The book, unfortunately, is more dogmatic. It develops the rational-choice perspective in an unflinching manner, with little attention to possible objections and alternative explanations. To mention only one point, which will have prove important later, he implicitly dismisses emotion as a possible motivation for human behavior. Although he does cite the important work by Jack Hirshleifer and Robert Frank on emotions as “guarantors of threats and promises”, to use Hirshleifer’s phrase, he trivializes their contribution in a footnote (p. 510) that refers to “crocodile tears”. Whereas crocodile tears are signs of “feigned” emotion (Oxford English Dictionary), the arguments offered by Frank and Hirshleifer turn crucially on the emotions being genuine ones.

In this note, I focus on Coleman’s treatment of social norms. I believe his treatment is deeply unsatisfactory. It is a piece of crypto-functionalism, in spite of his official rejection of that method (pp. 259-260) and his professed methodological individualism (p. 5). It is also somewhat panglossian, in its emphasis on the beneficial effects of norms. Although Coleman is fully aware of norms that benefit only a subset of the relevant community, at the expense of others, he completely ignores norms that make everybody worse off (examples are offered later). In my opinion, social norms cause vast amounts of pointless suffering.

According to Coleman, “a norm concerning a specific action exists when the socially defined right to control the action is held not by the actor but by others” (p. 243). Minimally, this means that A can sanction B’s action if it deviates from the norm, without A thereby becoming the target of disapproval or sanctions from third parties. Sometimes, third parties will say “Why are you meddling with him ? This is none of your business.” At other times they will not make such comments, or even positively encourage the interference. In that case A has the right to meddle, that is, to impose sanctions. Anticipating the sanction, B may or may not find it in his interest to conform to the norm. One reason for conforming arises if the sanction is so severe that he is better off by following the norm than by violating it. Another arises if he would himself want to invoke the norm on later occasions. Coleman says that “if residents of a dormitory attempt to establish a norm that one cannot use the public telephone for more than 10 minutes if others are waiting, then if one resident rejects the legitimacy of such control, he thereby rejects the norm and cannot claim the right to sanction others when they make long telephone calls” (p. 288). However, by violating a norm a person contributes only infinitesimally to its demise. Why could he not violate it and invoke it ? There seems to be a general norm to the effect that if a person violates a norm, he cannot also express disapproval of violators without receiving disapproval from third parties. They will say, “Why are you meddling with him ? You’re no better yourself.” This norm of consistency between words and behavior is itself in need of explanation. It’s not clear to me how Coleman would explain it.

Coleman argues that norms thus defined arise only when actions impose negative externalities on others (p. 275). In any given transaction, the effect of the norm is to favor others (the beneficiaries) at the expense of the actor (the target). The person who abides by the norm of not littering in the park suffers a net loss (the cost to him of not littering is greater than the benefit to him of an unlittered park), whereas all other users of the park derive a small benefit from his abstention. If we look at the same behavior in a collective perspective, however, the class of beneficiaries and the class of actors may coincide. If all users of the park abstain from littering all will be better off than if nobody abstained. But not all norms generalize in this way. For a norm such as “Children should be seen but not heard”, the targets of the norm and the beneficiaries are disjoint groups; hence Coleman refers to them as “disjoint norms” to be distinguished from the “conjoint norms” illustrated by the littering example.

I believe Coleman’s analysis of disjoint norms misses an important aspect of the phenomenon. To capture it, let me add a third group to beneficiaries and targets : the category of enforcers. Enforcers are those who sanction targets who violate the norm. Coleman assumes that the group of beneficiaries and enforcers coincide. In my opinion, however, targets can also act as enforcers of disjoint norms. Children who do not show the proper deference to adults may be penalized by other children. More importantly, social norms intended to keep the lower class in their place are often enforced by internal policing of members of this class by each other. Although Coleman does consider (p. 292 ff.) how targets may internalize disjoint norms, on his account the effect of internalization is merely self-policing by individual target actors rather than mutual policing within the set of target actors. This is related to his statement (p. 292) that internalization installs a propensity in the target to abide by the norm even when not observed by others. I believe this to be empirically wrong. In the language of emotion that Coleman carefully eschews, social norms operate through the emotion of shame rather than guilt. I shall return to that issue.

I have discussed norms that benefit all and norms that benefit some at the expense of others. But some norms do not benefit anyone. [3] Consider the norm in our society against walking up to the person at the head of the bus queue and asking to buy his or her place in the queue. This practice, if allowed, would not harm anyone. The person asked to give up the place is free to refuse. If the offer is accepted both parties to the transaction will be better off and no third parties will be hurt. By blocking such potential Pareto-improvements, the norm makes everybody worse off. Or consider the pointless suffering induced by norms of etiquette, which penalize people for wearing the wrong kind of clothes or having the wrong kind of haircut. The argument that these norms are useful in that adherence to them “will declare one’s group identity to other members and to nonmembers” (p. 258) may be adequate in some cases, but hardly in all. When a small girl comes home crying because her friends ridicule her purchase of the wrong sort of pram for her doll, no useful function is served. These may seem to be inconsequential matters, and in one sense they clearly are. Yet as Tocqueville [4] noted, although “nothing, at first sight, seems less important than the external formalities of human behavior [...], there is nothing to which men attach greater importance”. Proust and Edith Wharton would have concurred.

Some norms that are unambiguously consequential also fail to provide any benefits. In my view, norms of revenge fall in this category. [5] The Mediterranean and Middle Eastern societies that subscribe to these norms have levels of violence and mortality rates among young men far above what is found elsewhere. The idea that the practice of revenge is a useful form of population control is too arbitrary to be taken seriously. The idea that norms of revenge provide a functional equivalent of organized law enforcement in societies with a weak state is also fallacious, albeit more subtly. In many cases, the question “whether feuds created more disruption than they controlled” [6] may be answered in the affirmative. In dueling and feuding societies, many people do in fact engage in deliberate provocation, to insult or offend another. Moreover, one cannot achieve honor by insulting just anybody. In Iceland, “The possession of honor attracted challenges, because that was where honor was to be had.” [7] For a medieval knight, the “prime concern must be pursuit of distinction, and a challenge should never be rejected. Rather he should go out of his way to confront others.” [8] Montaigne refers to “what is said by the Italians when they wish to reprove that rash bravery found in younger men by calling them bisognosi d’honore, ‘needy of honour’ : they say that since they are still hungry for that reputation, which is hard to come by, they are right to go and look for it at any price–something which ought not to be done by those who have already acquired a store of it”. [9] Thus norms of revenge and the code of honor in which they are usually embedded may create as many fires as they put out.

In many small-scale societies there is a general norm against sticking one’s neck out. In Aksel Sandemose’s “Law of Jante” (a mythical small town in Denmark), the fourth of the ten commandments is “Thou shalt not fancy thyself better than we”. [10] Keith Thomas writes that in many primitive societies, beliefs in witch-craft “are a conservative force, acting as a check on undue individual effort. Similarly, in twelfth-century England the chronicler William Malmesbury could complain that the common people disparaged excellence in any sphere by attributing it to demonic aid.” [11] Unlike egalitarian norms that have a redistributive effect, the norm against sticking one’s neck out does not benefit anyone. Although Coleman recognizes this fact (p. 311), he does not try to meet the challenge it poses to his general claim.

Coleman argues that social norms are created and maintained by the rational self-interested action by individuals who exchange in voluntary exchanges with one another. Often, however, his accounts amount to little more than social science fiction. The discussion of voting (p. 289 ff.) is a case in point. Voting may be described as if it were based on exchange of rights, each person holding a little share in everybody else’s vote, but it is clear that no such exchange ever took place in any known society. It is story-telling, not explanation. The problem is not only that the second-order free rider problem (see below) is assumed to be solved. The transaction costs of carrying out the exchange would be prohibitive. Also, it’s often easy to abstain from voting without being observed by anyone.

Coleman recognizes and addresses the “second-order free rider problem” in the provision of public goods. A first-order problem is that of littering. A second-order problem is that of penalizing others for littering, by disapproving remarks, looks, and the like. The benefit to the sanctioner of sanctioning is negligibly small, but the costs and risks may be substantial. If you walk up to someone littering in the park and say “You really shouldn’t do that”, he might hit you in the face. Much ingenuity has been expended on “solving the second-order free rider problem”. Like others, Coleman appeals to “social capital” to explain why people find it in their interest to sanction others for norm-violating. I find this a case of obscurium per obscurius.

In a more specific argument (p. 283 ff.) Coleman refers to gossip as a way of overcoming the second-order problem. His argument, however, seems to involve a confusion between two senses of interest. When he writes that “Each person has an interest in the maintenance of the norm and the application of sanctions to those who violate it comes thereby to have an interest in the spread of information that can lead to a consensus on legitimate sanctions” (p. 284) he means an interest in an outcome. When he goes on to say that “This means that such a person will be interested in listening to gossip and interested in passing gossip” (ibid.), he must mean an interest in the action that will lead to that outcome. But we cannot infer the latter from the former. People gossip because it’s fun. Gossip may or may not have the effect of facilitating sanctions; if it does, that might be just a coincidence. Coleman owes us an account of why the sanction-facilitating effect of gossip can explain the fact of gossip. By offering no such account he leaves himself vulnerable to the charge of crypto-functionalism.

In my opinion, social norms are sustained by the emotions of shame and contempt. [12] By virtue of mechanisms that we only dimly understand, some actions are targeted for social disapproval. These may be actions that harm everybody, that harm some but benefit others, or that benefit everybody. The disapproval is conveyed by gestures or phrases that signify contempt; in Albania, for instance, it can take the form of passing a cup of tea under one’s left arm to a person who has failed to avenge his brother. The correlative feeling in the target of disapproval is the devastating feeling of shame. Thus in 1997 six people killed themselves in France after being exposed as consumers of pedophiliac material : presumably the shame was too much to bear. Prior to the exposure, however, these people presumably led more or less normal lives.

In such cases, we are dealing neither with a fully internalized norm that will be just as causally efficacious in the absence of observers, nor with purely material sanctions such as a loss of one’s job or a fine. Beyond a certain level of satisfaction of material needs, our need for the esteem of others is more important than anything else, except perhaps our need for self-esteem; and their withholding of esteem can be intensely painful. Thus A. O. Lovejoy quotes Voltaire as saying that “To be an object of contempt to those with whom one lives is a thing that none has ever been, or ever will be, able to endure. It is perhaps the greatest check which nature has placed upon men’s injustice”, Adam Smith that “Compared with the contempt of mankind, all other evils are easily supported”, and John Adams that “The desire of esteem is as real a want of nature as hunger; and the neglect and contempt of the world as severe a pain as gout and stone.” [13] Because Coleman focuses exclusively on (i) material sanctions and (ii) internalized norms, he misses what I believe to be the most important feature of the operation of social norms.

As indicated in this diagram, the causal structure of social norms differs from that of moral norms. There are other differences too. Anger is triggered by the action of the norm-violator, contempt by his or her character. Similarly, guilt is triggered by the belief that one has done a bad action, shame by the feeling that one is a bad person. The guilty person can hope to rid himself of guilt by making amends, but the person in the grip of shame can only hope to escape the contemptuous look of others. I do not want to make too much of the distinction. Littering in public may elicit anger in some observers and contempt in others. There are emotions that involve anger and contempt at the same time, as when a social inferior violates a moral norm. What matters for my purposes is that there are clear-cut cases on both sides of the large borderline area.

To conclude, I believe that Coleman focuses too much on effi-ciency-promoting social norms and that even for these norms he fails to show how their promotion of efficiency helps to explain them. The second problem is a pervasive one. Kenneth Arrow, for instance, makes the very same mistake in an even more transparent form :

Arrow and Coleman both exemplify a mind-boggling combination of rational-choice individualism and society-wide functionalism. In Arrow’s case, the lack of microfoundations is evident. In Coleman’s case, the fallacies involved are more subtle, but in the end I do not think he is any more successful.