Altruism tested

Three Mile Island

In 1979, I was in graduate school in Philadelphia, a city in which I had been living for seven years. As is the case with most people of university age, I had a cadre of close friends for whom I would give the world. Or at least so I thought.

As it happens, I also had a cousin who lived in Philadelphia, with whom I had had a sometimes frosty relationship. Despite the fact that we lived within 10 blocks of each other, we saw each other no more than a couple of times a year, events that I believe we both approached with equal amounts of dread and delight.

In March of that year, a nuclear reactor at Three Mile Island, less than 100 miles from where I was, experienced a partial meltdown. There were a few tense days. I remember one morning in particular being sufficiently charged with anxiety that I packed my car with some provisions, planning to leave town if things did not improve soon. The question was, who was I going to take with me, as none of my close friends had cars of their own.

Actually, I didn’t even ask the question. I knew. I called my cousin and told her that I was leaving town that afternoon, and that if she wished, she (and her husband) could come with me. In the end, things calmed down just before we were set to leave, and Three Mile Island became a memorable blip in the history of the nuclear power industry. But for me, it taught me a deep lesson about the power of genetic relatedness and altruistic behaviour, one which has now been formally tested.

A cornerstone of evolutionary psychology is that altruistic behaviour is based upon genetic relatedness. Made famous by Richard Dawkins in his 1976 book The Selfish Gene, the theory was given mathematical rigour in 1964 by W.D. Hamilton who formulated an equation which described the relationship between costs, benefits, and relatedness: whenever the benefit of an action times the degree of relatedness (1 for identical twins, 0.5 for siblings, etc.) is greater than the cost, altruistic behaviour makes sense. This became known as Hamilton’s Rule, and has been a dominant hypothesis in the field ever since it was published in the Journal of Theoretical Biology 47 years ago. The problem has been that Hamilton’s Rule, like so much of evolutionary psychology, was notoriously difficult to test. Until now.

In an ingenious set of experiments, a group of engineers from Lausanne designed robots that had several key features. One was that they could either act selfishly or unselfishly with respect to food that they could forage in their environment. Another was that their behaviour was controlled by a neural net which was subject to random modification with each generation: essentially, the neural net could evolve. Finally, the robots could also share their evolutionary accomplishments – essentially, they could produce hybrids which were a bit like one and a bit like the other (just like progeny in evolution). They then allowed the robots to do whatever they were going to do. What emerged was, remarkably, full confirmation of Hamilton’s Rule: over 500 generations, the robots became much better at gathering food, and the degree of cooperation bore strict fidelity to Hamilton’s mathematics – kin altruism appears to be real.

There are many reasons to laud the results, not the least of which is the elegance with which a long-standing hypothesis was tested. But one that also stands out is that the experiments demonstrate that the ‘just so’ stories of evolutionary psychology can become real experiments, tested with rigour and quantitative methods. It may not be the case that every theory in evolutionary psychology will be amenable to such experimental manipulation, but given the cleverness of the human mind, pretty much anything is possible.

You can watch all the fun on this YouTube video.

Empirical Zeal also had a great blog post about these experiments, which is well worth reading.

Image credit: The National Archives via Wired