The Prisoners’ Dilemma Tested On Real Prisoners

By Debra Kelly on Sunday, December 20, 2015
Man in handcuffs praying
“Remember those in prison as if you were their fellow prisoners, and those who are mistreated as if you yourselves were suffering.” —Hebrews 13:3

In A Nutshell

The prisoner’s dilemma presents two people who can’t communicate with a complicated test of how well people can work together and how much faith a person has in the idea that another will make a choice that will be mutually beneficial. It was only recently that University of Hamburg researchers put the scenario to the test with actual prisoners, pitting them against a group of students to see who was more likely to choose the route that would be mutually beneficial to both. In the end, the prisoners were about twice as likely to go the good route as the students were.

The Whole Bushel

The prisoner’s dilemma is a classic scenario that’s used in everything from social science to economics. It goes like this:

Two people are arrested and held in separate rooms for interrogation. There’s not enough evidence against either one of them for a conviction, so police offer a deal to each of them. They can choose to testify against the other person, or they can keep their silence.

If only one person testifies, that person goes free and the other is sentenced to three years in jail. If they both choose to testify against each other, they both get a two-year sentence. If they both keep silent, they each get a year on a lesser charge.

When you look at the game as a whole, the best thing they can choose to do for themselves as a pair is both keep silent, each serving one year in jail. But in order to avoid the possibility of drawing the shortest straw of all—the three-year sentence—an individual needs to testify against the other person. That’s the only way you can possibly get out of jail, and it’s the only way you know you avoid the worst scenario for yourself, making it the logical choice in the face of no real good choices.

The scenario was developed by Merrill Flood and Marvin Dresher of the RAND Corporation and Princeton mathematician Albert Tucker. It’s been applied to countless situations, but it was years before anyone actually applied it to the people named in the scenario: prisoners.

University of Hamburg economists Menusch Khadjavi and Andreas Lange decided to do just that, conducting an experiment where they put real prisoners to the test. They ran their study with a group of women from a primary women’s prison in Vechta, Germany (Lower Saxony just outside Hamburg). For a comparison, they used a group of average students. While the students were drawn from a pool of University of Hamburg students of all different majors, the prisoner group was made up of women serving sentences of anywhere from a few days to life, mostly for drug-related crimes.

There were a couple of different ways they did it. In simultaneous situations, the two parties needed to decide at the same time which action they were going to take, with no knowledge of the other person’s choices. In sequential situations, the second person knows what the first chose to do and must decide whether to cooperate for a mutually beneficial outcome, or to go all in for themselves.

The stakes were a little different—students were offered varying amounts of money, and the prisoners were offered cigarettes or coffee. The principles were the same, though. The research found that college students were much more likely to be out for themselves, while the prisoners were more likely to cooperate with each other.

Results were particularly telling in the simultaneous game, when the college students cooperated for the best mutual result 37 percent of the time.

In the case of the prisoners, though, they cooperated with each other 56 percent of the time.

When second parties knew the choice the first person had made, the likelihood of students to cooperate rose to 63 percent, while the prisoner group maintained their level of cooperation. In the long run, though, only about 13 percent of students tested ended up going the route that would get them the best outcome for them as a pair, while 30 percent of prisoners chose to be mutually beneficial.

Show Me The Proof

Business Insider UK: They Finally Tested The ‘Prisoner’s Dilemma’ On Actual Prisoners—And The Results Were Not What You Would Expect
Journal of Economic Behavior & Organization: Prisoners and their dilemma
The Concise Encyclopedia of Economics: Prisoners’ Dilemma