I think the common notion is that the survival of the individual is exactly what natural selection is selecting for.
However, it seems to me its more selecting for its own particular configuration of genes, that essentially select for a signature of themself.
Within a species the configuration between individuals is generally different enough to not ellicit altruistic behaviour.
In most animals, the 50% kinship gene relatedness threshold is definitely not enough to instigate self sacrificial behaviour. The children of a usocial species have to be like 11/16ths clones of eachother. And the runt suicide mechanism that triggers in rabbits requires a brood of true clones.
The reason I think it seems the genes select for individuals is because most circumstances will recquire that the current reproductive unit at the individual level needs to stay alive long enough for a net positive replication contribution to have been made.