What happens when actual prisoners play The Prisoner's Dilemma?




That college kids are prone to selfishness;
that prisoners come to believe cooperation is beneficial in the end.
I’m not surprised in the least…


When I learned this (in college, natch) it was always heavily implied, if not right outright stated, that prisoners wouldn’t cooperate with one another.


Somehow, that’s not surprising, either. Beliefs formed in ignorance / self delusion…


No surprise to me. 1. People in less secure, more stressful situations are more likely to cooperate because it increases the chances of mutual survival. 2. Anti-snitch behavior is profoundly promoted in prison culture. 3. As a university professor I know prisoners are much nicer than college students (ok, 3 is mostly snark).


laughs The researcher missed one crucial factor. Self preservation.

You don’t rat. Period.
I did two and a half years.


I think this has less to do with cooperation and more to do with a severe distrust of the police and prosecutors. Prisoners first think that their interrogator is lying, so would rather side with another prisoner than give up information. I imagine this is significantly different compared to people who are caught on their first offense, too, and would change depending on the person’s social background.


Fick as fieves ain’t they?


Its actually pretty obvious: Multi-round prisoners’ dilemma has an optimal and well known solution: tit-for-tat. Namely, “Cooperate until someone F-s you, and then you F-them up, and F-them up good”. Its only single-round prisoners dilemma that you actually benefit from being the first to be antagonistic. One of my goals in life is to make sure I’m only involved in multi-round prisoners’ dilemma games.

(Aside: I expect some game theory weenie to chime in with something about the Nash equilibrium. And I counter with “when its actually played in bulk in multi-round, tit-for-tat is optimal”. And “don’t do business with game theory weenies” )

In the prison context, everything is multi-round: everyone is in the same place. I’d be frankly surprised if prisoners were not more cooperative with each other than college students, given that random college students are more likely to treat it as single-round.


I agree with the commenter that spent time in prison. The disincentives for cooperating with the authorities against fellow prisoners are to say the least severe, I would not be surprised that when dealing with someone who has been immersed in such an environment to make choices that are not in their best interests simply to avoid cooperating with authorities. The conditioning is likely to be that strong.

I would also not be surprised to see prisoners deciding to work with fellow prisoners rather than with authorities as a conscious choice regardless of immediate consequence out of concern that post study, word would get out that they were the type willing to turn on a fellow inmate for personal gain (which could be a bad thing for their continued health and well being).


The mob had a strategy for this, never talk. Reward those who keep their mouths shut and punish those that don’t.

Jimmy Conway: I’m not mad, I’m proud of you. You took your first pinch like a man and you learned the two greatest things in life.
Young Henry: What?
Jimmy Conway: Look at me. Never rat on your friends and always keep your mouth shut.

In a way, that’s how that NSA does business.


Let’s not forget that the U.S. Justice system widely employs this practice to coerce information from people on trial. Which is deemed immoral and illegal in many other places.


A very diplomatic way of putting it.


Actually, when I’ve had college students play this in the game class that I teach, I’ve had to change the scenario (so that isn’t prisoners informing on one another) just so there’s a chance they will not cooperate. The social stigma for being a rat is so high among them that they simply won’t choose that option when playing the game. That said, I suspect upper-middle class white kids (who aren’t in my school’s typical demographic) might be more willing to screw one another over.


I’d argue that if you’re only giving them coffee and cigarettes, you’re not REALLY playing The Prisoner’s Dilemma game with actual prisoners. You’re playing a variant, but it’s not the same thing when you offer somebody “Hey, cooperate and you’ll get treats, screw the other guy and you may get a few more” as it is when you say “Hey, you’re going to spend a bunch of years behind bars, but you can save yourself that if you just snitch on the other guy.” It’s really a different dynamic. I’m not going to screw somebody over for money (unless they screwed me first), but if it meant potentially avoiding going to jail, I’d have to seriously think about it! (Assuming we both were guilty, if I knew we were both innocent I’d hold the line… If I was guilty and the other guy was innocent, I wouldn’t screw him over, and if I was innocent and knew the other guy was guilty, well, it depends a little on the severity of the crime and punishment… I’m not going to protect a murderer but for small things I might just keep my mouth shut and play dumb and hope they don’t have enough evidence on either of us)l


Well, since students aren’t really going to be spending time in jail, you’re not really playing The Prisoner’s Dilemma with them, either. It’s all theoretical.


Well, yes, but it’s an abstraction in that case, and usually acknowledged as such. But when you play the abstraction with prisoners, you’re still playing an abstraction… and it’s not really, IMHO, all that noteworthy… as opposed to when you ACTUALLY play the Prisoner’s Dilemma with actual prisoners who are actually in that situation.

I guess my point is, the fact that when you say “Actual prisoners playing the Prisoners Dilemma” you’re sort of setting up for the idea that the fact that they’re actually prisoners is relevant, that it’s somehow a more real result because they’re playing the actual thing. But it’s not, and they’re not. Just as the other group are people who happen to be college students playing the abstraction, this is a group who happen to be prisoners playing the abstraction. Neither are actually in the situation described in it, though, so their status as prisoners or college students and how they react is really only a slight curiosity rather than being some especially important aspect. (Well, it is important, but not for the same reason: Because the prisoners are more likely to have a non-optional extended interaction with each other, after the game, than the college students… but that’s not a part of the Prisoner’s Dilemma, as stated… in fact, the opposite might be true, in that, if one cooperates and the other doesn’t, the Prisoners might not have to interact with each other at all for years, whereas here, no matter what the case, they’re going to be in jail together.)


Privileged people find competition fun and exciting, because they know, no matter how badly they screw up, daddy will rescue them. But for the underclass, it’s all about the way the police treat the underclass - why in hell would they snitch, when they are almost guaranteed to be betrayed?


The inhibition against “ratting” in prison is way overstated in this thread. In fact, prisoners are quite frequently offered time off their sentences for ratting on their cellmates and they often take the deal. In fact, much of our “judicial” system is predicated on getting people to rat on each other. The idea of “omerta” being pushed here is from mob movies and tv shows. That said, the notion most of us hold of who and what constitutes a “criminal” is also simplistic and clueless.

Those who have been imprisoned are not necessarily those with “criminal” minds, and those who walk free are not necessarily “innocent”. Most white collar criminals – those who destroy economies through their own greed – are rarely even punished. Most people are incarcerated for non-violent crimes of desperation, where they are in terrible financial situations and they have to do something to pay the rent or feed their family that they otherwise might not do. And others, with quite a bit of overlap, are imprisoned for committing acts with no victim, such as the act of selling or imbibing certain classes of drugs that have arbitrarily been deemed illegal. The cartoon image of criminals as morally depraved, selfishly evil individuals is just that, a cartoon.


Exactly. (The iterated prisoner’s dilemma situation is, of course, exactly why prisoners treat snitches and so harshly… it’s a “crowdsourced tit-for-tat” solution.)