I actually wrote a program which showed behavior that appeared to be "social" after predictors were added to the simulation.

The postulation being organisms have a pressure (from predators) to form groups and eventually societies. Essentially, they'd either survive by evolving to be social, evolve defense mechanisms or die out. Social evolution may actually be the shortest path for non-aggressive species because they simply have to bare one another, as opposed to evolve long claws or something.

Very hard to prove, but our model showed given the options social interactions appeared more likely with basic reward circuitry.