First, altruism must be defined so we can see how it defies the conventional theory of natural selection. Perhaps more importantly we must then ask why, when inexplicable using Darwin's original model for selection, we still frequently observe altruism in nature. Three answers to this question have been put forward: kin selection, group selection and reciprocation. These will be examined in order.
When an actor's behaviour increases the fitness of the behaviour's receiver at the expense of the actor, then the behaviour is said to be altruistic. This definition is simplistic but provides a good starting point; throughout the discussion it will be sharpened. In contrast to so-called selfish behaviour increases the actor's fitness while reducing the fitness of the receiver. If both the actor and the receiver gain in fitness following a behaviour, then the behaviour is cooperative. We must keep in mind that these costs and benefits are measures of reproductive success, the number of offspring surviving to reproductive age and thus the number of genes passed to the next generation.
It is worth mentioning that much altruistic behaviour may directly benefit those performing the behaviour in a way that may not be immediately obvious to the observer. Guard watches in meerkats may be an example of this. It was always assumed that meerkats guards were giving up valuable foraging time to watch for predators to the benefit of their group. However, it has recently been suggested that in fact they only take turns on guard when they are not hungry and thus they gain a significant head start in the escape if they are the first to spot the predator. They also tend to guard while positioned close to their bolt hole, giving them yet more of an increased chance of escaping the predator relative to the other members of their group. Rhesus monkey (Macaca mulatta) allomothers provide another such example: the behaviour that was originally thought to be purely altruistic is now thought to directly benefit the allomothers, as when it come to their own reproduction they have had experience in offspring care. These behaviours can therefore be classified as selfish.
Why a gene encoding truly altruistic behaviour can not be selected for seems intuitively obvious: by its own action it will decrease its frequency in the population by decreasing the fitness of those who carry it. It will also increase the fitness of other members of the population who may not carry the same gene. Why then do we still see so many animals committing suicide, giving up time, effort and reproduction for the sake of other individuals? So far the most convincing answer and certainly the one with most supporting evidence is kin selection.
Kin selection does not contradict the conventional natural selection but rather applies a filter to its perspective, changing it from a view of selection based on the level of the individual to one based on the level of the gene. It states that individuals gain indirect' fitness benefits, coined originally by Hamilton, through the survival and reproduction of nondescendent offspring. These can even be quantified using r, the coefficient of relatedness. r for a brother, for example, is 1/2, as brothers share on average...