Two attempts to explain
We cannot call simply upon the collective benefit for the species because this argument is not relevant enough facing the inconvenience for the individual to give without expecting in return.
Another attempt to explain: John Miller Smith approached the evolution by the theory of game, showing that something could be done (for example giving an information) in order to get something else fair's fair approach). This needs to speak to those we believe are able to play the game but also to a have a system for detecting cheaters (approach developed by W.D. Halmilton).
But the « Green-Beard theory" images the difficulty of altruistic people to recognize each other:
"Let's assume that altruistic peolpe wear a green beard to identify each other. The few selfish persons from the same species who also wears a green beard will be tempted to cheat... And will succeed once again to the detriment of altruistic people !".
Yet, some researchers of the Ecology Lab of the Pierre et Marie Curie University (ENS-CNRS) and from the Royal Holloway College (London, UK) have shown recently that altruistic people could be one step ahead than cheaters by "modifying slightly and regularly the color of their beards". Simulations showed that in this special case, altruistic people could have a competitive advantage on selfish people but also on selfish cheating people.
Nevertheless: the fair's fair approach, if it enables to understand some altruistic behaviors, is not working with language because we often talk to a set of people.