An Evolutionary Approach to Norms

Abstract

Norms provide a powerful mechanism for regulating conflict in groups, even when there are more than two people and no central authority. This paper investigates the emergence and stability of behavioral norms in the context of a game played by people of limited rationality.

One interesting possibility is the employment of metanorms, the willingness to punsih someone who did not enforce a norm.

A wide variety of mechanisms […] can support norms, including metanorms, dominance, internalization, deterrence, social proof, membership in groups, law, and reputation.

Introduction

Leadership is itself subject to the power of norms, as Nixon learned when he violated political norms in trying to cover up Watergate.

Congress is shaped by […] reciprocity and apprenticeship.

Across many nations, tolerance of opposition is a fragile norm that has great impact on whether a democracy can survive

[I]n the domain of power politics, norms have virtually wiped out colonialism, inhibited the use of chemical warfare, and retarded the spread of nuclear weapons.

Large numbers of individuals and even nations often display a great degree of coordinated behavior that serves to regulate conflict. When this coordinated behavior takes place without the intervention of a central authority to police the behavior, we tend to attribute the coordinated behavior and the resulting regulation of conflict to the existence of norms. To make this appeal to norms a useful explanation, we need a good theory of norms. Such a theory should help explain three things: how norms arise, how norms are maintained, and how one norm displaces another.

[A]wareness of a given norm is most intense precisely when it is being challenged.

The Evolutionary Approach

The three most common types of definitions [of norms] are bsed upon expectations, values, and behavior.

Because for many purposes the most important thing is actual behavior, a behavioral definition will be used in this study.

Definition: A norm exists in a given social setting to the extent that individuals usuallya ct in a certain way and are often punished when seen not to be acting in this way.

This definition makes the existence of a norm a matter of degree, rather than an all or nothing proposition, which allows one to speak of the growth or decay of a norm.

While deductions about what fully rational actors will do are valuable for their own sake, empirical examples of changing norms suggest that real people are more likely to use trial and eerror behavior than detailed calculations based on accurate beliefs about the future.

In an evolutionary approach, […] there is no need to assume a rational calculation to identify the best strategy. Moreover, the evolutionary approach allows the introduction of new strategies as occasional random mutations of old strategies.

The evolutionary principle itself can be thought of as the consequence of any one of three different mechanisms. It could be that the more effective individuals are more likely to survive and reproduce. A second interpretation is that the players learn by trial and error, keeping effective strategies and altering ones that turn out poorly. A third interpretation, and the one most congenial to the study of norms, is that the players observe each other, and those with poor performance tend to imitate the strategies of those they see doing better.

The Norms Game

An individual i has an opportunity to defect.

This opportunity is accompanied by a known chance of being observed, […] or Seen, S.

If player i does defect, he or shee gets a payoff of T (the temptation for defecting), […] and each of the others are hurt (H) slightly. If the player does not defect, no one gets anything.

So far the game is similar to an n-person Prisoner’s Dilemma. The new feature comes in the next step.

If player i does defect, some of the other players may see the defection, and those who do may choose to punish the defector. If the defector is punished (P) the payoff is very painful […], but because the act of punishment is typically somewhat costly, the punisher has to pay an enforcement cost (E).

The strategy of a player thus has two dimensions.

• boldness (B~i~) determines when the player will defect. The player will defect whenever the chance of being seen by someone is less than the player’s boldness.
• vengefulness (V~i~) is the probability that the player will punish someone who is defecting. The greater the player’s vengefulness, the more likely he or she will be to punish someone who is spotted defecting.

Simulation of the Norms Game

Three completely different outcomes appear possible.

• In one of the runs, there was a moderate level of vengefulness and almost no boldness, indicating the partial establishment of a norm against defection
• On two other runs there was little boldness and little vengefulness
• [O]n the remaining two runs there was a great deal of boldness and almost no vengefulness–the very opposite of a norm against defection

All five of the runs begin near the middle of the field, with average boldness and vengfulness levels near one-half. The first thing to happen is a dramatic fall in the boldness level. The reason for the decline is that when there is enough vengefulness in the population, it is very costly to be bold. Once the boldness level falls, the main trend is a lowering of vengefulness. The reason for this is that to be vengeful and punish an observed defection requires paying an enforcement cost without any direct return to the individual. Finally, once the vengefulness level has fallen nearly to zero, the players can be bold with impunity. This results in an increase in boldness, destroying whatever restraint was established in the first stage of the process–a sad but stable state in this norms game.

Metanorms

a metanorm: establishing a norm that one must punish those who do not punish a defection.

The formulation of a metanorms game can help in the exploration of the effectiveness of this mechanism.

Requires also these additions:

• P', the cost of being punished for not punishing a defection
• E', the cost of punishing someone for not punishing a defection

The model makes the critical assumption that a player’s vengefulness against nonpunishment is the same as the player’s vengefulness against an original defection (also that P(Seen) is the same for defection as it is for failure to punish defection)

In all five runs a norm against defection was established. The dynamics are clear. The amount of vengefulness quickly increased to very high levels, and this in turn drove down boldness. The logic is also clear. At first there was a moderate amount of vengefulness in the population. This mean tthat a player had a strong incentive to be vengeful, namely, to escape punishment for not punishing an observed defection. Moreover, when each of the players is vengeful out of self-protection, it does not pay for anyone to be bold. Thus the entire system is self-policing, and the norm becomes well-established.

This result is dependent, however, on the population’s starting with a sufficiently high level of vengefulness. Otherwise the norm still collapses.

Mechanisms to Support Norms

Metanorms

As the computer simulations show, the existence of a metanorm can be an effective way to get a norm started and to protect it once it is established. By linking vengefulness against nonpunishers with vengefulness against defectors, the metanorm provides a mechanism by which the norm against defection becomes self-policing. The trick, of course, is to link the two kinds of vengefulness. Without this link, the system could unravel. An individual might reduce the metavengeance level while still being vengeful and then later stop being vengeful when others stopped being metavengeful.

What the evolutionary approach has done is raise the possibility that metanorms are a mechanism that can help support norms, thus suggesting the interesting empirical question of whether the two types of vengefulness are indeed correlated. My guess is that there is such a correlation. The types of defection we are most angry about are liektly to be the ones whose toleration also makes us angry.

Dominance

Another mechanism for supporting a norm is the dominance of one group over another. For example, it is no coincidence that in the South, white lynched blacks, but blacks did not lynch whites. The whites had two basic advantages: greater economic and political power, and greater numbers.

[I]n the metanorms version of the model, punishments for not punishing a defector would only occur within a group, as illustrated [in the example where a mob of white lynchers burned a courthouse to harm a black defendant, and when one of the mob commented it was a shame for the courthouse to have been burnt, was beaten by the mob for showing non-solidarity].

Without metanorms, even members of the stronger group tend to be free riders, with no private incentive to bear enforcement costs. This in turn leads to low vengefulness and high boldness in both groups. [this implies that whites who don’t exhibit racist behaviors will ‘defect’ more and more from racist behaviors as they discover their white community does not stigmatize non-racist behavior. This explains the tendency for polarization: some communities have a critical mass of racism that promotes racism to extremist levels; some communities lack that critical mass and extremism is then a property of individuals not of the group]

Internalization

[I]nternalization means that the temptation to defect, T, is negative rather than positive. If everyone internalizes a given norm this strongly, there is no incentive to defect and the norm remains stable. Obviously families and societies work very hard to internalize a wide variety of norms, especially in the impressionable young. They do so with greater or lesser success depending on many factors, including the degree to which the norm and its sponsors are seen as legitimate.

An interesting question for future modeling is, How many people have to internalize a norm in order for it to remain stable?

The logic of the norms game suggests that lowering the temptation to defect might not be enough. [I]f no one had an incentive to punish the remaining defectors, the norm could still collapse.

An increased incentive to punish, through internalization or by some other means, would lead some people to feel a gain from punishing a defector. For them, the payoff from enforcement, E, would actually be positive. Such people are often known as self-righteous busy bodies and often are not very well liked by those who enjoy a defection now and then. Given enough people who enjoy enforcing the norm, the question of its maintenance then becomes whether the chance is high or low that the defection will be seen.

Deterrence

While trial and error is a sensible way of modeling players of very limited rationality, it does not capture the idea that players may have a great enough understanding of the situation to do some forward-looking calculations as well as backward-looking comparisons with others.

Social Proof

The actions of those around us serve several functions. [T]hey provide information about the boldness levels of others, and indirectly about the vengefulness levels of the population. [They] contain clues about what is the best course of action even if there is no vengefulness. [T]he actions of others can provide information about how the population has been adapting to a particular environment. If we are new to that environment, this is valuable information about what our own behavior should be. The actions of others provide information about what is proper for us, even if we do not know the reasons. [B]y conforming to the actions of those around us, we fulfill a psychological need to be part of a group.

Membership

Contracts, treaties, alliances, and memberships in social groups all carry with them some power to impose obligations upon individuals. [I]t directly affects the individual’s utility function, making a defection less attractive because to defect against a voluntarily accepted commitment would tend to lower one’s self-esteem. [G]roup membership allows like-minded people to interact with each other, and this self-selection tends to make it much easier for the members to enforce the norm implicit in the agreement to form or join a group. [T]he very agreement to form a group helps define what is expected of the participants, thereby clarifying when a defection occurs and when a punishment is called for.

One might suppose it would be easy for a bold individual to join and then exploit a group […]. Actually, this does not usually happen, in part because the factors just outlined tend to isolate a defector and make it relatively easy for the others to be vengeful.

Law

A law supports a norm in several ways.

[Enforcement]: The most obvious is that it supplements private enforcement mechanisms with the strength of the state. Because enforcement can be expensive for the individual, this can be a tremendous asset. In effect, under the law the collective goods problem of enforcement is avoided because selective incentives are given to specialized individuals (inspectors, police, judges, etc.) to find and punish violations.

[Respect]: The law also has a substantial power of its own, quite apart from whether it is or can be enforced. In most cases, the law can only work as a supplement (and not a replacement) for informal enforcement of the norm. The failure of Prohibition is a classic example of an attempt to enforce a norm without sufficient social support.

[Clarity]: The law tends to define obligations much more clearly than does an informal norm. However, this clarity is gained at the expense of suggesting that conformity with the law is the limit of one’s social obligations.

[O]ften law is the formalization of what has already attained strength as a social or political norm.

[S]ocial norms and laws are often mutually supporting. This is true because social norms can become formalized into laws and because laws provide external validation of norms. They are also mutuallyi supporting because they have complementary strengths and weaknesses. Social norms are often best at preventing numerous small defections where the cost of enforcement is low. Laws, on the other hand, often function best to prevent rare but large defections because substantial resources are available for enforcement.

Reputation

This is an example of the signaling principle: a violation of a norm is not only a bit of behavior having a payoff for the defector and for others; it is also a signal that contains information about the future behavior of the defector in a wide variety of situations.

The signaling principle helps explaing how an “is” becomes an “ought.” As more and more people use the signal to gain information about others, more and more people will adopt the behavior that leads to being treated well. Gradually the signal will change from indicating a rare person to indicating a common person. On the other hand, the absence of the signal, which originally carried little information, will come to carry substantial information when the signal becomes common. When almost everyone behaves in conformity with a signal, those who don’t stand out. These people can now be regarded as violators of a norm–and dealt with accordingly.

A type of behavior with no direct payoffs can become a norm once it develops some sigaling value, as is the case when fashion leaders adopt a new style.

The Origin and Content of Norms

boostrapping: what is the content of behavior that might later turn into a norm?

Dominance and reputation: Doainance can work because if only a few very powerful actors want to promote certain patterns of behavior, their punishments alone can often be sufficient to establish it, even if the others are not vengeful against defections.

In fact, many norms obeyed and even enforced by almost everyone actually serve the powerful. This can happen in forms disguised as equalitarian or in forms that are blatantly hierarchical. An apparently equalitarian norm is that the rich and the poor are equally prohibited from sleeping under bridges at night. A blatantly hierarchical norm is that soldiers shall obey their officers. Both forms are “norms of partiality” to use the term of Ullman-Margalit.

Once started, the strong support the norms because the norms support the strong.

Behaviors will be easier to establish as norms if the optimal response of others is prompt and rewarding. Failing a prompt response, learning can also take place if the delayed punishment is explicitly cited as a response to the earlier defection.

Summary and Conclusion

Because norms sometimes become established surprisingly quickly, there may be some useful cooperative norms that could be hurried along with relatively modest interventions.