Cooperation Beyond Evolution?

Not just Selfish Genes, Mutualism, or Singer and Sidgwick's Group Altruism

In the previous post I wrote a lot about viruses “cooperating” - but viruses aren’t trying to cooperate. They just follow physical and chemical laws, and via evolution, those laws happen to lead to behaviors that resemble cooperation. They resemble it so much that the game theory which describes human behavior also almost perfectly describes the behavior of viruses. The “choices” they make aren’t intentional, and cannot be dictated by their non-existent brains. Neither are the responses of the bacteria, which are a bit more complex, but still lack some pretty basic requirements to have intentional behavior.

That doesn’t seem much like cooperation. But neither, I hope to explain, does selfless altruism.

Altruism and Mutualism aren’t Cooperation

There’s another framing for cooperative behavior, which is altruism. Peter Singer discusses this in his book, “Expanding Moral Circles,” where he talks about kin-altruism, reciprocal altruism, group altruism, and eventually, ethics. Reciprocal altruism and group altruism are one aspect of what I’d call cooperation - doing things knowing that it will help you as well. Kin altruism, though, is the narrower way that both animals and people will help their relatives, often even being willing to sacrifice their own lives to save those of their relatives, and it’s closely related to what we talked about for viruses.

Sociobiology has come a long way in explaining some of the effects mathematically, but Singer points out that kin-altruism can be understood directly via evolution. The math here is straightforward; if the benefit of a trait exceeds its cost, it will be beneficial. As W. D. Hamilton, among others, pointed out that it’s therefore possible for a trait that leads to self-sacrifice to be evolutionarily beneficial in those cases. No need for cooperation if we have a blind alien god to kill off whatever has the least beneficial strategy.

But one reason this is different than cooperation is that this all occurs at the level of the “Selfish Gene” - these aren’t being done in order to promote the general good, or to benefit others, they are genetic adaptations¹. If a species of bacteria evolves a mutually beneficial relationship with its host, it’s kin selection, not cooperation.

Similarly, it would be weird to say that humans are cooperating with their gut bacteria. But traditionally, biology differentiates between working with other organisms without appealing directly to evolution or altruism, and calls such positive interactions between members of the same species cooperation - though not in the sense I’d use - and calls the same sort of interaction between multiple species mutualism. 

You might remember the term “mutualism” from biology class. Perhaps you learned that there are three classes of long-term interactions between individuals of different species. In simple terms can be thought of as win-win, which is mutualism, win-lose, which is parasitism, and lose-lose, which typically leads to not interacting, since no-one wants this².

Parasitism is win-lose, where one organism benefits and the other loses out. Ticks and mosquitoes are human parasites, feeding on human blood, and the humans get nothing out of it. This is how most people think of games - there’s a winner and a loser, and people compete to be the winner, or at least try not to lose - almost the definition of a game.

But as noted above, there are some interactions, and some games, which you don’t want to play, because everyone is at least somewhat worse off. This last category is called interspecific competition. It’s “interspecific” because there are different species involved, whereas otherwise it’s just called competition. Most of the time, this is competition for resources - predators compete for a single type of prey, grazing animals compete for food, different birds compete for nesting locations, and humans compete with almost everything else to take land and build cities, for instance.

But humans don’t just compete - your interaction with your gut bacteria is mutualism, as is your interaction with your pets. But these aren’t really altruism either. Both types of terminology are unfortunate, I think, because they lose a key part of the difference between ecological interaction and direct interaction, and between intentional cooperative interaction and simple shared benefit. And none of this is altruism in the sense of giving up something you care about to benefit others without any return.

Shared Benefit

To explain why cooperation is different than altruism or mutualism, we need to talk about people. Humans have a complex internal mental model of not just their surroundings, but of the minds of those around them. So do many other mammals - dogs learn their owner’s quirks, and will treat different members of the family differently. And this points to something fundamental about what we mean by cooperation, and competition.

When two people decide to cooperate, they may be involved in something simple and one-time. When I walk to the store and buy bread and milk, I’m cooperating, in this case with the store owner. This type of cooperation is short-term and transactional, and it’s a fairly modern concept. On the other hand, they may be cooperating to build something together, due to shared interests. They might decide to work together to build a school or a church that everyone benefits from. This type of cooperation is longer-term, but still can be transactional. And when Singer talks about reciprocal altruism and human ethics, this is part of the picture.

But to move beyond reciprocal altruism, which I would call transactional cooperation, Singer needs to discuss altruistic motivation, and why people care about this. I think about it slightly differently than he does, less about why people want it and how we can justify ethics with biology, and more about how human relationships actually work. Because some relationships aren’t transactional. Friendships are built when people care about others and will help even when it doesn’t benefit themself. Romantic partnerships and marriages are (hopefully) the same type of mutually beneficial relationship. This requires some level of mental model of the other person - you act to benefit them. And yes, this can be understood as longer-term strategic mutual benefit, building alliances with others, but it isn’t usually experienced that way. And clearly, viruses and bacteria don’t have capacity for these types of relationships - ones that involve complex internal models of the other party, trying to do things that build relationships, and so on. But Singer follow’s Sidgwick’s hierarchy of benevolence, which tries to build an entirely justified mathematical framework for ethics based on utilitarianism. So Singer wants to claim that group altruism, and some form of enlightened self interest, leads to people cooperating in a Prisoner’s dilemma - and sometimes that is true. But sometimes it is false!

Cooperation isn’t Universal

Cooperation is everywhere, but it isn’t universal - it has sharp boundaries. In most of the examples of the natural world that were mentioned, cooperation was both with some group, and against some other group. Viruses cooperate with each other, against bacteria.

In other cases, cooperation is against something which isn’t playing along - but it’s still exclusive. Animals coordinate with each other to maximize their chances of survival, playing a cooperative game against nature. That might be cooperation to compete for resources, or for security, making it less likely that something will go wrong. But in either case, the cooperation is explicitly limited to the cooperating group - perhaps the herd of animals is cooperating to find water, helping the rest of the herd. 

Even in places where we might say no cooperation is occurring, it’s still a question of boundaries. A country might decide to eschew international norms and attack, but it’s a collection of people that are cooperating. A corporation might cheat their customers, or cheat workers out of wages, but the management is coordinating to do it. And even if a person decides not to cooperate with others, and lie, cheat, and steal their way through life, all along, his body is a collection of cells cooperating to do so. 

In each case, the failure to cooperate at a higher level is a lost opportunity, or even a tragedy, happening because of too-narrow cooperation. And this highlights a critical question that we’ll return to again and again in these essays - because the question is not whether to cooperate, but with whom. Though as we’ll see later, the nature of cooperation seems to create limits on the scale and types of cooperation which are possible.

Up Next?

So the next thing I need to talk about is what I think cooperation is, from a positive rather than a normative viewpoint - that is, what exists, not what should.

After that, if we want to follow the structure of a conceptual model for altruism, I could discuss more about how human cooperation works, and where. And I’ll definitely be writing about that! But before doing that, I plan to talk a bit more about the math of cooperation and stag hunts, in order to skip to some very different and interesting questions about the limits of cooperation, the problems of scaling, and cooperating with current and future AI.

1. And all of this depends on evolutionary stability, and has lots of interesting limits that relate to what is achievable with selfish cooperation - but that’s a discussion for later.

2. The last combination, lose-win, is just win-lose parasitism with the players reversed, so it’s not a different category.