Richard Dawkins and other proponents of "selfish gene" theories are faced with a problem. If organisms are merely ways for genes to propagate themselves, and if they must compete with all of their conspecifics for advantages in reproduction, why is it that so many animals cooperate with and even sacrifice themselves for others?

The answer they come up with is an elegant one. Organisms will forego their own reproduction, they explain, if doing so will aid someone else who carries some of the same genes. The point of living is to serve the genome's need to propagate itself, and the genome doesn't care if it gets propagated by you, your sister, or your second cousin. Altruism may make no sense from the individuals's perspective, but it serves the genome just fine.

Dawkins writes that if you were absolutely certain that someone else was genetically identical you would care as much for her as you would for yourself. This hardly sounds likely, but in less extreme forms the idea is widely accepted. The basic principle is known as Hamilton's rule, after a biologist of that name who formulated it back in 1964. Hamilton's rule states, loosely, that organisms are more likely to be altruistic the more closely they are related. (More formally, it states that "behavior increases in frequency when rb - c>0, where c is the fitness cost to the altruist, b is the fitness benefit to the beneficiary, and r is their genetic relatedness.")

A recent experiment from a group of Swiss researchers has received significant press coverage for its claims to have shown empirical proof for Hamilton's rule. It's an interesting experiment--but what may be most interesting about it is what it reveals about the hidden assumptions shared by the researchers and the enthusiastic science writers reporting on their work.

"Over hundreds of generations," the researchers concluded, "we show that Hamilton's rule always accurately predicts the minimum relatedness necessary for altruism to evolve."

But did they? Not by a long shot. The giveaway is how the researchers determined genetic relatedness. In their description of method they set out the structure of the neural net in each tiny robot and added, "The strength of these 33 connections was determined by 33 genes, whose values ranged from 0 to 255."

The unacknowledged assumption is concealed in this slide from genetics to the strength of the neural net's connections. The "genetic makeup" of the robots was nothing more than the character of their wiring. Since that wiring defined their behavior, the scientists--consciously or not--set up their experiment on the assumption that genetics completely determine behavior.

They must not have noticed this assumption--and it is an assumption, and a dubious one at that. If they had given it some thought they would have realized that they did not prove Hamilton's rule at all. Their experiment does not show that altruism is more likely with increasing genetic similarity. It showed that altruism is more likely with increasing behavioral similarity.

That's a very different thing. It's also highly plausible, and it's much easier to explain than any theory which would have organisms' being "aware" of genetic relatedness. In fact, running the experiment with computer-simulated robots makes the point even more clearly. Surely the robots can't calculate how closely they're related to other robots. What they "see" and respond to is how closely the other robots' responses are to their own. The better they mesh the more likely they are to cooperate and share.

What organisms do is the result of a complex dance among genetics, environment, upbringing, individual and social history, and many other factors. Their tendency to cooperate, coordinate, and sacrifice for others may very well increase when the other is not so Other--but otherness and similarity are multi-dimensional. Surely genetic similarity has something to do with similarity of behavior, but it's far from being the sole determinant.

The thinking behind the Swiss robot experiment, though, was that only genes matter, nothing else--and it "proved" this by the simple and unjustifiable tactic of making behavior and the genome identical by definition. The researchers basically assumed what they thought they were trying to prove.