Game Theory continued - the Prisoner's Dilemma
Pip's comment on my earlier post has reminded me of the Prisoner's Dilemma game, which demonstrates potential for cooperation in relationships. Again, it is taken from Robert Winston's book 'Human Instinct':
Prisoner's Dilemma
There is a simple game that can be used to examine the evolution of human cooperation. It is called the Prisoner's Dilemma and the game is non zero sum.
Imagine a robbery has been jointly committed by two people, Player A and Player B, and immediately after the crime they are caught and brought in for questioning. They are held in separate cells, and the police attempt to get each of the prisoners to implicate the other. For the purposes of the game, imagine there is no prior agreement between A and B; there is no plan to communicate during the game and there is no loyalty between the two prisoners or favours owed. Honour among thieves plays no role in the Prisoner's Dilemma. The aim for each player is simply to act in his or her own best interests.
Imagine you are Player A. The detective tells you that if you implicate Player B, and if B says silent, not only will you be set free, you will also receive a reward. Player B would, in this case, be given 10 years in prison. The reverse also applies: you will get 10 years in prison if Player B implicates you and you stay silent. Secondly, if both you and Player B implicate each other the each of you will be given 5 years in prison. Thirdly, if neither of you talks - in other words, if you co-operate with each other - the police will have no choice but to set you both free.
The best possible outcome, viewing the situation from the outside, is for both players to stay silent - in other words, to co-operate with each other. But what is the rational, self-interested thing to do? If Player B implicates you, the best thing for you to do is to implicate him, otherwise you could end up in prison for ten years rather than five; if Player B stays silent, once again the rational thing to do is to implicate him: you will not only stay out of prison, you will also get a reward.
So, whatever Player B chooses to do, the best course of action is to implicate him, or defect. Although staying silent (choosing to co-operate) would be better all round, rational self-interest tells you to defect. This is why the Prisoner's Dilemma is so called. Self-interested individuals do not necessarily choose the best overall outcome. Both players should choose to defect, implicate each other, and both will get 5 years in prison. Co-operation will be undermined by the tendency to defect, because defection pays.
Robert Winston (2002), pp 327-328.
posted by Nick Dymond @ 15:27
2 comments
2 Comments:
Yes, that's it. Seems like Winston may have borrowed it from our own King's College Professor Lawerence Freedman who used the same analogy in 1981 and then called it the "minimax strategy" rather than the "prisoner's dilemma" that Winston uses. See Crockatt page 146.
I like the Prisoner's Dilemma. It's a very useful teaching device if done properly but it really does depend, in my experience, on isolating the players. I have several times tried to do it in a lecture theatre and it never works. If the participants are able to communicate in even the most subtle ways then cooperation emerges. The nifty thing about that is that you can then illustrate the Liberal argument that international institutions can mediate the anarchy of the international environment by creating a forum for communication. The origin, by the way, goes back to Rousseau who called it the problem of the 'Stag Hunt'. It takes all the hunters working together to catch a stag--which they would all prefer. But if a hare should happen across the path of a hunter he will abandon the stag hunt--and condemn to hunger his colleagues--in order to satisfy his immediate need.
Post a Comment
<< Home