THESIS
2015
Abstract
The evolution of cooperation (or cooperation theory) is the study of how cooperation can emerge
and persist. Its modern developments were started by Robert Axelrod; a political scientist. Axelrod’s
work is based on Iterated Prisoner’s Dilemma (IPD). The Prisoner’s Dilemma is a classic
game where two players have the options to cooperate or defect when defection yields higher payoff
than cooperation whatever the other does. Iterated Prisoner’s Dilemma has the two players play the
same repeatedly. Several propositions about the game and collective stabilities of various strategies
provide realistic guidance for how to react in such situations and shed new lights on certain aspects
of evolutionary biology, moral and political philosophy.
In this thesis, we first design model that c...[
Read more ]
The evolution of cooperation (or cooperation theory) is the study of how cooperation can emerge
and persist. Its modern developments were started by Robert Axelrod; a political scientist. Axelrod’s
work is based on Iterated Prisoner’s Dilemma (IPD). The Prisoner’s Dilemma is a classic
game where two players have the options to cooperate or defect when defection yields higher payoff
than cooperation whatever the other does. Iterated Prisoner’s Dilemma has the two players play the
same repeatedly. Several propositions about the game and collective stabilities of various strategies
provide realistic guidance for how to react in such situations and shed new lights on certain aspects
of evolutionary biology, moral and political philosophy.
In this thesis, we first design model that covers a wider range of situations in the real world,
such as Iterated PD with more than two options and Iterated Continuous PD games. We then
conduct several computer tournaments on the new model to analyze how to react accordingly. The
results from the experiment show that the winner of Axelrod’s tournaments, TFT, still leads a strong
position in the games, but has an obvious drawback in the continuous model. Meanwhile, a new class of strategies, namely STEPPER, have an absolute advantage, and a strategy called Avenger
is proved to be collectively stable. We conclude that cooperation can hardly emerge or will take
longer time to emerge in the continuous case. This suggests that the players should be cooperative
or at least cooperative with its own kind, and at the meantime, first try to exploit other competing
players, and when this fails, restore mutual cooperation as soon as possible.
Post a Comment