I have three stories for you. They are all true stories. The first is the result of personal experience. I don't know how many of you know the Hamptons on Long Island, but my wife and daughter and I spend August on the Hamptons. There's a place there that used to be called the East Hampton Dump; it's now, not surprisingly, the "East Hampton Recycling and Disposal Center." If you go there in August, especially in late August, you'll see a lot of people there, even though this isn't a very poor neighborhood, East Hampton, and you'll see people with very nice cars and clothes and sweaters. What they are doing there is separating their garbage, patiently putting the green glass in the green glass bin and the newspaper in the newspaper bin and the other paper in the other paper bin, and so on. It's pretty complicated. If you look at their faces as they are separating their garbage in late August, they look happy.

The second story has to do with cigarette smoking. A lot of attention has been paid recently to the incidence of smoking among American teenagers; that of course is at the core of President Clinton's current concern with tobacco smoking. What hasn't received much attention is the dramatic decrease in cigarette smoking among African-American teenagers. The incidence of smoking among African-American teenagers has pretty much plummeted since the mid 1960s, while among white teenagers it's been pretty steady, at least in the last fifteen years. In fact, it's plummeted so much that it was down to about 4 percent in 1993, while it's something like 20 percent among white teenagers. The Centers for Disease Control says we have an unexplained and not understood public-health success story. I don't have a full account of what happened with respect to African-American teenagers and smoking, but we might get a clue if we look at subways in Harlem, which have posters in them that have a picture of a white skeleton looking like the Marlboro man hovering over an African-American child, and the caption below says, "They used to make us pick it, now they want us to smoke it."

The third story involves a game invented by economists. Economists haven't invented a lot of games, but here's one that they are playing. I've actually had my students play it. The game is called "The Ultimatum Game," and two people play it, one is the proposer and the other the responder. There's a pot of money allocated to the proposer--this isn't very complicated, so if it sounds like it's going to be, it isn't. The proposer gets a pot of money and the proposer can propose to the responder a division of the amount given. If the responder says okay, then they both keep the relevant amounts. If the responder says no, then the total amount goes back to the instructor. It's called the "Ultimatum Game," because it's a one-shot proposal and a yes or no. Now out of the ten dollar pot, the economists' proposition is that the proposer ought, as a rational self-interested person, to propose that nine dollars and seventy-five cents stay with him, and that one quarter goes to the responder. And the economists propose that the responder ought to say an enthusiastic yes to that offer, both acting in self-interested fashion. What would you expect to be observed? What is observed is that often proposals that offer less than fifty-fifty divisions are refused, that it's rare for proposers to propose less than a seventy-thirty division, and it's rare for responders to accept more than a seventy-thirty division. So systematically, proposers are fairer than would be anticipated by economic theory, and responders are more "prickly" than would be expected by economic theory. I did play this at the University of Chicago Law School, and we got the normal results. There were some violations, where we had enthusiastic yes responses by responders to weirdly unfair divisions, and proposers who were very self-interested and greedy. It turned out these people were all economics majors.

If these three stories are all about social norms and my topic is social norms and the relationship between existing ones and the United States government, the motivation is partly practical. If you look at the preventable risks of death in the United States, around 400,000 are from tobacco smoking, 300,000 are from inadequate diet and exercise, 100,000 are from abuse of alcohol, and another 100,000 (and it's on the rise) is the result of firearms, unsafe sexual behavior, and the use of illicit drugs. It shouldn't be controversial to suggest that in all of these contexts, an underlying contributor to the problem is norms, that are producing bad results. So that's the practical motivation.

The theoretical motivation is dissatisfaction with the notions of choice and preference as used in political science, economics, law, and ordinary day-to-day political conversation. That is, there is a pervasive tendency to reject paternalism as a dangerous and frightening specter. And the kind of analytical engine to the anti-paternalist case is the notion of choice or preference. Once we investigate these notions of choice and preference, I think it becomes clear that they are not transparent but confusing, that the notion of choice or preference often conceals the extent to which both choices and preferences are dependent on existing norms for which individuals can claim no responsibility and which they, on reflection, reject. So the suggestion is that the notions of choice and preference conceal the role of social norms in producing those things--social norms that people live according to but that they don't produce on their own individually or collectively.

My descriptive claims are these: Preferences ought not to be taken as bedrock but rather, preferences are a function of things including, importantly, social norms. In addition, many anomalies--both good ones and bad ones--occur in individual behavior, people cooperate when they wouldn't be expected to, and in other circumstances don't cooperate when we would really want them to. What produces the anomaly is existing social norms. We might think of social norms as taxes on, or subsidies to, choice. Where, for example, people might smoke, because they are subsidized by existing norms in their choice to do that, and now increasingly the choice to smoke is taxed by existing social norms. There are communities in which, to carry a gun, is easily subsidized by existing norms, substantially subsidized--and communities, like I'm proud to say my own, in which carrying a gun not only would be a pretty ridiculous and odd thing to do but is also taxed by social norms.
An additional claim is that existing social states are more fragile than they seem. Partly because what people want is a function of what other people are perceived to want. And once our perceptions of other people's desires shift, we can get cascade or bandwagon effects. I think we've seen something like this with respect to both the fall of communism and the attack on affirmative action, where, put to one side what you think of that attack, it's no doubt the case that the attack on affirmative action has been a kind of cascade or bandwagon, partly because social norms have shifted with respect to the legitimacy of criticizing affirmative action--and whence the norm has shifted, then the expression shifts as well.

Those are conceptual claims. My claim about the legitimate role of government is that there is room for government to do something about norms that people don't like but obey--and that predictably lead to shorter or worse lives. The reason there is room for governmental action is most simply that people face a collective social action in changing social norms. To shift a social norm with respect to revenge behavior or with respect to smoking is exceedingly difficult on one's own. In order to get the change we need to act collectively. Sometimes we can act privately, but sometimes public action is the only way to do it. And in the case like that, the objection from freedom to governmental meddling with people's choices or preferences, I suggest, is weightless, because people's choices or preferences are a product of norms that people don't like, but that they live according to. Also I suggest that government action designed to shift social norms is legitimate when it is designed to promote human autonomy or well being, at least in cases when people act as they do because of the reputational consequences of not acting that way. That is, the suggestion that our behavior is frequently a product of the reputational consequences of saying or doing something, and when the reason when we say or do what we do is fear of adverse reputational consequences, at least it's possible that governmental changes in reputational consequences will promote both autonomy and well being.

Maybe a simple way to make this point is that we want to ensure that people are autonomous and free, not only to satisfy their preferences, but also to form their preferences. And sometimes social conditions allow us to satisfy our preferences, again in the obscure term, but they don't allow us to form our preferences freely or autonomously. Think, for example, of someone who uses drugs, not because he really wants to, but because there is a high reputational cost of not doing so in his community. That's the kind of case I have in mind.

That's the introduction of all of my claims. I could end, but instead what I'm going to do is say a little bit more about the underlying concepts, then say something about their implications, and then say something about the legitimate role of government. I have three concepts really, and they are all simple, the main one is the issue of social norms, and the other two are social roles and social meanings--they're really derivative from the notion of social norms. By social norms, I mean social attitudes of approval or disapproval as captured in the notion, "it isn't done," a notion invoked when frequently the relevant thing is done, and what's sought by the term is to prevent its frequent occurrence. There are social norms about a wide range of things: when and how to show affection, how to dress, when and how to be angry, when to talk, when to listen, when to smoke, when to buy insurance, and just about everything else. A key fact about social norms is that they are not within the control of the individual agent--that particular people haven't generated social norms by which they live. There are people who like systematically to defy social norms. Teenagers--my daughter is six and a half and I can't really remember what it was like to be a teenager--but according to books, teenagers systematically like to defy social norms, they see that as an intrinsic good. But most people don't feel that way.

Now there are many consequences from the existence of social norms conditioning both talk and behavior. One thing that they tend to do is to drive a wedge between people's public talk and action and their private beliefs about appropriate talk and action. So, for example, up until recently in Hungary, people didn't buckle their seat belts in a taxicab because there was a social norm in accordance with which to buckle your seat belt was to accuse the driver of hazardous driving, and in that case, people privately often would like to buckle their seat belts, but didn't because of the norm in accordance with which buckling was a kind of accusation. There is a wedge driven between both private and public behavior, with respect to both smoking, and discriminating. It works in various ways. There are communities in which people don't discriminate on grounds of race or gender because there is a social norm in accordance with which that is very bad. And there are communities in which people do discriminate on grounds of race and gender, because there is a norm in accordance with which discrimination is very good. There is an interesting, kind of not much talked about aspect of the Civil Rights Act of 1964, where some of the southern restaurants and hotels actually sought a prohibition on their own discriminatory behavior. It is a kind of puzzle from the standpoint of public choice theory--what were they doing--supporting a ban on their own behavior. Well, one account that some of them give is that they were discriminating, but that they didn't really want to, they wanted to make money, and the only way to make money was by being race neutral. They discriminated because there was a powerful norm against non-discriminatory behavior, and what they wanted was the shielding effect of the law to enable them to do what in fact they wanted to do. They sought a ban on their own "voluntary" behavior, in order to enable to escape norms that were preventing them from maximizing their profits.

We might think then, that people's choices are a function of the intrinsic value of their underlying activity--the reputational effects of the choice and the effects of the self-conception of the choice. And one can think of consumption frequently as a product of a kind of intermingling of intrinsic value, reputational value, and self-conception value. Where you might for example, purchase a book, maybe it's by Stephen Hawking, that does a lot for your reputation and self-conception but it does not have a lot of intrinsic value. And you might not buy a book by Judith Krantz or Stephen King, even though you might enjoy it, let's suppose, because it would really damage your self-conception in order to buy it. Or you'd be afraid someone might see you buying it. Behavior in any case is a function pervasively not just of the intrinsic value, but of the reputational consequences and the self-conception consequences of the choice.

I have a colleague, an economist, with whom I was talking about some of these topics. He said, "You know, there is a puzzle I've been struggling with for years, which is that when my dog does what he does at night, I clean it up." And, he says, "The reason why this is a puzzle is that I don't enjoy cleaning up what the dog does at all, it's an intrinsic cost, not a benefit. I could understand cleaning it up during the day, because there is a severe reputational cost from the failure to clean up, but I clean up at night too and inconvenience myself in a context in which there is no reputational cost from not cleaning up. Why do I do that?"

My hypothesis is that the effects on his self-conception are such that he cleans it up. He sees a failure to clean up as a kind of antisocial act, and therefore, even when there isn't a reputational cost, he cleans up. Both reputational costs and self-conception costs are a function of social norms, that is the effects of our behavior on our self-conception and on our reputation is a consequence of the subsidy or tax on existing social norms. That's the master concept of social norms. Each of us occupies a range of social norms. We may be a spouse, an employee, an automobile driver, a customer, a citizen, and more, and each of our social roles is accompanied by a network of social norms. It is remarkable the extent to which social norms are internalized by each of us acting within our social roles. If you, for example, treated a waiter like you treat a police officer, this being very strange behavior, the waiter would be puzzled. If you treated a police officer like you treat an employee, the police officer might find that disrespectful or puzzling. Each social role is accompanied by a wide range of social norms, and some of those norms have legal sources, that is, some of the norms that govern roles are legally determined.

An especially interesting difference between social norms has to do with the role of the citizen and the role of the consumer. This is one area in which free-marketeers get into philosophically controversial territory. In our capacity as consumers we often make choices and judgments that diverge quite sharply from our choices and judgments in our capacity as citizens. There's a network of norms governing consumer behavior that is just different from the network of norms that governs citizenship behavior. So maybe in our capacity as consumers, partly because of the collective action problem we may not pay attention to altruistic or justice regarding values that in our capacity as citizens we hold dear. And it is, I think, not a function of hypocrisy, but just of prevailing norms, that people in their consumption behavior often make choices that they urge there ought to be laws against in their behavior as citizens. This isn't hypocrisy, but rather a reflection of different norms in a different context. I'll return to this point--the difference between citizens and consumers.

The last concept is maybe the most amusing, that is the notion of social meaning, and here the idea is that both words and actions have a wide range of meanings in the sense that they communicate certain attitudes or commitments to which the speaker or actor might be oblivious. An example with respect to a speaker, the use of the word "Ms." or instead "Mrs." to talk about a married woman, has a certain social meaning; it will signal a certain kind of person. So too would the words "black" or "African-American" have a certain signaling effect. And the signaling effect is one to which the actor may be oblivious. This is why sometimes someone from Europe will be taking a politically controversial stand by the use of certain terms. A stand to which the speaker is utterly oblivious. And there are actions that carry social meanings, too: if I were to light up a cigarette here, my guess is that would either be thought as an effort to exemplify the thesis or a rebellious act of some kind. Whereas twenty years ago, it might be an effort to exemplify the thesis too, but it wouldn't be a rebellious act of any kind. Right now, for a nonsmoker to ask a smoker, "Please don't smoke," is not considered discourteous or prude, or anything of the sort, but perfectly conventional. Whereas, twenty years ago, for someone to ask a smoker not to smoke signaled a kind of prudishness or fanaticism which is very different from its current social meaning. With respect to recycling behavior, drug use, fast, perhaps reckless, driving, there's a social meaning communicated by those acts, a social meaning which is a product of social norms. A particularly important place where this notion of meaning comes up has to do with the problems of safe sex, where there was an article in the New York Times a few months ago, in which some teenage boys said that they don't use condoms even though they really would like to--interesting use of the word "like"--they don't choose to even though they prefer to, and the reason is that use of a condom is an accusation or a confession, and neither is very romantic. That is, the social meaning of condom use is to say, you probably have AIDS, or I might have AIDS, and neither of those assumptions is very desirable to make in the relevant situation. The problem here is that the social meaning of condom use is quite different from what we would like it to be, and therefore condom use is taxed by existing social meanings. It would be really good if that social norm could be shifted; we could also use this as an argument with respect to the social meaning of abstinence, where, the social meaning of abstinence might be the kind of prudishness or fearfulness or unattractiveness. And it would be very good for a variety of reasons if that could be shifted.

Those are my concepts. Now for a series of conceptual or descriptive points and then some straightforward normative points. Often, social norms are more fragile than they seem, because what people do and say is a product to their beliefs about other people's judgments. And once those beliefs start to shift, surprisingly rapid change can occur. A Hungarian said in 1990, I think, "a year ago you couldn't find anyone saying anything bad about communism, now you can't find anyone saying anything good about communism." The astonishingly rapid and mostly nonviolent fall of both communism and apartheid, I suggest, are cascade or bandwagon effects, where people were silenced by their conception of prevailing social norms, and once that conception started to shift, we had a very rapid bandwagon, and then a tipping point. So that criticism of communism, first heavily taxed by existing social norms, then became heavily subsidized by existing social norms. And when we have very rapid large scale social change, it often has much less to do with the coercive effects of government action and much more to do with the kind of a malleable changes with respect to social norms.

A very interesting category of people in this context are what we might call "norm entrepreneurs." You can think of them as including William Bennett, Louis Farrakhan, Judge Bork (the Judge Bork of the most recent book, that is), and Catherine MacKinnon. All four of these people are norm entrepreneurs in the sense that what they are doing is signaling the existence of social norms with which they sharply disagree, and trying to overcome a collective action problem faced by individuals by lowering the costs or increasing the benefits of departure from existing social norms. And though it may be too early to tell with respect to Judge Bork (the book came out too recently), the other three, Bennett, Farrakhan, and MacKinnon, have had a fair degree of success because they have taken advantage of the collective action problem, where people have been living in accordance with norms, which on reflection, they reject.

Sexual harassment is an extremely good example of this, where behavior that was not taxed but perhaps subsidized twenty-five years ago, that is harassing behavior, now produces extremely high norm costs, and there has been a tipping point and a very dramatic shift with according shifts in behavior.
I don't want to spend a lot of time on some technical results in economics, but I'll just tell you a little bit about them. Often economists in experimental and real world situations observe high levels of cooperation because people are talking to one another and because are getting to know one another a little bit. That is, prisoners dilemmas and noncooperative action often can be overcome when people have had a little bit of contact with one another. It suggests that there is a kind of contagion or infectious quality to norms of cooperation under circumstances in which people aren't entirely strangers. These are said to be anomalies from the standpoint of rational choice theory. But if you think to yourself that what is rational to do depends partly on the social costs and benefits of action--that shouldn't be surprising--then it shouldn't be anomalous at all to find cooperation in circumstances in which norms emerge that punish noncooperative behavior. Many of the puzzles that economists have been struggling with over the last twenty years have not been puzzles of rationality, they are cases in which norms of rationality are coming to the fore.

I've suggested that the notions of choice and preference are harder to use for purposes of anti-paternalism once we attend to social norms; let me say a little bit about that. The notion of preferences as it is used by economists is ambiguous as between choices and wellsprings of choices. Sometimes what economists mean by preferences is actual choices, that is Samuelson's notion of the revealed preference. But if we use the revealed preference as a kind of foundation for social policy, we might think that there are problems when people's choices are an artifact of norms which they wish were different. So in a case in which a restaurant is discriminating, or a person is approving in talk and behavior of affirmative action, as a result of norms that that person rejects, the notion that we ought to respect that preference understood as choice runs into difficulty. The person wishes that the preference could be different, but takes the norm as decisive for a decision.

Sometimes the notion of preference doesn't refer to choice, but instead refers to the wellspring of choice. Gary Becker, the University of Chicago's Nobel Prize winner, rejects Samuelson's notion of the revealed and speaks instead of what underlies choices, what are the motivations for choices, that's what preferences refer to. But if we ask ourselves about the motivations for choices and what it means to respect that set of things called preferences, I think we'll rapidly run into conceptual problems. The wellsprings of choices are multiple. When people use drugs or engage in unsafe sex, or take school seriously or not, there's a range of underpinnings of those choices, there isn't a thing called preference that underlies those choices, there's not a physical entity nor is there a mental entity that's very easily used. The wellsprings of a choice, let's say, to take school seriously, or not to get pregnant, are extremely diverse. One of the things that's included in preference is the reputational consequence of the choice, and the reputational consequence depends on social norms. If people's choices are a function of a set of things, let's say intrinsic value, reputational value, and self-conception value, and if the latter two depend on norms which people live in accordance with, but on reflection reject, what's left of the anti-paternalist case?

If government wants to do something about existing social norms, it has a hierarchy of tools. And as good liberals (in the sense of the liberal political tradition), let's start with the less intrusive and describe the hierarchies. And let's prefer the less intrusive. The first is information, where we often see information-induced norm cascades. With respect to smoking or diet and exercise, that's very possible. So the first and least controversial thing for the government to try to do is just provide information. The second is to attempt to persuade, not simply to provide facts, but also to encourage behavior on the theory that it needs to inculcate new and better norms. With respect to drunk driving and littering and safe sex, the government has pretty self-consciously been engaged not just in the project of providing information by attempting to persuade. The government might try to counteract norm subsidies or norm taxes through taxes or subsidies of its own. So the government might use financial incentives, through say a cigarette tax, or a curbside charge on garbage disposal as a way of counteracting, let's suppose, bad norms. Last, the government might coerce, as in the case of a mandatory seat belt law, prohibition on the use of dangerous drugs, and so forth. So that's the hierarchy.

Now, let me conclude, and the conclusion isn't coming immediately, but it's not far away--the word conclude gets people's hopes up, so I have to qualify in that way--by saying something about categories of cases in which government action, I suggest, is justified, once we attend to the liberty-denying effects of norms by which people live and for which people have little enthusiasm. Okay, the clearest cases are those in which we have existing norms that fail to solve collective action problems, or that create collective action problems. That is, where we have norms of noncooperation, in cases in which people don't cooperate given existing social norms, but they would much rather they had prevailing norms which would enable them to cooperate with one another. There's an odd set of laws that I think pick up on precisely this problem. They're kind of odd laws that have a signaling effect, but they're extremely infrequently enforced. I have a dog, a German Shepherd, and so I noticed the fact that there is a law in the city of Chicago requiring people to clean up after their dogs. The existence of the law signals a kind of social judgment about the need for cooperation rather than noncooperation, but the law is just about never enforced. The dog poop law has helped play a role in an extraordinary norm cascade, where behavior twenty-five years ago--cleaning up after your dog, that would have seemed somewhat sick--is now very conventional and that's all to the good. With respect to seat belt buckling, we have a similar phenomenon, where the rate of seat belt buckling up until 1980 or so was about eleven percent and government's educational campaigns didn't do much. What would seem to have induced very dramatic increases in seat belt buckling was the enactment of laws that had a signaling effect, even though those laws were very infrequently enforced. The result is dramatically different social norms with respect to seat belt buckling in many communities now from what we had twenty years ago. For many people, not to buckle is intrinsically more costly because we're in the habit and the reputational cost of not buckling is often high (you look strange, and your self-conception value isn't very high).

There's an issue right now that needs some attention, the issue with respect to air bags exploding on children in the front seat and killing them. One possibility is that the government will get rid of the air bag law, even though my understanding is the aggregate data suggest that the air bag law is good, maybe not. You might think that parents really ought to have their young children in the back seat for various safety reasons--let's just posit that, it may not be true as a matter of fact. If it's so, it would be very good to inculcate a norm that would produce that change in behavior.

Okay, what I am referring to now is cases in which people want norms to be different from what they are. They act in accordance with them given what they are; they can't change them without collective action; and they want that. And the way to do it, the best way or the cheapest way to do it, is governmental. That I think is the easiest set of cases. A somewhat harder set of cases involve instances in which people engage in risk creating behavior, not because of the intrinsic value of the behavior, but because of the reputational costs of not engaging in the behavior. Here examples include carrying guns, smoking cigarettes, drinking too much, and engaging in unsafe sex, unsafe, let's say because of the risk of venereal disease or because of the risk of pregnancy. There are many Americans who engage in all of these forms of behavior not because of the intrinsic value of doing so, but because of the reputational costs of not doing so. And in cases of that sort, to say that we should respect people's choices or preferences is, I suggest, unresponsive to the nature of the problem--where risky behavior is produced not because of intrinsic value but because of reputational costs. There's an additional and related problem to that in which people engage in risk creating behavior because of reputational incentives, and that is in cases in which people's autonomy is at risk because of reputational incentives, and I'm thinking here of the problem of Americans who don't get educated, or otherwise autonomous, because to do that imposes a reputational price. In many communities, people who take schooling seriously face a reputational cost, and that's why they don't. Some communities, and maybe some of you are familiar with these, I'm not familiar with the first, but I am familiar with what I am about to tell you, where to go to law school imposes a kind of reputational cost. It's not that high and it can be overcome by countervailing values, and that's not the most serious problem facing America--the reputational cost of going to law school--but education is a very serious problem. With respect to women and education and work place involvement, there are many contexts, more outside of America than in America, but in America too, where for women to take education and employment seriously is for religious or other reasons to incur a very severe reputational price. In many countries outside of the United States, for women to get literate is to risk their reputation in very severe ways, and for that reason they don't. Have a little bit of involvement with this and when you ask such women, "What do you want? What's your preference?", there isn't a simple answer to that. There is a choice, but the choice is a product of existing norms that the relevant women see for what they are and wish could be different. It's a problem in the United States too, where there's a conflict between traditional norms about women's roles and women's autonomy, where processes of preference formation are adversely affected by existing social norms. There are caste systems all over the world, and the United States has some aspects of caste systems in the sense that, some people's skin color, or gender, or disability has a kind of signaling effect. A signaling effect which is connected with the social norm or social meaning of having that gender or skin color. Now one way to think about anti-discrimination law is that it is designed to overcome the signaling effect of the relevant, highly visible characteristic. Whether or not it works very well, there's no objection from the standpoint of liberty, where caste membership is a problem from the standpoint of freedom or autonomy and when that can't be changed without legal activity.

My last category has to do with the distinction between citizens and consumers. As citizens, people often want to change norms that govern their own behavior in their capacity as consumers. I actually liked former Senator Dole's "Hollywood" speech and I liked it partly because it was picking up on the fact that people in their capacity as citizens often adhere to values and aspirations that diverge from their values and aspirations in their capacity as consumers. To say that people as citizens want educational or public affairs or environmental protection that extends beyond their consumption choices is to point to something that may be an anomaly and that holds undoubted dangers. But there's no reason to think that people's consumption choices are real or harder or more honest than their choices in their capacity as citizens. It may be that their choices in their capacity as citizens threaten freedom more because they are coercing people who disagree with them--that's a valid argument. But it's not right to say that their consumption choices should be deemed, in some sense, as normative, or foundational, and always trump their citizen choices. In our capacity as consumers we are regulated by norms different from those that govern our behavior as citizens, and there ought not to be a kind of a priori objection to governmental responses to people's judgments and aspirations, let's suppose in their capacity as citizens, if those diverge from their judgments and aspirations in their capacity as consumers.

Now it's time really to conclude. Just a few more sentences. Much progress in social science might be made using something more fine grained and differentiated than the notion of preference as the engine of modeling. Gary Becker's most recent book, Accounting for Tastes, actually moves in this direction. It's revealing that our number one preference czar is, in his capacity as social scientist, apparently much less satisfied with using that notion as a foundation. He's aware of the fact that people's preferences are an artifact of things and if we want to do social science as well we might as well get to those things, rather than talking about preferences. On the normative side, it is said often everyday that we should respect people's choices and their preferences, but if what I say is correct, preferences and choices are an artifact of things including existing social norms, social meaning, and social roles, and these are things for which people in their individual capacity aren't responsible. Norms, roles, and meanings can be an obstacle to human autonomy and well being.

Now it should be unnecessary to say that there are private norm communities that can fulfill some of the freedom promoting and well-being promoting functions that government might promote. It would also be unnecessary to say that government action might be futile or counterproductive or invasive of rights, or motivated by self-interested private groups with their own agendas--all of that is undoubtedly true. But those questions can be resolved only by reference to the details, my suggestion here is that efforts by government to change social norms in the interest of freedom or well being ought not to be foreclosed by reference to slogans or mantras about the need to respect choice.

Cass R. Sunstein is the Karl N. Llewellyn Professor at the University of Chicago Law School.