Red, theory; black, fact
How does everyday human behavior eventually accomplish the biological function of dispersal for the human race?
Dispersal is things like dandelions shedding airborne seeds, slime molds developing into spore cases on stalks and releasing the spores into the wind, territorial systems of birds and mammals forcing the unlanded young to seek widely for their own territories, and humans going into space because our science fiction writers keep scaring us about the possibility of meteor crashes wiping out life on Earth.
The slime mold Dictyostelium is triggered into its dispersal program by the food supply running short; I will adopt the assumption that the human dispersal program is also triggered by the end of the good times, that is, the price of bread rising relative to wages.
Human neural pathways may potentiate aggression when the hard times come, but of an elaborate kind adapted for ensuring efficient dispersal (i.e., with minimal loss of life). It begins with a two-person feud of the sort illustrated in cultural references too numerous to mention. An arbitrary stimulus, made offensive by some piece of Pavlovian conditioning, is traded back and forth with rapidly increasing energy.
The emotional component is strongly threatening because the participants must be induced to seek allies, which people do when threatened, until all of society is eventually polarized. The acts of provocation being traded back and forth become progressively more outrageous, as they must, to keep the polarization process going. Eventually, one side gets the upper hand and forces the other to flee.
The result is a diaspora, i.e., dispersal. Because of the long polarization process, an entire group is expelled, not single individuals one at a time. Thus, members of such a group can assist each other to survive and relocate, thereby reducing the mortality associated with dispersal, thereby making the dispersal event more efficient in terms of number of people relocated. The group who flees is then seen by the international community as the blameless victim, and the group who stays is seen as the unprincipled aggressor. This tends to elicit a sheltering of the refugees and an intimidation of the "aggressor," who is deterred from pressing his advantage, that is, pursuing the refugees and slaughtering them to the last man, which is what each side would like to do to the other by this point. This, again, is an efficiency from the point of view of producing dispersal.
However, if each side is continually threatening the other, why don't they flee each other's presence during the very early stages? Humans may have a reflex that converts feeling threatened into a wish to injure the threatening party, possibly a behavioral leftover from some earlier adaptation, such as an anti-predation defense; to injure, you have to stick around.
Finally, settled refugees usually do not integrate completely into the host society, instead forming ethnic neighborhoods. Being seen as ethnic by the host society, due to slow integration, could improve the reproductive success of refugees because of disassortative mate-choice effects evolved to favor genes that produce dispersal.
The dispersal-producing dynamic just outlined is powerful, because it must overcome all the reasons a person would not leave their homeland forever at some arbitrary time: expense, risk of mortality in transit, opportunity costs, temporary loss of livelihood, need to learn a new language and customs, vulnerability to exploitation in the new country, etc.