Economic Experiments Reveal How Quickly Discrimination Can Take Root

Discrimination against different individuals can emerge on the basis of the most meaningless of distinguishing features, according to research by Shaun Hargreaves-Heap and Yanis Varoufakis, published in the latest issue of the Economic Journal.

Their experimental testing of the behaviour of 640 volunteers reveals that people can quickly sort themselves into rigid hierarchies, that discrimination is maintained by the complicity of its victims and that victims are more likely to behave co-operatively than the powerful are. The research seems to confirm Aristotle''s famous dictum: ''The weaker are always anxious for justice and equality. The strong pay heed to neither.'' Most people suspect that power is not always distributed fairly: those at the top of ''the pyramid'' do not always deserve to be there, towering over the rest and reaping the better rewards. Yet most people also believe this to be the exception; that systematic differences in socio-economic status and power must reflect systematic differences in intelligence and/or diligence; that most powerful people have earned their success, even if some acquired it through the mysterious operation of serendipity. These researchers argue that the conventional wisdom is too kind on hierarchies!

A laboratory experiment involving 640 volunteers shows that rigid hierarchies might emerge even among people who are identical. Of course, discrimination cannot emerge unless there is at least some distinguishing feature (e.g. some are ''left-hookers'' or have green eyes). In this experiment, the researchers ensured that subjects seemed identical to each other except for a tiny and wholly arbitrary difference: half of the subjects were initially allocated randomly the label ''blue'' and the rest the label ''red''. Would it make a difference?

Subjects interacted repeatedly in the context of a simple two -person game in which the aggressive player gained $2, the submissive player nothing, mutual aggression meant that both lost $2 and mutual acquiescence rewarded each with $1. Subjects never met the same opponent twice in a row and played 32 times. The question was: would the colour labels make a difference? And would one of the two colours emerge as dominant?

The answer is affirmative on both counts: colour labels did influence behaviour greatly and, yes, one of the two colours emerged as the dominant one, in the sense that subjects of that colour would claim, and be granted, the better reward when paired with a subject of the opposite colour.

The reason why this is remarkable is that all of them knew that the colours were arbitrary and, thus, meaningless! In effect, within 20 or 30 minutes, highly discriminatory conventions became established and determined whether a subject would get the $2 or receive nothing on the basis of his/her meaningless colour (as opposed to their personal characteristics, e.g. intelligence, aggression).

We now know that, at least in interactions in which an urge to be aggressive coexists with a mutual fear of conflict (e.g. any negotiation, or work environment, in which individual aggression brings significant gains as long as the other side backs down but, at the same time, conflict-avoidance is important), discrimination can emerge quickly even among ''identical'' people and be as systematic as it is arbitrary.

Discrimination starts because people are fearful of conflict and, when caught in unpredictable interactions, they try to condition their behaviour on any information that comes to hand – even meaningless information (e.g. on the relative aggression of ''blue'' players). Once they do so, an initial, random difference in the behaviour of the ''reds'' and the ''blues'' gets a bandwagon rolling, leading to stable discrimination that succeeds in minimising costly conflict despite being non-rational (why should the ''blues'' be dominant over the ''reds'' or vice versa?).

One striking result of this research is that discrimination, and its conflict-minimising impact, was maintained primarily due to the complicity of its victims. Another striking result was that, when the victims of discrimination met one another, they co-operated 90% of the time in ways that the economist''s standard analysis of self-interested behaviour cannot explain.

By contrast, when the dominant coloured people met, they behaved as the economists would have expected. It is almost as if the subjects wanted to confirm Aristotle''s famous dictum.

''Some Experimental Evidence on the Evolution of Discrimination, Co-operation and Perceptions of Fairness'' by Shaun Hargreaves-Heap and Yanis Varoufakis is published in the July 2002 issue of the Economic Journal. Hargreaves-Heap is at the University of East Anglia; Varoufakis is at the University of Athens.