Did you ever wonder, when you were a kid, whether your inane "summer camp" actually had some kind of elaborate hidden purpose—say, it was all a science experiment and the "camp counselors" were really researchers observing your behavior?

Me neither.

But we'd have been more paranoid if we'd read Intergroup Conflict and Cooperation:  The Robbers Cave Experiment by Sherif, Harvey, White, Hood, and Sherif (1954/1961).  In this study, the experimental subjects—excuse me, "campers"—were 22 boys between 5th and 6th grade, selected from 22 different schools in Oklahoma City, of stable middle-class Protestant families, doing well in school, median IQ 112.  They were as well-adjusted and as similar to each other as the researchers could manage. 

The experiment, conducted in the bewildered aftermath of World War II, was meant to investigate the causes—and possible remedies—of intergroup conflict.  How would they spark an intergroup conflict to investigate?  Well, the 22 boys were divided into two groups of 11 campers, and—

—and that turned out to be quite sufficient.

The researchers' original plans called for the experiment to be conducted in three stages.  In Stage 1, each group of campers would settle in, unaware of the other group's existence.  Toward the end of Stage 1, the groups would gradually be made aware of each other.  In Stage 2, a set of contests and prize competitions would set the two groups at odds.

They needn't have bothered with Stage 2.  There was hostility almost from the moment each group became aware of the other group's existence:  They were using our campground, our baseball diamond.  On their first meeting, the two groups began hurling insults.  They named themselves the Rattlers and the Eagles (they hadn't needed names when they were the only group on the campground).

When the contests and prizes were announced, in accordance with pre-established experimental procedure, the intergroup rivalry rose to a fever pitch.  Good sportsmanship in the contests was evident for the first two days but rapidly disintegrated.

The Eagles stole the Rattlers' flag and burned it.  Rattlers raided the Eagles' cabin and stole the blue jeans of the group leader, which they painted orange and carried as a flag the next day, inscribed with the legend "The Last of the Eagles".  The Eagles launched a retaliatory raid on the Rattlers, turning over beds, scattering dirt.  Then they returned to their cabin where they entrenched and prepared weapons (socks filled with rocks) in case of a return raid.  After the Eagles won the last contest planned for Stage 2, the Rattlers raided their cabin and stole the prizes.  This developed into a fistfight that the staff had to shut down for fear of injury.  The Eagles, retelling the tale among themselves, turned the whole affair into a magnificent victory—they'd chased the Rattlers "over halfway back to their cabin" (they hadn't).

Each group developed a negative stereotype of Them and a contrasting positive stereotype of Us.  The Rattlers swore heavily.  The Eagles, after winning one game, concluded that the Eagles had won because of their prayers and the Rattlers had lost because they used cuss-words all the time.  The Eagles decided to stop using cuss-words themselves.  They also concluded that since the Rattlers swore all the time, it would be wiser not to talk to them.  The Eagles developed an image of themselves as proper-and-moral; the Rattlers developed an image of themselves as rough-and-tough.

Group members held their noses when members of the other group passed.

In Stage 3, the researchers tried to reduce friction between the two groups.

Mere contact (being present without contesting) did not reduce friction between the two groups.  Attending pleasant events together—for example, shooting off Fourth of July fireworks—did not reduce friction; instead it developed into a food fight.

Would you care to guess what did work?

(Spoiler space...)

The boys were informed that there might be a water shortage in the whole camp, due to mysterious trouble with the water system—possibly due to vandals.  (The Outside Enemy, one of the oldest tricks in the book.)

The area between the camp and the reservoir would have to be inspected by four search details.  (Initially, these search details were composed uniformly of members from each group.)  All details would meet up at the water tank if nothing was found.  As nothing was found, the groups met at the water tank and observed for themselves that no water was coming from the faucet.  The two groups of boys discussed where the problem might lie, pounded the sides of the water tank, discovered a ladder to the top, verified that the water tank was full, and finally found the sack stuffed in the water faucet.  All the boys gathered around the faucet to clear it.  Suggestions from members of both groups were thrown at the problem and boys from both sides tried to implement them.

When the faucet was finally cleared, the Rattlers, who had canteens, did not object to the Eagles taking a first turn at the faucets (the Eagles didn't have canteens with them).  No insults were hurled, not even the customary "Ladies first".

It wasn't the end of the rivalry.  There was another food fight, with insults, the next morning.  But a few more common tasks, requiring cooperation from both groups—e.g. restarting a stalled truck—did the job.  At the end of the trip, the Rattlers used $5 won in a bean-toss contest to buy malts for all the boys in both groups.

The Robbers Cave Experiment illustrates the psychology of hunter-gatherer bands, echoed through time, as perfectly as any experiment ever devised by social science.

Any resemblance to modern politics is just your imagination.

(Sometimes I think humanity's second-greatest need is a supervillain.  Maybe I'll go into that line of work after I finish my current job.)

Sherif, M., Harvey, O. J., White, B. J., Hood, W. R., & Sherif, C. W. 1954/1961. Study of positive and negative intergroup attitudes between experimentally produced groups: Robbers Cave study. University of Oklahoma.

New Comment
65 comments, sorted by Click to highlight new comments since:

Was this before or after Lord of the Flies I wonder?

Anyway, I think children are different enough from adults that you can't conclude much about what adults will do from studying the behavior of children.


Well then, let's take some adults to summer camp!


Ethics be damned we need more experiments like this


How is this experiment unethical? It was just a regular summer camp, with counselors that happened to be taking notes.


What sort of long-term developmental effect do you think this experience had on the boys?

God could be the ultimate supervillian. Except it would make for a very small 'in' group.

If you count every murder, disease, rape, robbery, death for any other reasons, intellectual disability, and addition to uncyclopedia as his responsibility, he already is.

If those are the unfortunate downsides of policies that are worthwhile overall, then I don't think that qualifies for 'supervillain' status.

I mean, if you're postulating the existence of God, then that also brings up the possibility of an afterlife, etc, so there could well be a bigger picture and higher stakes than threescore years and ten. Sometimes it's rational to say, That is a tragedy, but this course of action is still for the best. Policy debates should not appear one-sided.

If anything, this provides a possible answer to the atheist's question, "Why would God allow suffering?"

"Policy debates should not appear one-sided" doesn't in this case give credence to the idea that a world with suffering implies the possibility of the God. Quite the opposite. That is a post-hoc justification for what should be seen as evidence to lower the probability of "belief in just and benevolent God." This is analogous to EY's example of the absence of sabotage being used as justification for the concentration camps in "Conservation of Expected Evidence"

I didn't mean to suggest that the existence of suffering is evidence that there is a God. What I meant was, the known fact of "shared threat -> people come together" makes the reality of suffering less powerful evidence against the existence of a God.

Except it really doesn’t, because a truly omni God (omniscient, omnipotent, omnibenevolent, etc) realistically wouldn’t help against the problem of evil. The point of the reality of suffering being evidence against God isn’t about omnibenevolence existing within a vacuum, but about omnibenevolence existing within the context of joint omnipotence and omniscience - Mackie’s inconsistent triad, so to speak. A God possessing both of the latter wouldn’t have to worry about puny human concerns like logic and the like, so logic or rationality-based theodicies (including those around finding a shared enemy) don’t, in my opinion, provide adequate arguments against the problem of evil.


Ian C.: Is there any reason in particular that you think that adults are so different from children? I would say that most adults most of the time act pretty childish, though they often couch it in a form that seems more mature.

Agreed. The main difference between adults and children, I think, is that adults are more capable of criticizing their own actions according to a moral framework. But they aren't necessarily inclined to do so. Adults who don't question their own thoughts and actions won't necessarily behave any better than children, just more within social convention, since they've had time to absorb those "rules".

I have also speculated on the need for a strong exterior threat. The problem is that there isn't one that wouldn't either be solved too quickly, or introduce it's own polarizing problems.

A super villain doesn't work because they lose too quickly, see Archimedes, Giorgio Rosa, et al.

Berserkers are bad because they either won't work or work too well. I can't see any way to make them a long term stable threat without explicitly programming them to lose.

Rogue AI doesn't work, again because it either self-destructs or kills us too quickly, or possibly sublimes, depending on quality and goal structure.

The best proposal I've ever heard is a rival species, something like an Ant the size of a dog, whose lack of individual intelligence was offset by stealth hives, co-op, and physical toughness. But it would be hard to engineer one.

My friend had the idea that we need a race of bunnies from another planet to infest Earth. They would be a nuisance, nothing more. They would breed and eat crops. But they would be enough trouble that we would have to work together to stop them.

My friend had the idea that we need a race of bunnies from another planet to infest Earth. They would be a nuisance, nothing more. They would breed and eat crops. But they would be enough trouble that we would have to work together to stop them.

Some others have Have Got A Theory that suggests the opposite approach!

(All but Giles)
What cant we do if we get in it?
We'll work it through within a minute,
We have to try, we'll pay the price,
Its do or die,

Hey i've died twice!

What can't we face if we're together?
What's in this place that we can't weather?
There's nothing we cant face....

.... except for bunnies.

You could just move plants and animals to continents where they don't belong. Image what would happen if kudzu was released in the US. Oh wait, it was.


You ever heard the phrase "X is like violence; if it's not solving your problems, it's because you're not using enough of it."? This is the very first time I've heard somebody propose "problems" as the value of X.

I don't want to say what it is for fear of spoilering it, but is anyone else thinking of the same groundbreaking comic book I am? Perhaps that's the supervillain Eliezer is thinking of...

Yes, but only once you brought up comics.

What comic book are you referring to? curious

Rot13 Jngpuzra. The main villain is trying to force the world powers to unite to fight his fake alien invasion, and you aren't supposed to find this out until the end.

Ahhh, thanks.

That... may make for some interesting reading.

last time we spoke about it, Eliezer was of the opinion that the last scene implies that A* V** failed. I thought it was more ambiguous than that.

"Is there any reason in particular that you think that adults are so different from children?"

I believe the main determinant of how people act is their ideas (as against biology or some other factor). So choosing a group of people to represent society who likely have a far narrower set of ideas than actual society is probably a bad experiment. Because it's not just any old difference, it's a difference in the main causal factor.

Children are a good representative sample for society. It is proven that and type of group, no matter the age, will act in the same way. with leaders followers and the middle ground people. Regardless if the group consists of children or adults, a group with that same common goal and ideas with have the same reactions.

I would believe the main influence the larger set of ideas has, is to provide a more extensive set of rationalizations as to why we did as we did, and to express those rationalizations more eloquently when we defend our actions.

"Now that we know who you are...I know who I am. I'm not a mistake! It all makes sense. In a comic, you know how you can tell who the arch-villain's going to be? He's the exact opposite of the hero, and most times they're friends, like you and me. I should've known way back when. You know why, D? Because of the kids. They called me Mr. G."

I love fictional evidence. Interpret as you will.

Eliezer - would you not say that humanity could take its pick of super-villains, but chooses not to do so because this would be akin to taking out flood insurance when there had been no floods in living memory? Nuclear war, near-Earth objects, global warming, grey goo, take your pick of vaguely-disturbing-but-comfortably-removed-from-real-life Doomsday Scenarios.

I fear humanity wouldn't unite, Independence Day-style, until our destruction was pretty much assured. Or, more likely, until the markets noticed that the end was nigh and sought to do something about it.

I've no doubt everyone's well aware of Phil Zimbardo's seminal 1970s prison guard experiments, but if not, http://en.wikipedia.org/wiki /Stanford_prison_experiment

As I pointed out before, Ronald Reagan had the idea of humanity uniting against an anthropomorphic menace long ago.

Didn't George Orwell preempt him in "1984"?

Reagan took office in 1981.

Is this a joke?

Just in case : "1984" was written in 1947. The original title of the book was to be "1948", the editor asked Orwell to change it so he reversed the numbers.Or so I have heard, I can't seem to find the confirmation, if anyone could confirm or infirm ?


I've thought that the single best thing that could happen our species is a hostile alien invasion (short of electronic transcendence, that is).

I don't feel this in/out group bias very strongly -- so I think it's possible to eliminate the mentality under certain circumstances. The question becomes, what are those circumstances, and how can they be reliably recreated?

Well when I look at the behavior of some sports fans it seems so strange. At a football game recently I saw a few people sitting behind the opposition bench and trying to bate the players into a fight.

Global warming shows that it is not so simple to create a common enemey.

"Sometimes I think humanity's second-greatest need is a supervillain."

Isn't this like saying the hurrican was so great it created all those contruction jobs? I agree it would be nice if we could work together more, but lets do it to be productive, not just to maintain status quo.

This may depend on how long the cooperation lasts after the external conflict occurs.

Why the hating on summer camp? The good ones are wonderful.

Please don't spoil important literary works in this thread. Spoilers will be deleted.

Great post. History's main supervillain has been the Devil -- unfortunately, the Rattlers inevitably decide that the Eagles do his dark bidding, and vice versa.

And for all that, The Devil is simply used as more rationalization for pack behavior and scapegoating.


Setting the conversation of a super-villain aside is there another important aspect to this study, such as the unification of two groups at odds through collaboration and teamwork? Segregation is polarizing and continues this 'us vs. them' attitude and often these ideas are challenged when collaboration occurs, voluntarily or forced.

When writing on the internet, it is best to describe children's ages using years, not their position in your local education system.

I wonder what would happen if you left 22 boys together, without an explicit split. Would the factionalize on their own?


If a misspent youth in Boy Scouts is any indication -- for American Midwestern boys, yes.

I wonder what would happen if you left 22 boys together, without an explicit split. Would the factionalize on their own?

Perhaps. They may even end up fighting over glasses. There might be dead pigs involved and maybe a great big glow stick. I think there was something to do with a rock crushing someone.

Now to be fair, the choir boys were already a subgroup. :P Also, Generalization from Fictional Evidence. But I have a feeling you were being facetious.


He was, but it's still good to remember not to argue from fictional evidence. There must be plenty of real-world examples of what happens to young boys fending for themselves; for example, we could look at the Ik who reportedly do basically that with their kids:

Children are minimally cared for by their mothers until age three, and then are put out to fend for themselves. This separation is absolute. By age three they are expected to find their own food and shelter, and those that survive do provide for themselves. Children band into age-sets for protection, since adults will steal a child's food whenever possible. No food sharing occurs within an age-set. Groups of children will forage in agricultural fields, which scares off birds and baboons. This is often given as the reason for having children.

..... How abjectly horrifying. Thank you, gwern. However, I'm not sure to what degree that's applicable... there's an obvious age disparity between each of these groups, which provides the impetus for social factioning.

Has this experiment been repeated since? On kids who weren't growing up in the near aftermath of a great war? It seems to me to be a bit of a stretch to take this as indicative of the nature of all humans everywhere at all times.

Has this experiment been repeated since? On kids who weren't growing up in the near aftermath of a great war?

Would you argue that there is some attribute that is fundamentally different between children growing up in the post WWII era and today (or any other era for that matter)? My very anecdotal evidence is that once any sort of division into groups occurs, children act in a matter very similar to the Ratters and Eagles. There was a gifted and talented program at my elementary school, which consisted of students from across the county who were bussed into the school and took classes seperately from other students. At the graduation pool party, an innocent slash contest escalated into a full out fight between over.. I'd say approximately 40 students, some of whom inflicted relatively significant injuries. Of course, in group bias had always existed throughout the school years, but violence associated with in group bias isn't something that I feel would be atypical in children of different eras.

Um ... all of human history? Does that count?



I think that this shows not just that splitting people into groups makes the people in one group like themselves and hate the people in the other group, but also that when people figure problems out together that they like eachother more.

Here http://www.abc.net.au/radionational/programs/hindsight/inside-robbers-cave/4515060 is a radio show on ABC radio Australia (roughly equivalent to the US PBS), casts a new light on the Robbers Cave experiment.

It is claimed that

  1. Those conducing the experiment came in with a preconceived agenda which they wished to prove ie that conflict easily arises based on quite trivial group differences etc.

  2. There had been two previous failed experiments in which attempts had been made to create the conflicts described in this study and involving in some cases quite blatant attempts to foment conflict (eg false flag attacks) (see at 12:15 in the audio).

  3. In this third experiment conditions were artfully manipulated to create and then defuse conflict. It appears the key issue in creating conflict is that the two groups must not be permitted to get to know each other and become friendly, and that intense competitive situations are needed, preferably with zero or negative sum outcomes. To then defuse the conflict, allowing socialization no longer sufficed and it was necessary to create a common threat or difficulty to bring the groups together again.

So the statement in the main post that

Well, the 22 boys were divided into two groups of 11 campers, and— —and that turned out to be quite sufficient.

Is perhaps overstated significantly.

It is a very interesting listen.

I don't know if the claims are true. Given how 'right' the results of the original experiment feel, and did feel after WWII, one should be on guard.

Edit; corrections - the experiment was not rigged to quite the degree I originally said. But still I would argue not quite as advertised.


See eg 12:15 where the experimenters were frustrated at fratanization and took steps to fan conflict.

[This comment is no longer endorsed by its author]Reply

Also worth noting in this context the great difficulty armies have in getting soldiers to actually kill the enemy. A lot of military training is aimed at desensitizing soldiers to the thought of killing the enemy.

In WWI informal truces kept breaking out along the front.

It appears the key issue in creating conflict is that the two groups must not be permitted to get to know each other and become friendly

Because then, of course, they might start attributing each other's negative actions to environmental factors, instead of assuming them to be based on inherent evil.

I know I'm very late to this thread, but I wanted to mention that this article also provides reasons to not place too much weight on Sherif's results. (Although of course the broad inferences drawn from his results might happen to be true anyway.) In particular, the article suggests Sherif had attempted a similar study earlier (in another location, with other boys), did not find the results he wanted (despite manipulation), and then suppressed that attempt's results:

Despite his pretence of leaving the 11-year-olds to their own devices, Sherif and his research staff, posing as camp counsellors and caretakers, interfered to engineer the result they wanted. He believed he could make the two groups, called the Pythons and the Panthers, sworn enemies via a series of well-timed “frustration exercises”. These included his assistants stealing items of clothing from the boys’ tents and cutting the rope that held up the Panthers’ homemade flag, in the hope they would blame the Pythons. One of the researchers crushed the Panthers’ tent, flung their suitcases into the bushes and broke a boy’s beloved ukulele. To Sherif’s dismay, however, the children just couldn’t be persuaded to hate each other.
After losing a tug-of-war, the Pythons declared that the Panthers were in fact the better team and deserved to win. The boys concluded that the missing clothes were the result of a mix-up at the laundry. And, after each of the Pythons swore on a Bible that they didn’t cut down the Panthers’ flag, any conflict “fizzled”. By the time of the incident with the suitcases and the ukulele, the boys had worked out that they were being manipulated. Instead of turning on each other, they helped put the tent back up and eyed their “camp counsellors” with suspicion. “Maybe you just wanted to see what our reactions would be,” one of them said.

That said, the article doesn't seem to provide evidence of substantial manipulation during the Robbers Cave study itself. So perhaps the conclusion to draw from Sherif's pair of studies is that, under such conditions, intense intergroup conflict will arise naturally roughly half the time.

Note that this article isn't included in the latest edition of Rationality: AI to Zombies, for roughly the reasons listed here (if I remember correctly).