Extortioners are only temporarily successful, game theory experiment demonstrates

May 30, 2014
Prisoner's dilemma in the laboratory: egoists are particularly successful when they meet others who are willing to cooperate. Credit: Bastian Ehl

(Phys.org) —Nice people are not always good people. Those who have always felt this intuitively can feel vindicated by the results of a recently published theoretical study: For the so-called Prisoner's Dilemma there are shrewd strategies a player can employ to force an opponent to cooperate before systematically exploiting this same cooperativeness in the end. But how successful is such exploitative behaviour in real life? Scientists from the Max Planck Institute for Evolutionary Biology in Plön have now developed an experiment to study this question in detail. It turns out that extortioners are only successful in the short term: many people do not allow themselves to be permanently exploited.

The Prisoner's Dilemma is a popular experiment among game theoreticians, which they use to study human social conduct. It involves two players, each of whom has no knowledge about the other, deciding either to cooperate or to behave selfishly. If both act selfishly, neither gets much out of it. If they both cooperate, they benefit together. However, the biggest profit is made by the one who decides to be selfish when the other is willing to cooperate.

If the players only come up against each other once, the selfish approach is the more successful. But given that we normally come across people more than once in life, it's more realistic to have each pair play several rounds against each other. In that case, cooperation yields the best result. Of all the game strategies tested to date, the "win stay – lose shift" strategy was long considered the most successful. This is a cooperative strategy in which the players repeat their successful behaviour from the previous round, only switching if the previous behaviour was not successful. For the past 20 years, the maxim for game theoreticians has therefore been: an unfair player cannot win; cooperation has the advantage over selfishness. In real life, however, things often look different.

American scientists recently discovered an even more successful, though morally objectionable strategy. In it, a player does cooperate with their opponent, but only so that they can seek to benefit themselves at the right moment. "Such a player cooperates on occasion, quite deliberately, thereby misleading their opponent to cooperate with greater frequency. The opponent is literally forced to cooperate, only for the other player to strike when the time is right and seek their own advantage," explained Christian Hilbe, who is now researching at Harvard.

Initially only a result of theoretical computations, the Max Planck researchers went on to put the extortionate strategies to the test. The Plön-based scientists devised an experiment in which subjects played 60 rounds of the Prisoner's Dilemma against an unknown opponent – unaware that the opponent was a computer program pursuing various game strategies.

The scientists' findings showed that a player who exploits the other's willingness to cooperate is, in actual fact, much more successful. Hilbe explains: "A player like this lures the other one in and then takes the payoff," said Hilbe. What makes it so deceitful: the opponent has no option but to enter into it and cooperate if they are to get any benefit out of it at all. "It's downright extortion: You're forced to cooperate more and more if you want to have any chance at all of increasing your own profit," explains Manfred Milinski from the Max Planck Institute for Evolutionary Biology. "Many of the test persons who took part in our experiments came out of the experience extremely frustrated and developed real feelings of hatred towards their unknown opponent. Of course what they didn't know was that their opponent was a computer."

The extortioner's approach is calculated to include just enough cooperation to prevent the opponent being completely non-compliant. Sooner or later though most people notice that the other player is acting in bad faith and stop cooperating. They forgo a small profit of their own, thereby depriving the extortioner of a bigger profit. The extortioner is punished, which, in reality, would be likely to make them give in. But not in the mindless computer strategy.

Extortioners damage themselves in the long run: calculated across all rounds of the experiment, the profit achieved through extortionate strategies is relatively small. Generosity is more successful. If the virtual player did not attempt to cheat its human , allowing them some success instead, the end result was always more profitable for the computer, too. "We actually appear to have the right strategies up our sleeve for spotting exploitative behaviour after a while and reacting to it," said Milinski. "We lose a bit to exploiters, but then we punish them as a means of disciplining them. We do need to be on our guard, even with nice people."

Explore further: Spatial order increases cooperation in quantum prisoner's dilemma game

More information: Christian Hilbe, Torsten Röhl, and Manfred Milinski." Extortion subdues human players but is finally punished in the Prisoner's Dilemma." Nature Communications, 29 May 2014. DOI: 10.1038/ncomms4976

add to favorites email to friend print save as pdf

Related Stories

Generosity leads to evolutionary success, biologists show

Sep 02, 2013

With new insights into the classical game theory match-up known as the "Prisoner's Dilemma," University of Pennsylvania biologists offer a mathematically based explanation for why cooperation and generosity have evolved in ...

Tax evaders prefer institutional punishment

Jul 05, 2012

(Phys.org) -- Selfish behaviour is a threat to successful coexistence and mutual cooperation. In many cases this human cooperation is based on punishing those who do not cooperate. There can be two different forms of punishment ...

Recommended for you

World population likely to peak by 2070

Oct 23, 2014

World population will likely peak at around 9.4 billion around 2070 and then decline to around 9 billion by 2100, according to new population projections from IIASA researchers, published in a new book, World Population and ...

Bullying in schools is still prevalent, national report says

Oct 23, 2014

Despite a dramatic increase in public awareness and anti-bullying legislation nationwide, the prevalence of bullying is still one of the most pressing issues facing our nation's youth, according to a report by researchers ...

Study examines effects of credentialing, personalization

Oct 23, 2014

Chris Gamrat, a doctoral student in learning, design and technology, recently had his study—completed alongside Heather Zimmerman, associate professor of education; Jaclyn Dudek, a doctoral student studying learning, design ...

User comments : 4

Adjust slider to filter visible comments by rank

Display comments: newest first

RobertKarlStonjek
3.7 / 5 (3) May 30, 2014
1] Ultimatum and other games are flawed and so give inaccurate and so irrelevant results because Western participants know that there will be no consequences of their actions after the game has ended. Thus they play it as a 'game', a special artificial environment with matching artificial behaviour.

When such games were tested on native people that did not have a good concept of 'game' and treated the 'game' as real, the results were totally unlike anything seen in modern western people indicating that when people treat the game as 'real' they behave entirely differently.

The only example of actual behaviour matching game theory behaviour is in stock market traders who are likewise isolated from the individuals they trade with and act in an entirely mechanical-logical manner.

The following paper proves my point. Just why game theory continued after it was shown to fail in real life is an ongoing mystery.

RobertKarlStonjek
3.7 / 5 (3) May 30, 2014
2] 'Economic Man' in Cross-cultural Perspective: Behavioral Experiments in 15 Small-scale Societies

Abstract (Partial)
Experimental behavioral scientists have found consistent deviations from the predictions of the canonical model of self-interest in over a hundred experiments from around the world. Prior research cannot determine whether this uniformity results from universal patterns of behavior, or from the limited cultural variation among the university students subject pools used in virtually all prior experimental work. To address the above questions, we undertook a cross-cultural study of behavior in Ultimatum, Public Goods, and Dictator Games in fifteen small-scale societies exhibiting a wide variety of economic and cultural conditions.

We found, first, that the canonical selfishness-based model fails in all of the societies studied.
http://www.sscnet...inal.pdf
peter_trypsteen
5 / 5 (1) May 31, 2014
@RobertKarlStonjek
Your link is broken, the working link to the pdf is:
http://www.sscnet...inal.pdf
ForFreeMinds
3 / 5 (4) Jun 01, 2014
This is evidence that people in business for the long haul, will seek mutually beneficial transactions (especially those transactions that can be repeated again and again) rather than seek to benefit at the expense of, or to no benefit for, their customers. And it's evidence of why greedy (and perhaps dishonest) businessmen have to keep starting new businesses as customers learn of their ways and avoid them.

This isn't what I learned in government schools, where they keep claiming that those in business are greedy, while those in government are not. The evidence points to most of those in government being greedy, with most in business seeking mutually beneficial transactions.