evolution of social norms jessica yung

13
The Evolution of Discriminatory Social Norms Jessica Yung August 2013

Upload: jessica-yung

Post on 16-Jul-2015

87 views

Category:

Documents


0 download

TRANSCRIPT

The Evolution of Discriminatory

Social Norms

Jessica Yung August 2013

The Evolution of Discriminatory Social Norms

IntroductionSocial norms can be neutral or unequal. Neutral social norms do not benefit specific parties more than others. Examples of relatively neutral social norms would be driving on the left/right

(depending where you are), or shaking hands using your right hand. Unequal social norms, on the other hand, are biased towards certain parties. An example would be the preferential treat-

ment of men through giving them higher pay for doing the same jobs, or giving them higher ranking jobs in general.

Today, many instances of discriminatory social norms, most noticeably gender and racial norms,

exist that affect our day-to-day dealings, and so shape societal functioning around the globe. How did these norms come to be, and what makes them evolve in the ways that they do?

A Prelude: Evolutionary Game Theory(Much of this essay will use ideas from evolutionary game theory to analyse social norms. So if

you’re unfamiliar with what that is, here’s a short introduction.)

Classical game theory assumes all players are rational. It also assumes common knowledge - that everyone knows that everyone else is rational, and that everyone knows everyone knows

everyone is rational, and so on. This model is called homo ludens. But in real life, do people ac-tually make completely rational decisions? In complicated situations, it's highly unlikely a person

can perform the necessary calculations in their head and go on to implement the optimal strat-egy. In simple ones, people are still often irrational. In the Ultimatum Game, Player 1 can

choose to give a proportion of, say, £10, to Player 2, and Player 2 has the right to accept the offer or refuse it. Refusing the offer would result in both players receiving nothing. The minimum

Player 1 would have to offer to be accepted in a non-repeated game with two completely ra-tional players would be 1p. Experimentally, however, players often offered a half of the initial

sum, and offering less than a third would frequently result in a refusal. (Güth, Schmittberger and Schwarze, 1982). So we'd say that homo ludens is not an incredibly accurate model of human

behaviour.

Jessica Yung: The Evolution of Discriminatory Social Norms

2

Evolutionary game theory, on the other hand, assumes limited or no rationality. In evolution, genes hardwire certain strategies into various living beings, and so animals within a species can

be divided into different types (by strategies). Different strategies arise through genetic muta-tions. If a certain strategy yields a high payoff, it is said to give the being a high fitness - i.e., it

can reproduce more. Relative fitness then determines the proportion of the population with each of the strategies in the long run.

This biological model, first put forward by Maynard Smith, can be transferred to model social

behaviour among humans as well. Durkheim suggested that beliefs and practices are learned and inherited, and that it is the process of socialisation that makes people into members of soci-

ety. People's values are often moulded by their upbringing, which is primarily determined by the beliefs and values of their parents and the people around them. In a way, then, it is analogous to

genetics assigning strategies. So behaviour can be approximated by dividing people into types according to their values. Mutations can represent the rise of new ideas, such as those of Rous-

seau, Locke and Montesquieu during the Enlightenment, or Marx in the mid-19th century. Peo-ple with higher payoffs can be said to be more successful, and so will not only be emulated by

their successors, but also by others that aspire to be successful, since it is assumed that suc-cess is something humans are attracted to (and that humans are not completely irrational).

People are assumed to learn through adaption, and shift towards practices that result in better outcomes without necessarily knowing why the practice is better. So with each iteration, the

proportion of people playing a certain strategy will increase or decrease depending on relative payoffs.

1. How discriminatory social norms came to be

To test whether discriminatory norms would evolve from random conditions, Hargreaves Heap

and Varoufakis (2002) played the Hawk-Dove game repeatedly with 640 people, matching them with random opponents using a computer system so they wouldn’t know who their opponents

were, only what strategies they ended up playing.

Jessica Yung: The Evolution of Discriminatory Social Norms

3

They then did this again, but randomly assigned half the people red labels and the other half

blue labels. This second part was repeated 15 times.

Player 1 / Player 2 Hawk (h) Dove (d)

Hawk (h) (V-C)/2, (V-C)/2 V, 0

Dove (d) 0, V V/2, V/2

Payoff matrix for Hawk-Dove game. e.g. when P1 plays d and P2 plays h, the outcome is dh and so P1 has a payoff of 0 and P2 has a payoff of V. Usually, C > V so hh is the worst possible

outcome for both players.

The theoretical predictions and actual outcomes are as follows:

hh hd/dh dd

Predicted (~%) 11 44 44

Actual, without labels (~%) 29 39.8 31.2

Actual, with labels (~%) 19 52 28

with ‘blues’ dominating (i.e. playing h against reds significantly more than reds did against blues)

9 times, and ‘reds’ 6 times. Once there was a method by which players could distinguish people, discriminatory norms evolved.

The reason for this is Game Theory's classic problem of indeterminacy. There is no obvious

choice as to what to play in Hawk-Dove - playing hawk is better in some cases and worse in others - so people try to latch on to distinguishing qualities to make some sense of how they

should act. If a blue player found that reds seemed to play dove more often in initial stages, they’d play hawk against red players more to try to take advantage of it. This differs from the

Prisoner's Dilemma, where the equilibrium is defect-defect since it's individually in the player's interest to defect no matter what their opponent does, assuming they only care about the pay-

offs in the box.

Jessica Yung: The Evolution of Discriminatory Social Norms

4

Player 1 / Player 2 Cooperate (c) Defect (d)

Cooperate (c) 2,2 -2,3

Defect (d) 3,-2 -1,-1

An example of the Prisoner’s Dilemma.

What were random differences at the beginning evolved into conventions as people related ac-tions with label colours. There was no 'reason' behind a certain group dominating another, which

is also why different colours prevailed in different instances of the experiment.

(One might find it strange that the people with red labels (say) accepted their fate as those play-ing dove, but since even a payoff of 0 was better than the payoff of hh, and they could forecast

that the blues would play h, it would still be better individually to play d.)

If discriminatory norms can arise when there does not seem to be any logic behind it, they are even more likely to when there are biological differences that may justify discrimination, such as

between sexes. In prehistoric times, when staying alive was based much more on hunting and gathering, men would seem superior to women because of their biological advantages in fulfill-

ing those roles. In a time where conventions were not yet established and humans were not sure how to distribute property rights etc., the difference in sex would thus become an obvious

focal point for people to use. From there, continuous self-reinforcement of the norm developed it into the perceptions of gender we know today.

Such selection of equilibria in games has a huge influence on social structure. Around half of the

human population was disadvantaged as a result of sex discrimination. As a result, the subtle differences in individual preferences that mould the initial process end up evolving into the huge

differences between different societies & cultures.

2. How social norms are upheld

It may seem strange that such strong racial & gender discrimination could be maintained for so

long, since employers would be losing out on human capital. However, had employers acted

Jessica Yung: The Evolution of Discriminatory Social Norms

5

individually, they would not have been able to profit from it. Deviating from the norm is analo-

gous to playing dove in the Hawk-Dove game (where V>=C), an inferior strategy if everyone else is playing hawk. If everyone forecasts that others will conform to the norm, it is in their best

interest to do that, too, so forecasts in early stages often become self-fulfilling. A result of play-ing dh would be other firms or customers, who may conform to beliefs about gender or racial

superiority, potentially discriminating against the firm going against the norm, causing heavy losses.

As such beliefs reinforce themselves, they may reach a stage where they are so entrenched

they become morals. Hume thought that, at a phase of behavioural evolution, conventions would sometimes become norms, perhaps under the schemes of politicians.1 People expect

others to follow the norm simply because it’s what usually happens, even though it may not be rational, giving these norms additional strength.

3. How they are overturnedTo better examine the collapse and sustaining of social norms, I developed an asymmetric

game to model what would happen under different circumstances. The model was based par-tially on Axelrod (1986), but differs from most other game interpretations of social norms in that

it is an asymmetric game that is not the Ultimatum Game.

A discriminatory social norm is treated as a two-player asymmetric game, with player type A be-ing treated preferentially over player type B under the current social norm. One person from

Type A and another from Type B from a large population are randomly paired, and the game is repeated many times.

Players can choose to either uphold the social norm (cooperate: C) or defect (D). If D is played

by both players, A & B are treated equally. B being treated preferentially over A will not be con-sidered here, since it can be thought of as a separate stage of evolution altogether.

Jessica Yung: The Evolution of Discriminatory Social Norms

6

1 COHON, R., 2010. Hume’s Moral Philosophy [Online]. The Stanford Encyclopedia of Philosophy (Fall 2010 Edition): Metaphysics Research Lab, Stanford University. Available at: <URL:http://plato.stanford.edu/cgi-bin/encyclopedia/archinfo.cgi?entry=hume-moral> [Accessed 21 June 2013]

If A cooperates and B does not, there is a chance A will observe B defecting. Given A does see it, there is a small chance A will be persuaded to defect as well. Usually though, that is not the

case and A may punish B for defecting, e.g. by lowering their wages.

If B cooperates but A does not, there is, again, a chance that B will defect. Otherwise, the norm ensues.

Model details

Payoff matrix, with acd = payoff A receives when A plays C and B plays D etc.

Jessica Yung: The Evolution of Discriminatory Social Norms

7

Data was obtained by varying the starting conditions and running. computer simulations.

A few observations:

1 Collective action on the part of the oppressed is important. In the model, the initial proportion

of people defecting seems to only decide the speed at which the model converges to its equilib-rium. However, a low rate of conversion would likely result in demotivation on the part of those

opting for change, as well as a chance for those in power to clamp down on resistant move-ments, reducing their effectiveness. So movements such as the African-American Civil Rights

Movement or the Feminist Movement helped by encouraging the necessary collective resis-tance.

2 The lower B’s utility is under the norm, the greater the incentive to defect despite the risks. If

the gap between equality and the norm is smaller, there is less of an incentive. This may be the reason why rights movements are more powerful at first when the division is stark (this may

temporarily increase the guilt factor as well), but less influential / significant when the grossest inequalities are wiped out. And that may be why discrimination against females, ethnic groups

Jessica Yung: The Evolution of Discriminatory Social Norms

8

etc. still remains and yet is not fought against nearly as much as it was before (e.g. during

apartheid) even by those affected.

3 The trend was not always consistent even within A or B. For example, A might seem to tend towards 100% cooperation before switching to a majority defection,2 and the proportion of B that

defects may temporarily increase before tending to 0. Changes are caused by significantly higher/lower proportions of the other party choosing to cooperate/defect: if more of B defect, A’s

relative payoffs upon cooperating are lower. This demonstrates quantitatively the idea that fo-cusing only on certain parties is never enough - society must move together as a whole, else

efforts to undermine unequal norms may be in vain.

Of course, in this model, many assumptions were made, and many factors left out. One key im-provement would be to include the fact that people are more likely to conform to the norm if oth-

ers do, even if it doesn’t affect their payoff. This could be introduced by varying norm according to the proportion of people cooperating.

ConclusionIt’s unfortunate that the arising of discriminatory social norms seems almost inevitable as hu-mans struggle to find something to hold on to in times of uncertainty. And yet, this does not

mean that it is as easy today as it was before for baseless atrocities to spring out of the ground. With the prevalence of equality movements, new pro-equality norms seem to be taking root and

becoming morals. Feminism is increasingly embraced (see response to recent Guardian arti-cle3), and the first line of the Universal Declaration of Human Rights, ‘all human beings are born

free and equal in dignity and rights’, is untouchable.

To understand social norms more, we could try to incorporate ways to account for politically en-gineered mutations as opposed to assuming mutations will only occur randomly with small

Jessica Yung: The Evolution of Discriminatory Social Norms

9

2 Input: b = 0.2; u = 0.1; p = 0.6; v = 0.5; g = 0.2; h = -0.3; X = 0.15; Y = 0.9; c = 0.5 which is quite realis-tic. See Appendix for figures.

3YOUNIS, J., 2013. What happened when I started a feminist society at school. [Online] The Guardian: 20 June 2013. Available at: <URL:http://www.guardian.co.uk/education/mortarboard/2013/jun/20/why-i-started-a-feminist-society> [Accessed 21 June 2013]

probabilities, or devise a method that involves the nuances of human psychology. The chaotic

nature of human behaviour will make this difficult, but advances will certainly be a milestone in the evolution of such studies.

(2387 words)

Jessica Yung: The Evolution of Discriminatory Social Norms

10

BibliographyImage on cover: Bazaarbrief, 2011. Faces. Photograph. Available at: <URL: http://blog.bazaarvoice.com/wp-content/uploads/faces.jpg> [Accessed 21 June 2013]

The following sources, especially Hargreaves Heap & Varoufakis’s text, were used exten-sively in the writing of this essay.

AXELROD, R., 1986. An Evolutionary Approach to Norms. The American Political Science Re-view, 80(4), 1096-1111.

HEIFETZ, A., 2012. Games and Evolution. In: Game Theory: Interactive Strategies in Econom-ics and Management. Cambridge: Cambridge University Press. pp. 260-283.

HARGREAVES HEAP, P., VAROUFAKIS, Y., 2004. The assumptions of game theory; Ch. 6 Evo-lutionary games: Evolution, Games and Social Theory. In Game Theory: A Critical Text (2nd Edi-tion). New York: Routledge. pp. 1-32, 211-266.

ZAITSEVA, I.V., 2010. Analytically Modelling Social Norms Using Evolutionary Game Theory. Bachelor of Arts Thesis, Wesleyan University.

The following sources contributed significantly to my understanding of Game Theory and (some) social theory:

BICCHIERE, C., MULDOON, R., ZALTA, E.N. (Ed.), 2011. Social Norms. [Online]. The Stanford Encyclopedia of Philosophy (Spring 2011 Edition): Metaphysics Research Lab, Stanford Univer-sity. Available at: <URL:http://plato.stanford.edu/archives/spr2011/entries/social-norms/> [Ac-cessed 15 June 2013]

BINMORE, K., 2007. Game Theory: A Very Short Introduction. Oxford: Oxford University Press.

BINMORE, K., SAMUELSON, L., 1994. An Economist’s Perspective on the Evolution of Norms. Journal of Institutional and Theoretical Economics (JITE), 150(1), 45-63.

DIXIT, A., NALEBUFF, B., 1993. Thinking Strategically: The Competitive Edge in Business, Poli-tics, and Everyday LIfe. New York: W. W. Norton & Co.

HUSEN, B., 2009. Markov Processes. [Online]. Ohio State University: Department of Mathemat-ics. Available at: <URL:http://www.math.osu.edu/~husen.1/teaching/571/markov_1.pdf> [Ac-cessed 15 June 2013]

JONES, P., BRADBURY, L., LEBOUTILLIER, S., 2011. Emile Durkheim; Structural-Consensus; Structural-Conflict. In Introducing Social Theory, 2nd Edition. Polity Press, 2011. pp. 7-15, 62-64.

MCKENZIE, A.J., 2009. Evolutionary Game Theory. [Online]. The Stanford Encyclopedia of Phi-losophy (Fall 2009 Edition): Metaphysics Research Lab, Stanford University. Available at: <URL:http://plato.stanford.edu/archives/fall2009/entries/game-evolutionary> [Accessed 14 June 2013]

Jessica Yung: The Evolution of Discriminatory Social Norms

11

The following sources were consulted briefly: BERREBY, D., 2012. Human Irrationality is a Fact, not a Fad. [Online] Mind Matters: Big Think. Available at: <URL:http://bigthink.com/Mind-Matters/human-irrationality-is-a-fact-not-a-fad> [Ac-cessed 16 June 2013]

COHON, R., 2010. Hume’s Moral Philosophy [Online]. The Stanford Encyclopedia of Philosophy (Fall 2010 Edition): Metaphysics Research Lab, Stanford University. Available at: <URL:http://plato.stanford.edu/cgi-bin/encyclopedia/archinfo.cgi?entry=hume-moral> [Accessed 21 June 2013]

COWDEN, C.C., 2012. Game Theory, Evolutionary Stable Strategies and the Evolution of Bio-logical Interactions. Nature Education Knowledge [online], 3(10):6 Available at: <URL:http://www.nature.com/scitable/knowledge/library/game-theory-evolutionary-stable-strategies-and-the-25953132> [Accessed 15 June 2013]

EASLEY, D., KLEINBERG, J., 2010. Evolutionary Game Theory. In: Networks, Crowds, and Markets: Reasoning about a Highly Connected World. Cambridge: Cambridge University Press. pp. 209-227

FAGE, J., 2013. Private communication on evolutionary strategies in biology.

MIEKISZ, J., 2008. Evolutionary Game Theory and Population Dynamics. Arxiv [online] Avail-able at: <URL:http://arxiv.org/pdf/q-bio/0703062.pdf>SANDHOLM, W.H., 2007. Evolutionary Game Theory. University of Wisconsin

SCHWEIZER, U., 1994. Binmore’s and Samuelson’s Perspective on the Evolution of Norms: Comment. Journal of Institutional and Theoretical Economics (JITE), 150(1), 64-67.

Jessica Yung: The Evolution of Discriminatory Social Norms

12

AppendixData re: footnote [2]Input: b = 0.2; u = 0.1; p = 0.6; v = 0.5; g = 0.2; h = -0.3; X = 0.15; Y = 0.9; c = 0.5Key: {prop of A cooperating, prop of A defecting, prop of B cooperating, prop of B defecting, 1}

Jessica Yung: The Evolution of Discriminatory Social Norms

13