What Would Happen If Prisoners Actually Did The “Prisoner’s Dilemma”?
By now, we’re probably all familiar with the concept of the Prisoner’s Dilemma. Here’s a visual representation just to serve as a reminder as as a reference.
Essentially, the Prisoner’s Dilemma involves two prisoner’s suspected of being guilty for several crimes, but there’s not enough evidence to pin them for their biggest crime. In this scenario, the individuals are taken into two separate rooms and given the chance to defect on their partner, with the incentive that they walk if they confess. The outcomes are depicted in the picture, with “Not Guilty” representing “Not confessing” and “Guilty” representing “confessing”. As we’ve seen, the Nash Equilibrium Dominant Strategy would be for both players to confess, and here’s why:
Both players look at the outcomes and see that 2 years of jail time is better than 5, and that 1 year is definitely better than 3. Therefore their dominant strategy would be to confess. This would be considered the Nash Equilibrium, and we can check and see if it benefits anyone to deviate from the Dominant Strategy. If either player were to deviate from confess, they would be facing 5 years as opposed to 3 if the other decided to still confess, so it’s unlikely that an individual would picked to not confess.
But this is only in theory, and as history has shown, humans tend to not follow predicted theories. So we ask, how would this scenario play out if actual prisoners were placed in this game? In Game Theory, the idea of defecting is always a dominant strategy. But what if you and your partner are incredibly loyal to each other?
Two University of Hamburg economists decided to test this game out in two different scenarios: one with actual prisoners at a women’s prison with the payoff being coffee or cigarettes (no lessened jail time, unfortunately) and students with their payoff being Euros. Your first impression might be that prisoners are more jaded, and therefore are more willing to defect, but the result might just surprise you. At the end of the experiment, the economists concluded that 63% of students chose to defect, while only 44% of inmates chose to defect. Similarly, only 13% of students achieved the best mutual outcome, whereas 30% of prisoners achieved the best mutual outcome.
What can you take away from this? Maybe you learn that prisoners aren’t so bad, untrusting, and “mastermind” calculating after all. In a game that depends on all blind trust and where there’s no chance to retaliate or avenge betrayal, the prisoners are still significantly more willing to cooperate than the students game. What does this conclude for Game Theory in general? Possibly the fact that as mathematically appealing as the Nash Equilibrium is, it doesn’t all line up with behavioral theory that well, and that we can’t use a model to predict people’s trust and distrust for each other.
Source: http://www.businessinsider.com.au/prisoners-dilemma-in-real-life-2013-7