained wide currency among social scientists in general and economists in particular. Why is this so? What is it about the utility framework and its three principles (optimality, commensurability, and sufficiency of preferences) that social scientists find so appealing?

I don’t think the answer is at all difficult to locate -though it seems to have been eclipsed in recent writings. Baldly stated, this framework -in regulating the conflicting and chaotic claims on an individual -offers a solution to the vexing problem of how to single out a particular action from a multitude of possibilities. Schelling’s [70] term “focal-point” captures the nub here: once we admit the existence of the entity utility underlying this framework, special status is conferred on the decision that maximizes utility. The utility framework has thus managed to bring order out of the chaos of decision making which self-conscious beings confront.

One of the clearest explications of the issues here is given by Nussbaum [55], using Greek philosophy as a backdrop. In Plato’s The Protagoras [57], Socrates and the famed sophist Protagoras debate the basis of a “techne of practical reasoning” [55:57]. Socrates key point is that an ethical “science of measurement” is required to avoid the dilemmas that would otherwise accompany multiple choices with no clear basis for decisions. Nussbaum [55:55] captures the essence of the difficulty here:

If ethical values are all commensurable, differing from one another only in quantity, what difference does this make? Plato gives us a stark and simple answer. The adoption of an “ethical science of measurement,” at the heart of which is the belief in commensurability, is both necessary and sufficient for “saving our lives,” i.e., for giving human beings a life that will be free of certain intolerable pains and confusions.

According to Plato (through Socrates), an ethical science of measurement, which is able to place all competing ends on a single scale, is essential to escape from the agonizing choices incommensurability poses.21

Using The Protagaras as a starting point, Nussbaum [55:117-9] weaves an exceptional thought experiment that captures the power of this ethical science of measurement:

In spite of the gifts of Zeus, [human beings] lacked precise control over their choices and actions . . . when they were confused about justice, nobility, or goodness, they had no art to arbitrate their quarrel . . . Even when they could agree, the primitive confusion of their value system often gave rise to serious conflicts: piety could seem to dictate one course of action, courage or love another . . . Worse still, even when they did have settled beliefs about the right course of action . . . they could still . . .be swayed by the lure of other nearer goods that were the object of some passion,

Nussbaum has identified three sorts of ethical crises the Greeks faced: (i) they could not agree on the goals that were ethically desirable; (ii) even when they agreed, they could not determine the relative intensity of the claims of different sorts of desirable goals (the precise problem of commensurability);22 and finally, (iii) even when the relative weights of certain desirable goals had been decided, there was always the possibility that weakness of will could prevent taking the proper action.

Faced with these imposing difficulties, Plato [57:357A] argues that our salvation must lie in bringing the precision of measurement to the ethical sciences:

What would assure us a good life then? Surely knowledge, and specifically a science of measurement, since the required skill lies in the estimation of excess and defect; or to be more precise, arithmetic, since it deals with odd and even numbers.

In Nussbaum’s [55:118] parable, then, this salvation, like many good things in Greek myth, comes from the gods:

Now Apollo, god of sunlight and rational order, god of numbering and of firm boundaries between one thing and another . . looked down at [humans’] plight. He did not wish the entire race to perish. Through his resourceful messenger Socrates, he revealed to them his marvelous gift: the art or science of deliberative measurement, together with the single metric or good, which is pleasure . . . The significance of this new orderliness for all their ways of life was immense . . . Every circumstance by which the condition of an individual can be influenced being remarked and inventoried, nothing was left to chance, caprice, or unguided discretion.

This deep desire of the Greeks23 to be spared the intolerable pains and confusions of ambiguity24 has crystallized into the principles of the utility framework.

Nussbaum [55, 56] illustrates the power of the utility framework with her vastly simplified but highly instructive, stylized bagel world. Imagine that all one cared about were bagels with cream cheese -and the more bagels, the better. If one is presented with a choice between two identically prepared bagels, the utility framework and the assumptions of the situation imply that the bagel lover will be completely indifferent between the two alternatives. After all, one cares only about bagels with cream cheese, and here are two such identical bagels. Now, imagine that the choice is between one bagel (with cream cheese) or two bagels (with cream cheese). Again, the preferred choice is determined. The two-bagel alternative, in a world where only bagels matter, must be better than the one-bagel alternative. Clearly then, at least in this bagel world, we can answer questions concerning how bagel lovers will choose, the ethically correct decisions for bagel lovers, and how bagel lovers should choose actions to maximize the consumption of bagels.

The world, unfortunately, presents us with more than bagels. Individuals can choose more leisure and less money, more travel and less food, more sex and less sleep, or perhaps more money and no honor. The possibilities for combining the various dimensions of life are limitless and the decision-maker’s dilemma is precisely to arrive at a decision when faced with this plethora of possibilities. If the real world problem could somehow be converted into the much simpler bagel world problem, the dilemma of decision making would disappear. It is precisely this conversion that the utility framework guarantees.

The Principle of Sufficiency of Preferences ensures that the decision-maker confronts the external world and individual preferences as fixed, with the individual decision-maker serving merely as a locus of preferences. As Sen and Williams [77:4] put the issue, in referring specifically to ethical utilitarianism:

Essentially, utilitarianism sees persons as locations of their respective utilities -as the sites at which such activities as desiring and having pleasure and pain take place. Once note has been taken of the person’s utility, utilitarianism has no further direct interest in any information about him . . . Persons do not count as individuals in this any more than individual petrol tanks do in the analysis of the national consumption of petrol.

To such preferences, completely exhausting the description of the individual, external reality (presumably fixed and independent of the individual decision-maker) is added to round out the description of the decision-making environment. The Principle of Commensurability is also required to complete the utility framework. This principle ensures that the decision-maker can convert all outcomes into the common currency of utils, or dollars, or playing cards, because the relationship between competing preferences is fixed. Once this conversion has been made, the remaining task of applying the Principle of Optimality becomes trivial: the decision-maker or social scientist counts up the options and determines which one yields a bigger number to learn what he:

(I) Will do, if the decision is considered from the viewpoint of microeconomics;

(II) Should do, under ethical utilitarianism (care must be taken to ensure that utility is properly summed);

(III) Should be advised to do in order to be as happy as possible, if the goal is to improve decision making with insights from decision theory.

Once all possible outcomes can be reduced to a single number -and it is self-evident that we prefer more to less according to this index – then decision making presents, in principle, no insoluble dilemmas, Rawls [61:555-6] points to the advantages of the determinateness of utilitarianism in his discussion of hedonism:

[Under hedonism, the] rational agent knows exactly how to proceed in determining his good: he is to ascertain which of the plans open to him promises the greatest net balance of pleasure over pain. This plan defines his rational choice, the best way to order his competing aims. The counting principles now apply trivially, since all good things are homogeneous and therefore comparable as means to the one end of pleasure. Of course, these assessments are plagued by uncertainties and lack of information . . .

This counting principle is simply the observation that if we have a single measure (utils or bagels, for example) and agree that we wish to maximize it, then the problem of choice disappears as soon as it is stated. The answer immediately follows from the single scale. Taylor [84:129], too, recognizes the powerful pull of a doctrine that solves the decision-making dilemma in his discussion of ethical utilitarianism:

A utilitarian ethic seemed to be able to fit the canons of rational validation as these were understood in the intellectual culture . . .of the [rationalistic] seventeenth century . . .

Moreover, using the utility framework, it is possible to:

. . .[validate] an ethical position by hard evidence. You count the consequences for human happiness of one or another course, and you go with the one with the highest favourable total.

The advantages of such a neat and tidy rule are immediately apparent if the utility framework is contrasted with other candidates for explicating or moralizing about decision making. Consider first an ethic based on Kant’s categorical imperative, one version of which runs [36:91]:

Act in such a way that you always treat humanity, whether in your own person or in the person of any other, never simply as a means, but always at the same time as an end.

On the surface, this appears to be an admirable and defensible moral position. Others should always be treated as beings worthy of some sort of respect. In the process of filling in the content of this maxim, however, insoluble practical difficulties arise. Does ordering a meal in a restaurant violate this rule by using the waiter as a means to food? Are car drivers not simply using autoworkers to get around? Are students using their teachers to increase their knowledge? Indeed, despite Kant’s enormous influence on the theories of morals and law in contemporary Western society, all formulations of the categorical imperative suffer from this intractability. We must only obey rules that can be universalized. This may mean we should generally tell the truth, but does this imply that we should tell the truth at all times, or only when it does not hurt others, or only when it is expedient for us? Unfortunately, Kantian ethics simply do not always help to answer the pressing practical questions we face.

Now, consider an ethic based upon the communitarian25 view of reaching a moral position through a self-conscious interpretation of underlying community values. This procedure certainly holds a great surface appeal, for the particular context of a particular community should undoubtedly be considered when seeking moral guidance. What happens, however, if no relevant (or if competing) principles are forthcoming? Once the community has gone through the exercise of self-consciously searching for the appropriate extant values, there is nothing else to do. This risk of under-or over-determinacy can never be entirely eliminated. And the same difficulties arise with alternative social science explanations of observed action, whether these explanations are couched in Weberian, Marxian, Freudian, or other terms.

Put starkly, alternatives to the utility framework all seem so damnably unfocused -there is enormous attraction to an approach that can yield decisions and use hard facts (such as fixed preferences and external events).

It is also easy to see why the concepts of utility and rationality have become so closely intertwined within economics.26 The existence of a single scale of measurement provides the rational basis upon which a decision can be made. Once the scale has been accepted, and the commensurability of all outcomes assured, potential actions receive a rational justification in light of the expected utility. In the absence of a single scale, grounding a decision becomes an impossible task.

The same considerations leading to the belief that ethical utilitarianism solves the decision-maker’s ethical dilemmas also lead to the view of microeconomics as an explanation of action. If a certain action has been chosen because it maximizes utility, there is now an explanation -and one that the social scientist can grab hold of and examine.

This crucial aspect of the utility framework has deep roots in Western thought.27 In addition to the debate between Plato and Aristotle (illuminated by Nussbaum [55,56], Bentham,28 and Sidgwick29 recognized the inherent appeal of an approach that could solve the decision-maker’s dilemma.30 Edgeworth31 extended this early work, focused on ethical utilitarianism, to the positive domains where it was developed by economists and decision theorists. Among contemporary thinkers, Becker [4] has been most responsible for adding to this particular line of thought, and I will return to his work throughout this thesis.32

Finally, modern cost/benefit or welfare analysis crystallizes all three branches of the utility framework:33 principles of microeconomics and decision theory are used to decide social issues within a framework of ethical utilitarianism. Cost/benefit analysis reflects two deeply felt considerations: (i) individual preferences have a standing that must be considered; and (ii) limited resources are not to be squandered. The bite of cost/benefit analysis arises from both the Principle of Commensurability and the Principle of Sufficiency of Preferences. Commensurability is essential, because disparate types of benefits must be directly compared to arrive at a decision. Sufficiency of Preferences is essential, since both preferences and the external world must be items of datum that the decision-maker can gather up and take back to his computer. Individuals count only insofar as they are depositaries of these detachable preferences. The Principle of Optimality expresses itself as the decision rule that the decision-maker choose the alternative that yields the greatest net benefit.

Without the utility framework underlying the whole structure, cost/benefit analysis would not be able to get started: either preferences (or combinations of preferences) would not be the stable entities one needs to make sensible decisions or such preferences would not yield a unique solution. In both cases, the key advantage of cost/benefit analysis -the ability to render a decision acknowledged as having a sound basis in individual preferences and a conception of social justice -would evaporate.

This section has analyzed the reasons behind the utility framework’s immense popularity among economists and other social scientists. The capacity to bring order out of the conflicting claims of decision making is an enormous advantage in both social science description and ethical or non-ethical prescription. In my conclusion to this chapter, I outline the remainder of this thesis.

1.5 Conclusion In this chapter, I have sketched the principles underlying the utility framework and demonstrated the link between this framework and three separate (though related) lines of thought. In the remainder of this thesis, I will closely examine the principles underlying this framework. My key proposition is that inconsistencies in these principles circumscribe the usefulness of the utility framework and the strains of thought deriving from it.

Chapter 2 addresses the role of the Principle of Optimality. Starting from the relatively simple observation that a conceptual framework is always necessary for decision making in the world, I argue that most of the important and interesting elements of decision making are stripped away by the utility framework. This framework must take for granted those prior decisions that set the stage for later decisions or actions. In this way, the utility framework, in its use of the Principle of Optimality, cuts into the decision problem half way, and explains much less about ethics or actions than proponents of this approach claim.

Chapter 3 addresses the role of the principles of Commensurability and Sufficiency of Preferences in the utility framework. I argue that our capacity for self-reflection implies the ability to examine and, perhaps, alter our desires (Frankfurt’s [21] concept of “second-order desires” is useful in pinpointing this dilemma). And I shall show how the neat solution guaranteed by the utility framework disintegrates once the implications of this capacity for reflection are considered.

Chapter 4 examines more closely the Principle of Sufficiency of Preferences and addresses two issues:

(I) The role that intentions play in the utility framework. Implicit in this framework is the notion that decision-makers consciously strive after preferred outcomes, or at least would increase utility by striving. A little thought, however, should convince one that there are many desired outcomes that cannot be attained by direct pursuit. To seek spontaneity, for example, is a project doomed to failure and Chapter 4 examines this phenomenon in more detail.

(II) Our imperfect ability to choose between various preference rankings -a capability discussed at length in Chapter 3. There is inevitably a tension between our attempts to will what we desire and our inability to always do so successfully. The considerations raised in this chapter suggest that the emphasis of the utility framework on the examination of fixed preferences and the external world may diminish the very satisfaction that is to be maximized. In Chapter 4, then, the object of the critique is the framework itself; if pursuing utility maximization and actually maximizing utility bear only a minimal relationship, then all bets must be off.

In the fifth and final chapter, I gather together the strands of the previous three chapters to suggest a positive theory of individual decision making. By this, I do not exactly mean an ethical theory -although moral considerations will certainly play a role -but rather an approach to individual decision making that will address some of those identified issues that have not been attended to by the utility framework. I promise no solutions, and the raising of issues in this chapter is as salient as the answers I will be able to provide.

I must also state that I am not attempting to assimilate microeconomics, ethical utilitarianism, and decision theory into a single body of thought. Though all three rely upon the premises of the utility framework, the use to which they are put varies widely. Indeed, were this not the case, they would all be the same subject. As I raise problems with the utility framework, I will also comment on the applicability of a particular criticism to each of the derivatives of this framework. For example, we will find that the considerations raised in Chapter 3 have a great deal to do with ethical utilitarianism and microeconomics, and not much to do with decision theory.

In addition to this distinction among microeconomics, ethical utilitarianism, and decision theory, a second contrast will appear throughout the thesis: that between the utility framework and the concept of profit maximization. Recall Elster’s remark in Section 1.1 that rationality can be defined by either the maximization of a notional function, like utility, or a real function, like profit, suggesting that profit and utility are parallel. This, however, is not the case; the principles underlying profit maximization do not present the same conceptual difficulties as utility maximization. We will find, for example, that the issues raised in Chapter 2 apply to both the utility framework and profit maximization, while those raised in Chapter 3 concern only the utility framework.

Before turning to Chapter 2, I wish to re-emphasize the enormous value of the utility framework. More than perhaps any other approach to human choice, it has provided a usable benchmark for thinking about decision making. It scrupulously avoids the creation of useless categories or ad hoc explanations of behavior that have proved all too easy to construct when decision-making is examined from the viewpoint of other intellectual traditions. It posits a certain type of rationality -the ability to do a reasonable job of choosing the means to a given end -that seems to be borne out with many types of individual choices. It possesses a surprising elegance and simplicity and rests on only a few principles. The goal of this thesis, then, is not to discard the utility framework. Rather, it is to draw a number of finer distinctions regarding the decision-making reality in which many people find themselves -distinctions that can be obscured within the utility framework.

ENDNOTES TO CHAPTER 1

1. In his widely acclaimed work, The structure of scientific revolution [39], Thomas Kuhn introduced the term paradigm to describe an integrated conceptual framework generally used within a scientific discipline. Paradigms provide a recognized procedure for answering questions that fall squarely within the accepted limits of a particular intellectual approach. The term is now, however, generally overused; as a result, I have chosen the less baggage-ridden term framework to describe the integrated intellectual approach that has gained prominence in economics and other areas of social science.

2. For those readers who are skeptical about my claim that the utility framework is trying to become the dominant theory of human behavior, Gary Becker [4:6] provides the most direct argument:

I am saying that the economic approach provides a valuable framework for understanding all human behavior . . . If [my] argument] is correct, the economic approach provides a unified framework for understanding behavior that has long been sought by and eluded Bentham, Comte, Marx, and others.

Only time will tell whether Becker will rank among the intellectual giants in Western thought.

3. The two are never clearly separated in this framework, although analytic philosophy has often explored this distinction. See Ryle [64:Chapter III] for an interesting overview of the issues.

4. Perhaps not unlike other investigators, adherents to the utility framework sometimes lose sight of the fact that other ways of approaching human behavior exist. Four quick (and dated)examples will demonstrate the variety of very different starting points that one can adopt for examining human action:

     (I) Weberian analysis -action is most fruitfully considered through an interpretation of the action’s meaningfulness for the individual concerned. An explanation of an action consists of reconstructing the context of the decision from the decider’s viewpoint by determining what meanings the individual has given to the action. Thus, the emphasis is on the endowment of meaning by the individual concerned. This is radically different from the utility framework’s requirement that preferences are essentially given for the individual.

(II) Marxist analysis -decisions are considered as the outcome of an individual’s class position resulting from his relation to the means of production in society. Knowing an individual’s relation to his means of production should, if the (classical) Marxist view is correct, tell us what individual actions to expect given the circumstances. Thus, action is definitely not the outcome of an interaction between the external world and individual preferences.

(III) Structuralist (or functionalist) analysis – an individual’s decisions arise from one’s particular role in the social system. One’s “conceptual map” of the world is built up from data provided by one’s role. Action, in this model, can be linked to the particular position an individual occupies within the society. This line of thought also leads to the view that getting the structure of social institutions right is the key factor in improving society. Poverty, for example, can be eliminated only when the structural aspects of ghetto life are eradicated. Structuralist analysis is most closely associated with Durkheim.

(IV) Freudian analysis -the starting point here is the psycho-sexual development of the individual and the ability to cope with the traumas produced by frustration with the external environment. A Freudian analysis of actions consists of tracing behavior back to previous life incidents producing trauma. Actions are not then the outcome of limits placed by the external world; rather, the limits arise from an individual’s particular complexes or sublimations.

The interesting aspect of this list is not the content but the fact that such widely varying interpretations of human action can coexist, however uneasily.

I should also add that as an approach to individual behavior, the utility framework, far from having a monopoly on explanation (as some members of the economics seem prone to believe), offers us a relatively impoverished view of individual decision making in contrast to other paradigms. Consider an explanation of individual choice couched in Freudian terms. Such an explanation would trace certain patterns of behavior to earlier traumatic events or show how sublimated drives manifest themselves in behavior. Even if we disagree on whether such an analysis is a fruitful explanation, at least the language used is richly descriptive. Or consider an explanation in Weberian terms, which would start by understanding the meaning of the action from the viewpoint of the individual. Other approaches to behavior, then, seek to uncover the “why” of actions. The utility framework, if offering an explanation of action at all, allows us much less insight.

5. It’s worth adding that a framework such as this also brings with it a unique language. Were our starting point a Marxist analysis, terms such as ’’class,” ’’relations of production,” and ”superstructure” would be essential to our thinking for the concepts used by Marxists (where I use the term in its social science, not political, sense) are only captured with the appropriate Marxist language. Similarly, Freud’s “superego,” “regression,” and “complex” are equally applicable only to those analyses of behavior that begin with a Freudian perspective. The utility framework, not surprisingly, also shares this feature: “choice,” “preferences,” “utility,” and “maximization” are the only legitimate currency in the world of the utility framework.

6. The reader may have noted by this point that the utility framework shares many of the features of a view of individual decision making often labelled the rational choice model. I am going to avoid using this term.

“Rational,” using Connolly’s [11] descriptor, is an essentially contested concept. Applying the term “rational” to behavior, or to a particular approach to understanding behavior, carries with it some measure of approval in much the same way that applying the term “democracy” to a nation’s political system does. It would be premature, therefore, to use “rational” to describe the utility framework.

Indeed, it doesn’t require much acquaintance with moral philosophy to realize that no consensus on the usage of “rational” has emerged. Consider the enormous gulf between Kant’s view that rationality requires one to not necessarily do what one desires and that of the rational choice model that requires one to do what one desires. A greater contrast in usage is difficult to imagine. How to apply the term “rational” to behavior is something about which we must debate; much of this thesis argues that the approach to decision making implied by the utility framework may not be entirely rational.

7. It is interesting to note that though economists are the social scientists who have placed the most emphasis on the act of choice, revealed preference theory strips away much of the possible leverage of choice theory. As rationality is always imputed (indeed, a convoluted demand function to justify any actions can always be constructed), the economist has no foothold for claiming any action is not rational. For this reason, the theory of revealed preference suffers from precisely the opposite problem than the stronger version of utility theory, and rationality has now become an exceedingly weak notion. To see this, let us ignore preferences for the moment and only consider probabilities regarding the expected states of the world of the individual decision-maker. Any observed decision can be rationalized through the construction of an appropriate utility function and subjective probabilities. Here are two examples:

In 19th century southern Africa, the Xhosa slaughtered their cattle and burned their fields in the expectation of a new Messiah. The preferences -greater material prosperity and the withdrawal of the British -were indisputably reasonable, while the subjective probabilities as to what was going to happen in the world were tragically lethal.

Witch trials, depending upon one’s view of the existence of witches, could be another example of rationalizing decisions based on dubious subjective probabilities. If the world contained witches, their expulsion might be legitimate. Nonetheless, one might deem that the subjective probabilities needed to legitimize witches’ existence are irrational. Revealed preference theory is not sufficiently powerful to be able to say much about such irrational subjective probabilities. I consider this a serious drawback. 8. Elster’s arguments are in response to the following example from Milton Friedman [22:35] of “must be” thinking characteristic of the Chicago School:

. . . unless the behavior of businessmen in some way or other approximated behavior consistent with the maximization of returns, it seems unlikely that they would remain in business for long.

To put forth this conjecture without a description of the evolutionary process that is to bring about the required response is simply so much scientific balderdash. Winter [99:96-9] also criticizes Friedman’s assertions.

Curiously, Elster uses this insight for the wrong purposes, arguing that “in creating man natural selection has transcended itself” and that man is, indeed, “a globally maximizing machine” [17:16]. His insight on the limits to evolution are correct, but he fails to recognize that bounded rationality ensures that these limits also apply to human behavior in many complex situations.

9. Indeed, demonstrating this is a project of Nelson and Winter [48], Nelson and Winter note that the claim of “satisficing as maximizing” requires the optimal amount of information be gathered and processed. This leads to an impasse, however, since examining the issue from the viewpoint of firm behavior (quoted in [17:59]):

. . . the choice of a profit-maximizing information structure itself requires information, and it is not apparent how the aspiring profit maximizer acquires this information, or what guarantees he does not pay an excessive price for it.

We will address this question in much greater depth in Chapter 2. For the moment, we should simply note that the arguments of those who defend satisficing are powerful, and those who wish to defend any version of maximizing behavior, even “satisficing as maximizing,” face a thorny task.

10. There is also the view that microeconomics is not searching after a description of the decision-making process at all. There can be no general presumption that individuals do choose the maximal element from the feasible set. Rather, our interest as economists lies in general economic laws. For these purposes, the assumptions of utility maximization provide tractable results and useful predictions. Moreover, in situations in which the assumption of utility maximization is not useful, we are not bound to stick to it. This is exactly the standard position of Friedman [22:15]:

. . . the relevant question to ask about the ‘’assumptions” of a theory is not whether they are descriptively ’’realistic,” for they never are, but whether they are sufficiently good approximations for the purpose in hand. And this question can be answered only by seeing whether the theory works, which means that it yields sufficiently accurate predictions.

Whether utility theory is an accurate descriptive model of choice is beside the point; if it is up to the tasks we assign, then it is the appropriate tool.

This is, in many ways, a sensible position for those in the discipline who are concerned with the immediate application of economics to real-world problems. The a priori constraints that I mentioned earlier are irrelevant, and it is only the ex post performance of the utility approach that determines to what use utility theory will be put.

This is not to say, however, that all economists agree that satisfying these ex post constraints is sufficient. Again, it is the iconoclast Winter [99:93] who has something to say here:

It is really a simple matter of the unacceptability of procedures that yield the right answers for the wrong reasons -when it is known that the reasons are wrong, and even the general character of the right reasons is also known. While there may be practical men who will stoutly contend that it is only the answers that matter . . . it is hard to think of another area of scientific or scholarly activity where this sort of bottom-line pragmatism holds sway . . . the general principle that scientific inquiry ought to pursue the “right reasons” speaks clearly and decisively.

A further objection that might be raised is whether we can avoid the issue completely simply by directly invoking the end result of the theoretical analysis that begins with utility theory. If we are only interested in predictive power, why do we need utility theory and some mild technical constraints to derive downward-sloping demand curves? The world tells us this already. The key point is that if we are going to take this pragmatic approach and evaluate concepts solely in practical terms, then utility theory in any form may be superfluous. Although this issue deserves further investigation, it will not be addressed here.

11. For example, implementing ethical utilitarianism as a guide to action immediately raises numerous practical issues: How large is the relevant group? A town? A nation? A state? The world? How will future generations be treated? Do we wish to maximize average or total happiness? Is it legitimate, as Hare [26] suggests, to rule that certain preferences are inadmissible? Adding to these issues is the old debate concerning act and rule utilitarianism: should we work with general rules expressing general principles or should each situation be examined individually? A seemingly simple question (most philosophers lean to rule utilitarianism), the distinction appears to break down as we push hard on the issue. Parfit [56] is a readable introduction to many of these issues.

12. Modern utilitarianism has attempted to divorce itself from makeshift psychological theories regarding pleasure and pain. Rather than defining utility by referring to the pleasures and pains arising from an action, as Bentham [6] did, the modern utilitarian points to the choices themselves as the observable manifestation of preferences. The goal is to avoid a commitment to any particular psychological theory. This leads to some confusion, as Sen and Williams [77] point out, since the term “preference” can refer either to desires that are not acted upon (I bought the red car, though I really preferred the yellow), or to the actual choices that are then imputed to represent preferences.

13. Sen and Williams [77:5] point to autonomy and integrity, for example, as two possible ethical considerations that can have no place in a utilitarian ethic.

14. Rawls [63] discusses the types of Information that are necessarily omitted under ethical utilitarianism.

15. Raiffa [60] is among the best general treatments of this topic.

16. A fourth area of enquiry with which the utility framework has affinity is game theory -the extension of the decision making of microeconomics to situations where individuals must consider the response of other rational agents. The unique contribution of game theory, as Elster [18:13] puts it, comes from the insights it lends in those situations in which “the action of each depends upon the action of all, by strategic reasoning.” The prisoners dilemma and chicken are well-known examples of game-theoretic situations.

Despite the close affinity between game theory (which has been developed largely by economists and mathematicians) and the utility framework, the original promise here has not been delivered. Consider a two-person game of chicken. In this game, the payoffs are such that a player is always better off if he takes an action different from his partner/opponent, though he is best off if he can force his partner to swerve. The notion here is one of two individuals driving at high speed towards each other.

If we are to apply the utility framework to predict or understand the outcome of a play of this game, we begin by specifying the preferences and the external situations faced by our two individuals. Specifying the preferences is simple -they are given by the parameters of the game. Specifying the external situation, however, creates a problem because of the game’s peculiar nature. For Player 1, the external situation is essentially what Player 2 is thinking. That in turn depends upon what Player 2 thinks Player 1 is thinking, ad infinitum. Strictly speaking, of course, what Player 2 thinks Player 1 is thinking may have nothing to do with what Player 1 actually thinks; nevertheless, there seems to be an erosion of the Principle of Sufficiency of Preferences (and External Events) as Player 1 realizes that his way of thinking has somehow become data for the game.

The end result is a lack of determinateness that defeats the major attractiveness of the utility framework. In Section 1.4, I discuss at some length the advantages of the utility framework that arise from its ability to always provide an answer.

17. It’s worthwhile at this point to distinguish among three sorts of “should.” The first type arises when we have a particular goal, or at least some notion of a goal, and we wish to attain this goal as painlessly as possible. If empirical research shows that individuals systematically ignore the possibility of improbable events, then we ’’should” take this into account. This corresponds to Kant’s hypothetical imperative: If you wish to obtain goal A, you should ensure that action X is taken. The “should,” however, is in no sense moral.

A second “should” arises from ethical claims. If we accept some ethical command, then we are accepting a command of the sort “Thou shalt do X” unconditionally. It Is more difficult to see how research (in its broadest sense) can help here. Certainly, research can uncover insights that are useful to our moral thinking. In the end, however, such insights can only be supplements to moral predispositions we already have.

The third “should” is one that has been neglected but it is central to the work of this thesis: the “should” that arises from the general query: What actions should I take to lead the best possible life? This isn’t quite the first type of “should,” since we have also thrown open the question of the goals themselves. No longer are we merely asking what are the appropriate means to a given end. Nor is this the “should” of moral imperatives, since how one should live one’s life isn’t necessarily always a moral question. This is because much of our activity is really non-moral in character (Do I prefer to play golf or tennis? Should I live in New York, London, or Nepal?). An illustration might make this third “should” clearer. Consider the issue of regret. Regret arises when the decision-maker suffers a loss of well-being because a good ex ante decision turned out relatively poorly ex post. A positive theory of regret -one describing the incidence, intensity, or the type of decision involved – is clearly a contribution to the theory of individual decision making. Viewed as a normative issue, there are two possible responses to the existence of regret.

The first is simply to acknowledge that regret will occur, and to account for it in advance. Thus, if I know that I will be unhappy if my equity investments drop (even though they were a good purchase at the time), then perhaps I should buy a fixed income security instead -though, again, if I then grow unhappy with my fixed income security when the stock soars, I cannot win. This response is to see regret somehow as a normal part of the utility function (though this introduces some technical difficulties -usually tastes and feasible sets are separate).

A second and more sophisticated response is that I should endeavor to eliminate regret from my preferences -that I can increase my overall well-being if I somehow stop being regretful. This response, however, directly attacks one of the key principles of the utility framework -the Principle of Sufficiency of Preferences. Preferences are now open to revision by the decision-maker himself. Chapter 3 examines this possibility in some depth. Here we should note that the payoff from controlling preferences is enormous. Among preferences that should be controlled, envy, regret, spite, counter-adaptiveness, and perhaps conformity come to mind.

18. There is much to recommend this view. Undoubtedly, all of us suffer a diminution in utility from failure to apply sufficient thought to our situation. The potential gains from a little extra effort in our lives’ decisions are enormous. Thus, the goal of decision theorists to help us improve our choices is admirable.

There is, however, a nagging doubt here. Despite enormous efforts to increase the attention paid to the dictates of rationality, surprisingly little tangible use is made of decision theory. Legions of business school students learn the rudiments of decision trees and subjective probabilities and promptly forget these tools on graduation. Perhaps the decision theorists are right, and those who ignore their teachings ultimately lose out. This last notion, however, is at odds with another idea that strongly appeals to economists: business generally does as well as possible, and easily corrected mistakes are quickly rectified. If decision theory is as helpful as its proponents claim, then we should see more use of it. As Bell, Bradley, and Haimo [5:2] state the problem:

Management science techniques should be a panacea for strategic planners. What better way to deal with the complexities of the modern world[?] . . . yet it is apparent that despite a few successes, strategic planners prefer the more qualitative insights.

For whatever reasons (and Chapter 2 will explore one possible explanation), decision theory has not delivered its prima facie potential.

19. Incidentally, it is worth noting that almost all that has been said here concerns the choosing of subjective probabilities and not the choosing of preferences.

20. This is not entirely necessary, however. One can imagine an attempt to increase personal welfare by discouraging the adoption of counterproductive preferences. For example, the common emotion of regret can be viewed as always causing a diminution in utility, A role for decision theory, then, might be to highlight the losses in utility such preferences generate to encourage an individual to alter his preferences.

One difficulty with this tactic is that the line is blurred between an ethical and positive theory of decision making. A decision theory, which only helps in making better decisions with existing preferences, always work in the hypothetical mode: if one desires x do y. Replacing this with a theory of proper preferences would move us to the categorical: do x.

Nonetheless, I think it is tempting for decision theorists to move in this direction. If it is good to help people attain their goals, then surely it is better to help them have more desired goals?

21. It is also worth noting that Plato [57:356A-B] (again through Socrates) provides perhaps the earliest version of the mechanics of making moral choices based on an ethic of utilitarianism:

. . . like an expert in weighing, put the pleasures together and the pains together, set both the near and the distant in the balance, and say which is the greater quantity. In weighing pleasures against pleasures . . . choose the greater and the more; in weighing pains against pains, the smaller and the less; whereas in weighing pleasures against pains, if the pleasures exceed the pains . . . one must take the course which brings those pleasures; but if the pains outweigh the pleasures, avoid it.

Many later philosophers diverge from Plato by drawing a clear distinction between ethical utilitarianism and individual utility maximization. Given his particular view of the polity and society, Plato would not have recognized that following the dictates of ethical utilitarianism and maximizing individual utility do not lead to the same action. This thought is shared by Rousseau.

22. Indeed, choosing the proper goals and choosing the relative weights may very well collapse into a single question since asking, “is a particular goal desirable?” may be meaningless if it is not joined to the question, “How desirable?” Pursuing this issue strikes me as leading more to semantics than substance.

23. Nussbaum [55:294-7] has an interesting passage concerning Aristotle’s backtracking from Plato’s position. According to Nussbaum, Aristotle explicitly rejected the commensurability of all ethical values.

24. Forcing the recognition of the necessity of the ambiguity of our ethical value systems has been a lifelong crusade of the contemporary English philosopher Isaiah Berlin. Berlin credits Hobbes, Vico, and Mill, among others, with keeping the flame of ambiguity alive in the face of constant attempts to find a single framework to cope with man’s multiple commitments. Four essays on liberty [7] and Against the current: Essays in the history of ideas [8] are excellent collections of some of Berlin’s finest essays.

25. Modern communitarianism is largely an update of Aristotelian political and moral philosophy. Alisdair MacIntyre [42], Michael Walzer [94], and Michael Sandel [67] are among the most eloquent proponents of this view.

26. Or even, for some, within ethics -see Harsanyi [27].

27. It is interesting to note that Stigler [82], in his excellent survey of the history of utility theory, neglects to fully motivate the issue by explaining why utility theory is such a major advance over its more wishy-washy theoretical cousins. This article contents itself with following the development of some of the details of the theory (resolution of St. Peterburg’s paradox, the issue of complementary goods, etc.), without fully exploring the rationale for the foundations of the theory.

28. According to Bentham [6:17], the principle of utility is:

. . . that principle which approves or disapproves of every action whatsoever, according to the tendency which it appears to have to augment or diminish the happiness of the party whose interest is in question . . .

This principle has been couched in terms of moral approbation or disapprobation, not in those of choice; the point of the doctrine is to evaluate actions and not to measure utility. As a guide to action, Bentham’s ethical utilitarianism works only because utility is to offer the final and decisive word for all moral questions (quoted in [54:58]):

Every circumstance by which the condition of an individual can be influenced, being remarked and inventoried, nothing . . . [is] left to chance, caprice, or unguided discretion, everything being surveyed and set down in dimensions, number, weight, and measure.

Ethical utilitarianism is meant to solve a fundamental epistemological problem. “Discretion” and “caprice” can be discarded in favor of exact calculation. Finally, morality can be based on rules that give answers, rather than relying upon the ephemeral support of Divine Law or human nature.

Bentham’s utilitarianism (not unlike a large part of philosophy) was a response to a set of particular historical exigencies. Bentham was trying to put forward the view that the state, in deciding among competing policy options, should choose the one that maximizes overall happiness. The precise definition of happiness, and the exact status of individual choice, would not have been of much interest to Bentham as long as some type of utilitarian principle of utility replaced the rather shaky alternatives. It is in this sense that Bentham’s project, like Plato’s, lay in providing a solution to the seemingly intractable issues in decision making.

29. Henry Sidgwick’s The methods of ethics is properly regarded as the most systematic treatment of ethical utilitarianism arising from the 19th century English Utilitarians, Sidgwick, too, was motivated by this over-riding concern to find a rule that could fix action [79:406]:

If we are not to systemise human activities by taking Universal Happiness as their common end, on what other principles are we to systemise them? . . . we have a practical need of determining not only whether we should pursue Truth rather than Beauty, or Freedom or some ideal constitution of society rather than either . . . but also how far we should follow any of these lines of endeavour.

Sidgwick was also careful to backtrack from J.S. Mill’s muddled position on higher and lower pleasures. See the following note.

30. As an historical aside, it is interesting to note that it was J.S. Mill, perhaps the philosopher most closely associated with 19th century utilitarianism, who backtracked from the determinacy demanded by his predecessors. Mill [47:8] combined ethical utilitarianism with a theory of aesthetics that resisted the levelling claims of Bentham, Sidgwick, and others:

It is quite compatible with the principle of utility to recognize the fact that some kinds of pleasure are more desirable and more valuable than others. It would be absurd that, while in estimating all other things quality is considered as well as quantity, the estimation of pleasure should be supposed to depend on quantity alone.

Mill, by introducing higher and lower pleasures, rejects the notion of commensurability (Bentham’s “pushpin” and “poetry”). No longer can utilitarianism be a solution to the dilemmas of decision making; the indeterminacies that originally drove thinkers toward utilitarianism have been reintroduced by Mill. The advantage of utilitarianism lies precisely in being able to provide an answer independent of caprice and discretion, as this section has repeatedly stressed. Introducing individual aesthetic judgment into the picture puts this project in peril. Mill [47:8] tries to repair the holes his aesthetics have introduced with a patch and the neatness of the system seems in danger of collapse:

If I am asked what I mean by difference of quality in pleasures, or what makes one pleasure more valuable than another . . . there is but one possible answer. Of two pleasures, if there be one to which all or almost all who have experience of both give a decided preference, irrespective of any feeling of moral obligation to prefer it, that is the more desirable pleasure.

Mill’s claim that there are individuals to whom judgment can be deferred -those who have experienced both -rings hollow. Even if we accept this principle (and there are good arguments for utilitarians not to do so), we still face difficulties when the condition of “all or almost all” individuals choosing one type of pleasure over another breaks down. One of the buttresses of utilitarianism -that it gives us an answer to ethical questions -disintegrates in Mill’s hands. Calculation can now give way to opinion.

31. Edgeworth [16:97-8] recognized how concepts built for ethical utilitarianism could be wrought into tools for the modern social scientist:

. . . the materials with which exact social science is concerned are no metaphysical shadows, but the very substance of modern civilization . . . Quantity of labour, quantity of pleasure, equality of sacrifice and enjoyment, greatest average happiness . . . [are] the thoughts of leading Englishmen and corner-stone conceptions upon which rest the whole systems of Adam Smith, of Jeremy Bentham, of John Mill, and of Henry Sidgwick.

The insistence here is most certainly on measurability and determinacy.

32. Becker [4], as the most able contemporary defender of the utility framework as applied to microeconomics, deserves to have his views directly addressed. There are four key ideas in Becker’s “economic approach to human behavior”:

(I) The economic approach views decisions as the maximal choice from among the decision maker’s feasible set. Indeed, the relevant costs are defined in such a way to make this true by definition. So far, the thesis has no predictive content.

(II) The economic approach is applicable to all decisions, independently of the nature, size, or importance of the decision once one accounts for all relevant costs.

(III) Individuals generally care about ”deep” goods -love, self-esteem. These deep goods bear an elastic relationship to the “shallow” goods -washing machines, perfume, sun vacations -that are traded in the marketplace.

(IV) All individuals, independently of position within society or even the society to which they belong, have the same deep goods preferences.

Using this model, given the relevant constraints and the relationship between deep and shallow goods, the decisions that will be taken can be predicted. The fourth assertion, that all individuals have identical tastes in deep goods is the bold stroke that moves Becker’s model from being a vacuous description to one having some bite.

33. Interestingly, cost/benefit analysis seems a much more useful and tenable point for economic analysis than the usual position that interpersonal comparisons are impossible. To use the language of utility, a language which values all outcomes in a common currency, without invoking interpersonal comparisons seems to contradict the spirit of the exercise. Moreover, cost/benefit analysis has the advantage of relying upon the ethical doctrine that all count equally, a proposition that may be disputed but certainly has merit.