Like family: Network coevolution and the emergence of altruism

ABSTRACT

Altruism is difficult to explain in an evolutionary context—the mechanisms of selection appear to favor selfishness. Existing scholarly work addressing this puzzle relies on structural and behavioral assumptions that severely limit generalizability. The model presented in this article circumvents the need for such assumptions by incorporating an evolving network component to natural selection. Through a process of coevolution of individual and relational traits, many of the exogenous assumptions of previous models of the evolution of cooperation are realized endogenously within a simulated population. Such endogeneity allows a more precise examination of conditions under which cooperation arises, but also a clearer understanding of how those conditions themselves arise. The model is specified analytically, and interpretation is carried out on simulated outcomes.

KEYWORDS:

• Altruism
• coevolution
• networks
• simulation

Notes

¹ For a population of size n and 2m possible values for C and K, the chain has possible states. Furthermore, the transition probabilities are highly dependent (inseparable). For example, letting n = 5 and m = 2, the size of the state space is approximately |X| = 9.3825 × 1012, and calculating the stationary distribution would require inverting an |X| × |X| matrix.

² There may be a slight tendency for the less cooperative cluster to be more heavily selected as primary parents, at least early on in the bifurcation.