It is often analytically convenient to abstract from the phenomenon of changing tastes in explaining or evaluating economic phenomena. Alfred Marshall for example defended the assumption of given wants as a useful, if crude, starting point in developing utility theory. Since the 1930s, however, the assumption of given wants has hardened increasingly into dogma. A notable step in that direction was taken in 1932, when Lionel Robbins gave wide currency to the definition of economics as the study of the relations between ends and means, the ends taken as given (Robbins 1932, Chapter 2).

Nobody, of course, supposes that tastes for goods and services are literally biological givens (Stigler and Becker (1977) do deploy this claim, perhaps for its shock value, but it is natural to read them as saying that certain deep-lying wants, as for nourishment and self-esteem, are given and these deep-lying preferences interact with prices and incomes to explain changes in tastes for particular goods and services.)

The more common view is that it is efficient to divide the intellectual labour between economists who study the consequences of given tastes and sociologists, psychologists and others who explain formation of and changes in tastes. Yet on the whole, economists (with a few notable exceptions like Scitovsky (1976)) have shown little interest in what psychologists and others have had to say about preference formation and change. At a minimum, there would seem to be a need for interdisciplinary collaboration on problems, such as long run economic development or cross-national comparisons in consumption patterns, where both causes and consequences of tastes are likely to be important (see, e.g. Felix 1979). If the causes of the taste changes are non-economic, the economists on such teams might usefully concentrate on analysing their consequences.

Sometimes, however, the division of labour will not be so neat. First, tastes may change because of changes in economic variables, making tastes endogenous to the economic system. The alleged influence of advertising on tastes is a standard example; another is the claim that the extension of market production into traditional societies affects tastes for material consumption relative to communal ‘leisure’ activities (Galbraith 1958; Hefner 1983). When such interactions are empirically important, adequate economic models need to include them.

Second, when a consumer’s tastes differ at different points in time, problems of temporal inconsistency in preferences and of intrapersonal preference conflict arise. A consumer who expects to have different preferences in the future faces a planning problem between his present and future self somewhat analogous to that involved in allocating resources between two consumers with different tastes. The challenge of extending the theory of rational intertemporal choice to cover such cases – with the consumer making plans he or she will actually carry out and will not regret – is considerable (Hammond 1976). Further problems arise when preferences change rhythmically or recurrently (Winston 1980). A consumer who is periodically assailed by an impulse to spend or to overeat may take steps in advance to control the consequences of those impulses, burning credit cards or locking the refrigerator. Thomas Schelling has given the label egonomics to the study of such strategic attempts to reconcile intrapersonal preference conflicts (Schelling 1978, 1980; Elster 1979, 1985).

The taste changes discussed so far may simply ‘happen’ to consumers, without their active participation or indeed sometimes without their knowledge. A further degree of complexity is introduced when it is recognized that consumers may have preferences regarding what their tastes should be. A consumer may, for example, prefer that she lose the taste for smoking, say, or acquire a taste for jogging. An adequate economic theory of consumer choice should include such second-order preferences or meta-preferences (Frankfurt 1971; Sen 1977). Such an extension is necessary, most simply, in order to explain the non-trivial expenditures that consumers in advanced societies make on deliberately changing their own tastes – for example, through music-appreciation classes, weight-control clinics, and SmokEnders. More broadly, tension between consumers’ everyday expenditure patterns and their larger views about how they should live – what one might call their values rather than mere preferences (Hirschman 1985) – plays a role in explaining various features of human experience, including ambivalence, ideological commitment and the capacity for self-discipline and self-criticism. These features of behaviour will not always matter for economics, but sometimes they will. Savings behaviour and workplace relations (where willingness to identify with the organization’s goals is an important variable) are obvious applications; Albert Hirschman (1982) has argued that the explanation of fluctuations in the ‘taste’ for political involvement relies heavily on the formation and revision of ideological metapreferences towards politics.

The fact that preferences can change raises issues for the normative as well as the explanatory dimensions of economics. One set of questions involves the role social policy should play in resolving questions of intrapersonal conflict among preferences. Society can provide support for self-control devices, for example by allowing people to enter contracts that bind their future behaviour (such as agreeing to be confined to an alcohol treatment facility). This would then involve overruling the person’s future demands to be released. It is difficult to know from what standpoint to judge the effects of a decision either way on the person’s welfare or liberty (Schelling 1984). Social policy can also undermine self-control, as state-run lotteries show.

Severe difficulties also arise if economic institutions or policies can change preferences. To evaluate such policies or institutions in terms of either the preferences ex ante or ex post (unless the two happen to agree) seems arbitrary. But to decide which set of preferences is itself ‘objectively’ better seems to most contemporary economists just as arbitrary, as well as threatening to liberty. (Many earlier economists, including notably Marshall and J.S. Mill, were much more willing to make and defend value judgements about desirable preferences). An appeal to meta-preferences may help – basing judgement on what preferences people themselves prefer. But of course meta-preferences may themselves depend on economic institutions and policies, so this solution doesn’t go very deep.

From a normative standpoint, both kinds of problems – those of preference conflict and of endogenous preferences – suggest the need to move away from the strictly ‘want-regarding’ moral systems that underlie most neoclassical welfare economics. Such moral systems, labelled ‘welfarist’ by Sen (1979), exclude as morally irrelevant all information about a society except the degree to which individuals’ preferences are satisfied. Other information enters the analysis only insofar as it affects the amount of preference satisfaction achieved. (Utilitarianism and Paretian welfare economics are the most important examples of welfarist moral systems.)

Relaxing the welfarist constraint admits several kinds of information that may help with ‘preference change’ problems. First, procedural issues about how preferences are formed may be recognized as morally important. Processes of preference formation or change that rely on misrepresentation or distortion of facts, or on emotional manipulation, may be morally downgraded irrespective of any judgement about the worth of the preferences that result (McPherson 1982, 1983). Second, measures of individual well-being may be constructed that depend on the resources available to an individual or on the capabilities he or she can exercise, rather than (only) on the amount of preference satisfaction attained (Sen 1982; Rawls 1982). This may partly free evaluations of states of affairs from dependence on existing preferences. Finally, a society can use its public deliberative processes to come to agreement on the objective value of promoting (say, through education) certain preferences or common values (Scanlon 1975). Such agreement need not presuppose that those same preferences would be valuable in other societies or times, and procedural protections could be applied to the means by which these values are promoted.

All the problems discussed in this essay – the dependence of preferences on institutions and policies, the presence of conflicting desires within the person, and the human capacity to evaluate one’s own preferences – were well known to Plato and Aristotle, and were recognized by them as central issues for social theory. That the problems have remained central and largely unresolved for 25 hundred years no doubt makes some economists think it wise to define them out of the discipline, at whatever cost in realism and relevance. Others, however, welcome the resurgence of interest in problems of changing tastes as an opportunity to reestablish links with the other social sciences and with political philosophy.

See Also