Being a hero by jumping on a grenade is a pretty terrible survival trait for an individual for instance, but essential for the group to have at least one in any decent sized population. All the non-grenade-coverers will strongly support such folks, as long as it doesn’t hurt their own survival chances somehow.
So if we look at individual tendencies as coming more from die rolls than anything else, with a wide distribution, a lot of outlier behavior makes a lot of sense.
A hardcore survivalist most of the time is going to be selected against, for instance, for many reasons when things are going well. They’re dumping all their stats points in the wrong categories!
But any population that doesn’t have at least a few is going to completely disappear on those rare long tail events (an actual nuclear war?).
And if those survivalists are actually capable, they get the benefit of ‘seeding’ the next generation without any competition! Long odds, but potential huge payoff biologically.
Also many of the traits might also only show up under extreme stress, and we’ve learned to frown on doing destructive a/b tests on entire cultures.
[0] one suggestion is the "gay uncle hypothesis" which posits that people who themselves do not have children may nonetheless increase the prevalence of their family's genes in future generations by providing resources (e.g., food, supervision, defense, shelter) to the offspring of their closest relatives; this hypothesis seems to be consistent with the evidence without being sufficient on its own.