Friendship is one of the phenomena that evolution psychology has long had in its sights. Charles Darwin himself pondered the adaptive advantage of sympathy in The Descent of Man, though it worried him it must come from a 'low motive', namely survival.
Game theory provided a big boost, the scenarios in which doing to as you have been done unto - tit-for-tat - suggested that it is in your self-interest to act as if in another's interest. The concept of reciprocal altruism was born, you scratch my back if I scratch yours. That there is still something of a low motive in this idea is mitigated, the evolutionary theorists say, by what's sometimes called the Beethoven Error: the maestro hardly lived the beautiful life his music reflects, though as that doesn't destroy the value of the music, so the self-interested origins of friendship doesn't reduce its value to us now.
Another element often referred to is the Banker's Paradox, which is the observation that often the people most in need of a loan are the worst credit risks. Similarly, in evolutionary history, the people most in need of friendship were presumably the ones least able to return it, being at the point of death or some such. So where is my self-interest benefited in showing them friendship, as empathic individuals will do? That must be accounted for. The answer proposed is a kind of excess of feeling: in small groups my welfare can be so caught up with yours that it leaves a legacy so that even when my welfare will be compromised by helping you, I will do so. Charity was born and the rest is history.
Bit the science of friendship doesn't add up. Two aspects reveal flaws, I believe. First, is the level of speculation involved: how can we possibly know that tit-for-tat operated on the Savannah. As the credit crunch revealed, economists can't predict what human beings will do when they're right in front of them. There are scant grounds for what the biologists would have us believe of our behaviour in the distant past.
Second, is that the accounts produced by evolutionary theory, even as they stand, don't seem to me to add up to friendship anyway. The fundamental problem is that they are all utilitarian - more or less fancy bits of cost-benefit analysis. But the closest friendships we have in life - the quintessential type, according to Aristotle, by which we recognise other kinds of relationship as friendly at all, and not merely contractual - are defined by being non-utilitarian at heart: loving someone for whom they are, not what they deliver. A friend who feels used ceases to be a friend, which is, of course, a world apart from the friend who is happy to be useful. It's that difference that the science fails to accommodate.
The Beethoven Error misses the point then: it's not the 'low motive' per se that's objectionable, it's that the logic of game theory won't get you from its utilitarian world to the non-utilitarian one that is so important for friendship. And I don't think the Banker's Paradox answer works either, seeing true friendship as a kind of overflow or excess of feeling. For one thing, putting a clearly defined phenomenon down to an excess is obfuscation not explanation. For another, it doesn't account for the non-fungibility of friendship: my sense of connection to my friend doesn't become so deep because our mutual welfare becomes so entwined, though it may do; it becomes so deep because it's you, the person who is my friend. No-one else will do.
There is another possibility: that emotions for others have moral worth because to work - that is to deliver the evolutionary advantage - we have to believe them to be true for their own sake. But that surely has all the moral integrity of the salesman, whose disingenuousness we spot a mile off. It's not the love we hope for called friendship.