In the next week or so I'll publish a piece describing a tiny act of altruism I performed for someone I'd never met before—and did not expect to meet again.  This happens to us all the time.  The fact is that people are nicer than they should be—at least according to the dictates of evolution.  When you meet someone only a single time, and that person is nice to you, you tend to be nice to them back. This is a frequent observation in laboratory studies of human behavior. 

As I said, that behavior contradicts what we’d expect evolution to produce.  If you have a one-shot encounter with someone, and that person is nice to you, the best evolutionary strategy would be to defect—to not return the benefit.  Since you won’t meet the person again, you have nothing to lose by not reciprocating, and you don't incur the cost of doing something nice (i.e., sacrificing something).  But people in games—and in life—tend to reciprocate in these situations, even when they’re sure that they’ll never meet that person again.  Why do we do this? Why, for example, do we leave big tips even in a restaurant we know we’ll never visit again?

There are two explanations for  such one-shot generosity.  One is cultural: we have simply reasoned our way into extending the generosity and help we have evolved to bestow on our kin and groupmates to other people we don’t even know.  This is the thesis of Peter Singer in The Expanding Circle. He suggests that rationality tells each of us that we hold no special moral status over the rest of humanity, and so we should help others, even at some cost to ourselves, when we don’t know them.  It’s the Golden Rule, but derived through secular reason.

Another explanation is that human generosity in one-shot situations is simply a byproduct of instincts that have evolved for other, more adaptive, reasons. That is the thesis of a new paper in the Proceedings of the National Academy of Sciences by Andrew Delton, Max Krasnow, Leda Cosmides, and John Tooby.  What they show though evolutionary modeling is that humans can evolve to be generous to strangers in “one-shot” situation because they never really know whether a one-shot situation could actually be the beginning of multiple encounters with that same person, encounters in which you can each reciprocate, enhancing your mutual fitness in a big way.

It turns out that in many situations, the small price you pay in being nice to a person on an initial encounter is more than compensated for by the greater benefits you’ll get if, by being nice initially, you establish a reciprocal and mutually beneficial relationship that pays off in subsequent encounters.  (If you defect on the first encounter, that person won’t be inclined to be nice to you in the future.) And since one-off meetings might have been relatively rare during most of human evolution, when we lived in small groups, we might have evolved the tendency to be nice even when we meet somebody for the first time—explaining at least part of our current generosity.

The paper of Delton et al. involves making evolutionary simulations of behavior that include many pairs of people interacting in either one-shot or multiple situations, with each person deciding whether or not to be generous or to defect. The payoffs to each person constitute his evolutionary fitness, and determines the frequency with which his behavior is seen in the next generation (defecting and generosity are assumed to be genetic traits). They show that under many situations individuals will evolve to reciprocate generosity to another person, even when they think they’ll never see him again.  This is based on two assumptions:

1. The small cost you incur by being generous to someone else in a one-shot situation can be more than compensated for by the larger benefits of a reciprocal and mutually beneficial relationship that forms if you happen to meet that individual again.
2. Critically, over most of human evolution we didn’t know for sure whether a one-shot encounter might turn into a multiple-shot relationship.  As Delton et al. note:

Using agent-based simulations, we show that a propensity to make contributions in one-shot games (even those without reputational consequences) evolves as a consequence of including in the architecture an overlooked computational step necessary for guiding cooperative decisions: the discrimination of one-shot from repeated interactions.

Imperfect discrimination is the biologically realistic case because real computational systems, such as human minds, cannot know with certainty whether an interaction is one-shot at the time the decision about cooperating must be made. Indeed, given the stochastic nature of the world, it might be correct to say that, at the time of the interaction, the interaction is not determinately either one-shot or repeated.

The authors simulated 500 pairs of individuals who interacted under a variety of situations (one-shot, repeated encounters, different payoffs, etc.), leaving descendants based on the fitnesses realized from their interactions. This evolutionary game extended over 10,000 generations.  They also made two types of models.

Model 1. Cognitive component fixed, motivational component allowed to evolve.  That is, the actor’s belief that an interaction was either one-shot or repeated was a fixed and unchanging value, and what could evolve was the probability of cooperating in a given interaction.

Model 2. Motivational component fixed, cognitive component allowed to evolve.  That is, actors would never cooperate if they thought an interaction was one-shot, and would always cooperate if they thought an interaction would be repeated—these behaviors were fixed and not allowed to evolve.  What was allowed to evolve was the weight of evidence required to convince an actor whether an interaction was likely to be one-shot or repeated.

The results were pretty much the same in both cases: if interactions between a given pair were frequent enough—somewhere between two and five encounters—the motivational component would evolve in the first case: individuals in populations would become more likely to cooperate even if they thought an interaction was a one-shot deal.

Likewise, under the same frequency of repeated interactions, the cognitive component evolved in case 2: as the authors note, “Cognitive architectures that are highly resistant to concluding that interactions are one-shot are favored by selection over architectures that are cognitively more accurate.”  In other words, people’s minds evolved to deceive themselves about the reality of interactions, for that type of self-deception yields a higher long-term payoff than would more accurate assessments.

The big conclusion, then, is that we can evolve to be cooperative with someone even when we think we’ll meet them only once, for we’re never certain that it’s going to be just a single meeting, and we’d best be nice in case it isn’t.  It’s a sort of Pascal’s wager on interactions.  As the authors put it:

In short, the conditions that promote the evolution of reciprocity—numerous repeat interactions and high-benefit exchanges—tend to promote one-shot generosity as well. Consequently, one-shot generosity should commonly coevolve with reciprocity. This statement is not a claim that direct reciprocity is the only force shaping human cooperation—only that if reciprocity is selected for (as it obviously was in humans), its existence casts a halo of generosity across a broad variety of circumstances.

Note that this paper explains one-shot generosity in modern societies as the result of evolution that occurred in the past, over millions of years when we lived in small groups.  The situation we live in today, when we can be absolutely certain that we’ll never meet someone again (as when we’re travelling), has not been in place long enough to alter our evolved behaviors.

The authors also note that one-shot generosity is even more likely to evolve if other people observe your behavior, for such observations can give you a reputation as either a cheater or a generous person—a reputation that can spread and affect your future interactions with a lot of other people.

It has not escaped my notice, though the authors don’t mention this, that a similar mechanism could generate an initial “altruistic” act when you meet a stranger in a presumed one-shot interaction.  The price you might pay if the other person doesn’t reciprocate might be more than compensated by that person’s reciprocation, which could then trigger a beneficial longer-term relationship.

All of us, I bet, have both performed and been the recipient of such altruistic acts.  This morning I went to the Finland Station, where the locomotive on which Lenin entered Russia during the revolution was reputed to be housed.  When I asked around about it, nobody spoke English, and everyone was baffled at my inquiry, even when I drew a locomotive with a little Lenin guy on it.  But a student who spoke broken English came up to me and, after a long and difficult conversation, he realized what I wanted. He then took me around the station to speak to various officials so I could be allowed onto the train platform, where the locomotive indeed resided.  They finally let me out there, and I saw and photographed the locomotive (pictures will come soon).  What a nice person he was, and of course he got nothing for his actions save my fervent thanks and a handshake.  Perhaps those acts of kindness, too, reside partly in our genes. (Now this isn’t really true altruism, in the sense that the donor gets nothing back, but it’s “altruism” that may have evolved in our ancestors when you really did expect, on average, to benefit from the behavior.)

This isn’t really my area of expertise, but I think the paper is an important one. It shows that under the right conditions a previously inexplicable type of human behavior—at least inexplicable from an evolutionary standpoint—can evolve as an inevitable byproduct of human interactions.

There are two potential problems with this paper.  The first, and less serious, is that we realize one-shot generosity as an evolutionary byproduct only if interactions between pairs of individuals are frequent: five is a good estimate.  That’s not so unrealistic, though, since for millions of years our ancestors probably lived in smallish groups. Over most of our evolutionary history, repeated interactions with others were probably the norm.

The other problem is genetic: the authors’ model is asexual, that is, each behavior or cognition is simply transmitted as a whole from one generation to the next.  This simplification makes the model tractable, but is rather unrealistic.  Behaviors are not all-or-none traits produced by single genes, but are usually based on several to many genes that can recombine with the genes of mating partners, producing intermediate behaviors.  It’s not clear whether a more realistic genetic model would produce results similar to that of Delton et al. I suspect it would, but that’s only a guess.

Nevertheless, and though I haven’t been a huge booster of the Toobey/Cosmides brand of evolutionary psychology, I see this as a pretty important paper.  It explains, in a simple and comprehensible way, how a puzzling human behavior might evolve.  Of course, as the authors note above, reciprocity isn’t the only force shaping human generosity, and cultural evolution remains as an alternative or supplementary explanation.

But I still leave larger tips in bars and restaurants when I know I’ll be going back. That’s simply me overcoming the tyranny of my unselfish genes.

h/t: Alex Lickerman

________

Delton, A. W., M. M. Krasnow, L. Cosmides, and J. Tooby. 2011. Evolution of direct reciprocity under uncertainty can explain human generosity in one-shot encounters. Proc. Nat. Acad. Sci. USA (early edition): doi:10.1073/pnas.1102131108