Anti-social Punishment

This is a cross post from


There’s a trope among Slo­vak in­tel­lec­tual elite de­pict­ing an av­er­age Slo­vak as liv­ing in a village, sit­ting a lo­cal pub, drink­ing Borovička, crit­i­ciz­ing ev­ery­one and ev­ery­thing but not will­ing to lift a finger to im­prove things. More­over, it is as­sumed that if you ac­tu­ally tried to make things bet­ter, said in­di­vi­d­ual would throw dirt at you and place ob­sta­cles in your way.

I always as­sumed that this car­i­ca­ture was silly. It was partly be­cause I have a soft spot for Slo­vak ru­ral life but mainly be­cause such be­hav­ior makes ab­solutely no sense from game-the­o­ret­i­cal point of view. If a do-gooder is stupid enough to try to al­tru­is­ti­cally im­prove your life, why go into trou­ble of ac­tively op­pos­ing them? Why not just sit safely hid­den in the pub, drink some more Borovička and wait un­til they are done?

Well, it turns out that the things are far more com­plex then I thought.

Public goods game

Benedikt Her­rmann, Chris­tian Thöni and Si­mon Gächter did a study of how peo­ple from differ­ent so­cieties deal with co­op­er­a­tion and pun­ish­ment. You can find the pa­per here and sup­port­ing ma­te­rial here.

The study is based on the “pub­lic goods” game. The game works as fol­lows:

There are four play­ers. Each player gets 20 to­kens to start with. Every par­ti­ci­pant ei­ther keeps them or passes some of them into a com­mon pool. After all the play­ers are done with their moves, each of them, ir­re­spec­tive of how much they con­tributed, gets to­kens equal to 40% of all the to­kens in the com­mon pool. The par­ti­ci­pants can­not com­mu­ni­cate with each other and are un­aware of each other’s iden­tities. The game is re­peated, with the same play­ers, 10 times in a row.

The earn­ings, ob­vi­ously, de­pend not only on sub­ject’s move but also on the will­ing­ness of the other play­ers to co­op­er­ate and put to­kens into the com­mon pool. But free rid­ers get an ad­van­tage. They keep their origi­nal to­kens but also get their share from the pool.

To get a feel­ing of the pay­offs, let’s have a look at the sin­gle-round earn­ings in the ex­treme case where each par­ti­ci­pant ei­ther puts all their to­kens into the pool (“co­op­er­a­tor”) or keeps all the to­kens for them­selves (“free-rider”):

Public goods game with punishment

There’s a var­i­ant of the “pub­lic goods game” where play­ers are able to pun­ish each other af­ter each round of the game. The mechanism is sim­ple. When the round ends the par­ti­ci­pants are in­formed about how much each of them has put into the com­mon pool. Then they de­cide whether to spend some of their to­kens to ad­minister pun­ish­ment. For each to­ken spent on pun­ish­ment you can sub­tract 3 to­kens from the earn­ings of an op­po­nent. The play­ers know that they’ve been pun­ished but they are not in­formed about who ex­actly has pun­ished them.

Par­ti­ci­pant pools

The re­searchers were in­ter­ested in com­par­ing the re­sults of the game among differ­ent so­cieties:

Our re­search strat­egy was to con­duct the ex­per­i­ments with com­pa­rable so­cial groups from com­plex de­vel­oped so­cieties with the widest pos­si­ble range of cul­tural and eco­nomic back­grounds to max­i­mize chances of ob­serv­ing cross-so­cietal differ­ences in pun­ish­ment and co­op­er­a­tion. The so­cieties rep­re­sented in our par­ti­ci­pant pools di­verge strongly ac­cord­ing to sev­eral widely used crite­ria de­vel­oped by so­cial sci­en­tists in or­der to char­ac­ter­ize so­cieties. This vari­a­tion, cov­er­ing a large range of the wor­ld­wide available val­ues of the re­spec­tive crite­ria, pro­vides us with a novel test for see­ing whether so­cietal differ­ences be­tween com­plex so­cieties have any im­pact on ex­per­i­men­tally ob­serv­able dis­par­i­ties in co­op­er­a­tion and pun­ish­ment be­hav­ior. … To min­i­mize so­ciode­mo­graphic vari­abil­ity, we con­ducted all ex­per­i­ments with uni­ver­sity un­der­grad­u­ates who were similar in age, shared an (up­per) mid­dle class back­ground, and usu­ally did not know each other.

Speci­fi­cally, the ex­per­i­ment was con­ducted at the fol­low­ing places:

  • Athens, Greece

  • Bonn, Germany

  • Bos­ton, US

  • Chengdu, China

  • Copen­hagen, Denmark

  • Dnipropetro­vsk, Ukraine

  • In­stan­bul, Turkey

  • Melbourne, Australia

  • Minsk, Belarus

  • Mus­cat, Omman

  • Not­thing­ham, UK

  • Riyadh, Saudi Arabia

  • Sa­mara, Russia

  • Seoul, Sourg Korea

  • St. Gallen, Switzerland

  • Zürich, Switzerland

Re­sults: Public goods game with­out punishment

The re­sults from this ex­per­i­ment are ex­actly as you would ex­pect. The co­op­er­a­tors found out that there was no way to pre­vent free-rid­ing and the amount of re­sources they’ve put into the com­mon pool steadily de­creased. This re­sult repli­cated across differ­ent par­ti­ci­pant pools. Par­tic­u­lar pool may have started with a high or low co­op­er­a­tive be­hav­ior, but as the time went on the co­op­er­a­tion always de­creased.

Re­sults: Public goods game with punishment

Abil­ity to pun­ish free-rid­ers in­creased the co­op­er­a­tive be­hav­ior in most par­ti­ci­pant pools. Free-rid­ers learned that free-rid­ing doesn’t pay off and started con­tribut­ing to the com­mon pool.

How­ever, in­tro­duc­tion of pun­ish­ment had no effect in some of the pools. The con­tri­bu­tions stayed more or less the same through­out the ex­per­i­ment in Minsk, Sa­mara, Dnipropetro­vsk, Mus­cat, In­stan­bul, Riyadh and Athens.

Now that’s an in­ter­est­ing re­sult. What’s go­ing on there? Are mem­bers of some so­cieties re­sis­tant to pun­ish­ment or what?

The re­al­ity turns up to be even more in­ter­est­ing than one would ex­pect.

Anti-so­cial punishment

Her­rmann, Thöni and Gächter found out that par­ti­ci­pants in some so­cieties were en­gag­ing in what they’ve called “anti-so­cial pun­ish­ment”. They were pun­ish­ing co­op­er­a­tors!

In fact, they were pun­ish­ing co­op­er­a­tors so much that the co­op­er­a­tion-en­hanc­ing effect of pro-so­cial pun­ish­ment was en­tirely can­celed.

To make it even more con­fus­ing, the anti-so­cial pun­ish­ment, un­like the pro-so­cial pun­ish­ment which had roughly similar level in all the par­ti­ci­pant pools, differed widely among the pools. While it was al­most non-ex­is­tent in the West, it was com­mon in Eastern Europe, in Mid­dle East and in Greece.

The au­thors then try to find out which as­pects of the so­ciety are cor­re­lated with the high anti-so­cial pun­ish­ment rate:

With re­spect to an­ti­so­cial pun­ish­ment, we found that both norms of civic co­op­er­a­tion and rule of law are sig­nifi­cantly nega­tively cor­re­lated with pun­ish­ment (at P < 0.05). In other words, an­ti­so­cial pun­ish­ment is harsher in par­ti­ci­pant pools from so­cieties with weak norms of civic co­op­er­a­tion and a weak rule of law. Ad­di­tional analy­ses show that an­ti­so­cial pun­ish­ment also varies highly sig­nifi­cantly with a va­ri­ety of in­di­ca­tors de­vel­oped by so­cial sci­en­tists in or­der to char­ac­ter­ize so­cieties. Thus, the ex­tent of an­ti­so­cial pun­ish­ment is most likely af­fected by the wider so­cietal back­ground.

Why on Earth?

I wouldn’t have much to add to the fas­ci­nat­ing re­sults above, I am not a so­ciol­o­gist af­ter all, but I hap­pen to come from a coun­try that is prob­a­bly af­fected by this prob­lem. Slo­vakia hasn’t par­ti­ci­pated in the study, how­ever, it’s a former Ost­block coun­try and as such it is very likely to have re­sults similar to Ukraine, Rus­sia or Be­larus. More­over, lo­cal folk wis­dom, as already men­tioned, has it that the phe­nomenon does re­ally ex­ist. There­fore, hav­ing all the rele­vant con­text and all the in­tri­ca­cies of the lo­cal cul­ture in my head, I should be able to come up with an psy­cholog­i­cally plau­si­ble ex­pla­na­tion of why it would make sense to pun­ish co­op­er­a­tors. It was hard to em­pathize with some­one I dis­agree with on a very fun­da­men­tal level, to put my­self in their shoes, but I think I’ve suc­ceeded and what fol­lows is what I came up with.

Her­rmann, Thöni and Gächter spec­u­late that the anti-so­cial pun­ish­ment may be a form of re­venge. You’ve pun­ished me for free-rid­ing so now I’ll pun­ish you just that you know how it feels! And given that I don’t know who the pun­isher was, I’ll pun­ish all the co­op­er­a­tors who were likely to ad­minister the origi­nal pun­ish­ment in the first place.

While that, I be­lieve, is a part of the equa­tion, the psy­chol­ogy of anti-so­cial pun­ish­ment may be some­how more nu­anced. Let me give you an ex­tremely sim­plified toy ex­am­ple, just to get grip of what may be go­ing on.

When I came to Cara­cas, the first thing I’ve done was to buy a cup of coffee from a street ven­dor. The coffee was very good but when I drank it I was left with an empty plas­tic cup. I’ve car­ried the cup with me for sev­eral hours look­ing for a trash bin. I haven’t found one. Fi­nally, I threw the cup at one of the piles of trash that were heaped against the walls ev­ery­where. If, at that point, some­one chas­tised me for lit­ter­ing I would be ex­tremely an­gry and I would yell at that per­son. In other words, I would ad­minister counter-pun­ish­ment.

Psy­cholog­i­cally, I would be an­gry be­cause, ap­par­ently, ev­ery­one else was lit­ter­ing but it was just me who was picked for the pun­ished. It would be un­just. Also, there were no trash bins so I couldn’t had be­have even if I wanted to. That dou­bles the in­jus­tice. More­over, I was car­ry­ing the cup for hours, you do-gooder mo­ron!

If I was a lo­cal there may have been an ad­di­tional rea­son to over­re­act: I would prob­a­bly be sub­limi­nally an­gry for hav­ing to live among the trash all along. This would be a great op­por­tu­nity to let some of that steam off!

To get back to Eastern Europe, we’ve used to live un­der com­mu­nist regime where all the com­mon causes were ap­pro­pri­ated by the state. Any gains from a con­tri­bu­tion to a com­mon cause would silently dis­ap­pear some­where in the dark cor­ners of the bu­reau­cracy.

Quite the op­po­site: Peo­ple felt jus­tified to take stuff from the com­mons. We even had a say­ing: “If you don’t steal [from the com­mon prop­erty] you are steal­ing from your fam­ily.”

At the same time, steal­ing from the state was, legally, a crime apart and it was ranked in sever­ity some­where in the vicinity of mur­der. You could get ten years in jail if they’ve caught you.

Un­sur­pris­ingly, in such an en­vi­ron­ment, re­port­ing to au­thor­i­ties (i.e. “pro-so­cial pun­ish­ment”) was re­garded as highly un­just — re­mem­ber the coffee cup ex­am­ple! — and anti-so­cial and there was a strict taboo against it. Rat­ting of­ten re­sulted in so­cial os­tracism (i.e. “anti-so­cial pun­ish­ment”). We can still wit­ness that state of af­fairs in the highly offen­sive words used to re­fer to the in­form­ers: “udavač”, “donášač”, “práskač”, “špi­cel”, “fízel” (roughly: “nark”, “rat”, “snoop”, “stool pi­geon”).

I also re­mem­ber how, when I moved to Switzer­land, a lot of my friends said things like: “I’ve heard that Swiss will rat on you at any oc­ca­sion.”

Swiss peo­ple would not un­der­stand. What’s so bad about pun­ish­ing free-rid­ers af­ter all?