How ethnocentrism evolves: a simulation of evolutionary dynamics

Cooperation is a paradox—it just doesn’t make sense. Why should I help you when there’s no direct benefit for me? Artem, Professor Tom Shultz, and I have been working for quite some time on a paper about cooperation, and we’re psyched to announce that it’s finally been published in The Journal of Artificial Societies and Social Simulation (JASSS). JASSS is an open web journal, so you can view the full text of our article for free on their website. Or you could skip the 8000 or so words and check out this summary post.

Once again, cooperation is weird. The classic prisoner’s dilemma game shows us why. You and an accomplice are arrested in ancient Babylonia on charges of—I don’t know—goat thievery. But the authorities don’t have enough evidence to convict you. They plan to ring you up on lesser charges, which will result in each of you losing a thumb. But there’s a twist. You and your accomplice are each taken into a separate room and offered a plea deal. Rat out your partner and you’ll get off scot free, but your partner will lose both her hands. If you both rat each other out, you’ll each just lose one hand.

	Cooperate (stay silent)	Defect (rat out partner)
Cooperate (stay silent)	(lose thumb, lose thumb)	(lose both hands, scot free)
Defect (rat out partner)	(scot free, lose both hands)	(lose one hand, lose one hand)

Prisoner’s dilemma payoff matrix: You’re in blue and your partner is in red.

Think about it. No matter what your partner does, you are better off telling on them. If your partner stays silent, you get off with all 10 fingers intact. If your partner sells you out, at least you get to keep one hand. But the best option for both of you is to cooperate and stay silent. You can see why this is a dilemma.

The weird thing is, in the real world people do cooperate, all the time, even though it is not in their own rational self-interest. You can try and chalk it up to honor among thieves or good values instilled at a young age, but that still doesn’t cover the whole picture.

Animals cooperate too (Chase, 1980), albeit in less contrived scenarios. Some plants actually display cooperative behavior (Dudley & File, 2007; Runyon, Mescher, & De Moraes, 2006)—even slime molds and species of bacteria (Lenski & Velicer, 2000; Velicer, 2003).

Seems like natural selection may have a hand in this. But how do can test that? Computers offer a cheap solution. Instead of waiting millions of years to observe evolution in action, we can create computer simulations of evolutionary processes that run in seconds.

Hammond and Axelrod (2006) put together a neat simulation that sheds new light on the paradox of cooperation. Imagine life on a 50 x 50 square grid. You’re born into a square. Every year there’s a chance you’ll clone an offspring into a neighboring square (reproduction is asexual), and a chance you’ll die. That’s basically it, you never move from your square. Boring huh? But there’s one small sliver of excitement. Every year you play a prisoner’s dilemma game with each of your neighbors. You’re not playing to keep you fingers this time, you’re playing to increase your reproductive capacity. Winning is in your own genetic self-interest. The more you win, the more likely you’ll have offspring and keep your genes alive.

	Cooperate	Defect
Cooperate	(reproductive capacity +2, reproductive capacity +2)	(reproductive capacity -1, reproductive capacity +3)
Defect	(reproductive capacity +3, reproductive capacity -1)	(no change in reproductive capacity, no change in reproductive capacity)

Payoff matrix for prisoner’s dilemma game played in Hammond and Axelrod’s (2006) simulation.

Your neighbors are basically the same as you. You’re each born with one of four arbitrary tags (kind of like a skin color). You’re also born with two preset prisoner’s dilemma strategies: a behavior towards individuals with the same tag (cooperate or defect), and a behavior towards individuals with different tags (cooperate or defect). When you play with a neighbor you look at your tag and their tag. If your tags match you use your same tag strategy, if they differ you adopt the other strategy. A simple conditional strategy — that’s all there is to it.

The stakes have changed, but the principle is still the same. Cooperation costs. This time it’s not your thumb, but a little bit of energy that could be better spent improving your own reproductive potential (i.e. working out, learning French, buying a new hat, whatever). Being cooperated with is awesome. You get a reproductive boost without doing any work yourself.

It turns out there are 4 possible strategies you can have in this world:

• Humanitarian – cooperate with anybody regardless of their tag
• Ethnocentric – only cooperate with those who share your tag
• Selfish – cooperate with no one
• Traitorous – cooperate only with those who don’t share your tag

If you have offspring, they inherit your tag and strategy, subject to a small mutation rate. This allows for certain strategies to dominate over time (For a more detailed description please see our published article, or check out this post).

So who wins? Given enough time, ethnocentrism is always by far the most successful strategy, followed by humanitarianism, selfishness, and traitorousness in that order. This seems reasonable enough. A lot of the competition we see in humans and nonhumans does have an ethnocentric flavor (LeVine & Campbell, 1972; Chase, 1980).

Why though? What gives ethnocentrism its edge? This is what our paper attempts to explain. By modifying aspects of Hammond and Axelrod’s model and analyzing it in new ways, we uncover the underlying dynamics that make ethnocentrism so adaptive.

To understand this we first need to rewind history, which, lucky for us, is just a matter of scrolling up a spreadsheet. A typical simulation (or “world”) lasts 2000 generations. That’s 2000 chances to reproduce, 2000 chances to die, and 2000 rounds of prisoner’s dilemma. Oh and one more thing—our world is totally empty to start, so to toss a bit of grease in the wheels we throw a new “immigrant” into the mix with a random tag and strategy at the start of each generation.

To our surprise, when we took a closer look at the data, early worlds turned out to be very different from later ones. While ethnocentrism always winds dominating in the end, the first few hundred generations are marked by intense competition between ethnocentrics and humanitarians. In some worlds, ethnocentrism wins out right away. In others humanitarians attain a fleeting dominance. Still in others, these two strategies are neck and neck, all until about 300 generations, when ethnocentrism invariably begins to step ahead of the pack.

Animation of a world where humanitarianism dominates early.

So the question now becomes, why does ethnocentrism beat humanitarianism? If cooperation benefits the greater good, why stop with ethnocentric cooperation, shouldn’t more cooperation be better?

There are two possible explanations that come to mind:

1. Ethnocentrism beats humanitarianism because ethnocentrics do a better job at suppressing selfish free riders. If an ethnocentric group comes across a group riddled with selfish individuals, they’ll refuse to cooperate. Over time, thanks to the ethnos’ mutual cooperation and the selfish group’s total refusal to even help themselves out, ethnos will reproduce faster than the non-cooperators and thus expand at the selfish group’s expense. Meanwhile those nice humanitarian fellows blissfully waste their precious reproductive potential helping out free riders, who are all to happy to receive their favor, giving nothing in return. We call this idea, that ethnocentrism beats humanitarianism because it is better at suppressing free-riders, the “mediation hypothesis,” and it is the mechanism favored by Hammond and Axelrod in their original paper.
2. Another possibility is that ethnocentrism beats humanitarianism outright. Imagine an ethno group next to a humanitarian group. Individuals on the group boundary benefit from the cooperation of their own group-mates behind them. But the ethnocentrics at the front doubly benefit from the cooperation of those hapless humanitarians. Might this give the ethnos the edge they need? We call this the “direct hypothesis”.

We know the tipping point comes at around 300 generations. But what happens at that point that changes the world so suddenly? Turns out, 300 generations is about when the world starts to fill up. Early on, when population is sparse, group lines are few and far between, and there is little opportunity for ethnocentrism to beat humanitarianism either through the mediated or direct mechanism. The fact that the ethnos’ rise to dominance coincides with population saturation suggests that competition between groups is a factor, but it doesn’t tip the scales toward either hypothesis. So we’re closer but not really.

To get to the bottom of this, we ran restricted simulations where certain strategies were outlawed. For example, we ran worlds with just humanitarian, selfish, and traitorous individuals, or just humanitarian and selfish, or even just selfish individuals.

The results were clear: free riders do not spell doom for humanitarians as was suggested by Hammond and Axelrod. In the absence of ethnocentrism, humanitarianism is by far and away the dominant strategy. The difference is clear even when strategies are viewed in isolation. 2000 generations of no cooperation leads to a much lower population than 2000 of indiscriminate cooperation.

Neither humanitarians nor ethnos really need to do anything to suppress free riders. In a “viscous” environment, where individuals only interact with their neighbors, free riding does a pretty good job of suppressing itself.

This is somewhat counterintuitive. Cooperating with everyone sounds nice in theory, but it seems naively idealistic in practice. Because ethnocentrism has an in-built mechanism for suppressing foreign free riders, it may feel like a more practical strategy for the real world. There’s this idea that ethnocentrism protects us from strangers who may take advantage of our kindness. This simulation, however, suggests the exact opposite. Humanitarianism is nearly as good as ethnocentrism at suppressing free riders. Ethnocentrism is the best strategy, but only because ethnos takes advantage of humanitarians!

Of course, this all hinges on viscosity (i.e. the extent to which movement is restricted). If you are able to freely move and interact with any other individual, the effect of free riding will be much greater. Viscosity is easily apparent in simple animal populations (Gadgil, Joshi, & Gadgjil, 1983; Seppä & Pamilo, 1995)—less so in modern humans what with Megabus, Couchsurfing and all that. Still, in spite of all the moving about we seem to do, psychologists are quick to remind us that spatial and cognitive closeness play a large role in human social relations (Festinger, Schacter, & Back, 1963; Hipp & Perrin, 2009; Katz & Hill, 1958; Kubitschek & Hallinan, 1998; Nahemow & Lawton, 1975). Also, the environments we evolved in offered considerably fewer travel opportunities (Foley, 1995). So it’s not totally far-fetched to apply our understanding here to the evolution of human societies.

Animation of a world where ethnocentrism dominates early on.

Okay. So we know why ethnocentrism beats its closest competitor, but there’s still something odd here. Why do humanitarians dominate at all early on? And why does this prelapsarian period of humanitarian dominance only appear occasionally? When we plot the results from a number of worlds, we see that early strategy dominance varies according to a normal distribution, which is a fancy way of saying it’s probably due to chance.

Remember that each generation an immigrant with a random strategy is placed randomly on the grid. It seems reasonable that a cluster of early humanitarian immigrants in a sparsely populated world could grow rapidly at the expense of later immigrants. Evolution takes any advantage it can get, and in this case a lucky head start can be all that’s needed to give humanitarians an edge—at least until population maxes out and inter-group dynamics kick in.

So-called founder effects like this are not uncommon and can sometimes play a role in human populations. A founder effect is basically the overrepresentation of certain traits that occurs when a large population grows out of only a few select founders. The unusually high proportion of twins in Cândido Godói (often misattributed to cultish Nazi experimentation) and the high proportion of Quebecois descended from the original French pioneer settlers of the 17th century are two real world examples.

We can test this too. If we make just the first 20 immigrants ethnocentric or humanitarian, we very significantly influence which strategy will dominate early on. If we do the same thing 100 generations later, when clusters of like individuals, the effect is insignificant.

At this point, if you’re still paying attention, you’re either bobbing your head in comprehension or banging it against the wall wondering what the heck we’re doing studying the behavior of computer models. (I would imagine a small subset of you are stroking your goatees). This whole business of studying the behavior of unthinking bits of code may a tad reductive—but that’s the whole point. What’s great about Hammond and Axelrod’s model, and many other computational models of evolution, is that they show how complex behaviors can evolve in even the simplest creatures. In our case, all that’s required is a single visible trait (a tag), and the ability to distinguish others based on their tag. Any comprehensive model of ethnocentrism needs to be simple enough to work with bacteria, yet rich enough to apply to human societies.

Still, while this model can explain some human behavior, it’s not at all broad enough to cover the full scope of human interaction. We need more simulations, taking into account more human aspects like cultural transmission and learning, as well as other games besides prisoner’s dilemma, to more fully understand why ethnocentrism evolves in human societies.

Be sure to check out Theory, Evolution, and Games later this week. We’ll show you how to evolve ethnocentrism on your own computer using the same code found in our paper.

References

Chase, I. D. (1980). Cooperative and noncooperative behavior in animals. The American Naturalist, 115, 827-857.

Dudley, S. A., & File, A. L. (2007). Kin recognition in an annual plant. Biology Letters, 3, 435-438.

Festinger, L., Schacter, S., & Back, K. W. (1963). Social pressures in informal groups: a study of human factors in housing. Stanford, CA: Stanford University.

Fogel, D. B. (1993). Evolving behaviors in the iterated prisoner’s dilemma. Evolutionary Computation, 1(1), 77-97.

Foley, R. (1995). The adaptive legacy of human evolution: A search for the environment of evolutionary adaptedness. Evolutionary Anthropology: Issues, News, and Reviews, 4(6), 194-203.

Gadgil, M., Joshi, N. V., & Gadgjil, S. (1983). On the moulding of population viscosity by natural selection. Journal of Theoretical Biology, 104(1), 21-24.

Hammond, R. A., & Axelrod, R. (2006b). The evolution of ethnocentrism. Journal of Conflict Resolution, 50, 926-936.

Hartshorn, M., Kaznatcheev, A., & Shultz, T. (2013). The Evolutionary Dominance of Ethnocentric Cooperation. Journal of Artificial Societies and Social Simulation, 16 (3)

Hipp, J. R., & Perrin, A. J. (2009). The simultaneous effect of social distance and physical distance on the formation of neighborhood ties. City & Community, 8(1), 5-12.

Katz, A. M., & Hill, R. (1958). Residential propinquity and marital selection: A review of theory, method, and fact. Marriage and Family Living, 20(1), 27-35.

Kubitschek, W. N., & Hallinan, M. T. (1998). Tracking and students’ friendships. Social Psychology Quarterly, 61(1), 1-15.

Lenski, R. E., & Velicer, G. J. (2000). Games microbes play. Selection 1, 1-3, 89-95.

LeVine, R. A., & Campbell, D. T. (1972). Ethnocentrism. New York: Wiley.

Nahemow, L., & Lawton, M. P. (1975). Similarity and propinquity in friendship formation. Journal of Personality and Social Psychology, 32(2), 205-213.

Runyon, J. B., Mescher, M. C., & De Moraes, C. M. (2006). Volatile chemical cues guide host location and host selection by parasitic plants. Science, 313, 1964-1967.

Seppä, P., & Pamilo, P. (1995). Gene flow and population viscosity in Myrmica ants. Heredity, 74, 200-209.

Velicer, G. J. (2003). Social strife in the microbial world. Trends in Microbiology, 11(7), 330-337.