Geometry of ‘standoffs’ in lattice models of the spatial Prisoner’s Dilemma and Snowdrift games

The Prisoner’s Dilemma and Snowdrift games are the main theoretical constructs used to study the evolutionary dynamics of cooperation. In large, well-mixed populations, mean-field models predict a stable equilibrium abundance of all defectors in the Prisoner’s Dilemma and a stable mixed-equilibrium of cooperators and defectors in the Snowdrift game. In the spatial extensions of these games, which can greatly modify the fates of populations (including allowing cooperators to persist in the Prisoner’s Dilemma, for example), lattice models are typically used to represent space, individuals play only with their nearest neighbours, and strategy replacement is a function of the differences in payoffs between neighbours. Interestingly, certain values of the cost–benefit ratio of cooperation, coupled with particular spatial configurations of cooperators and defectors, can lead to ‘global standoffs’, a situation in which all cooperator–defector neighbours have identical payoffs, leading to the development of static spatial patterns. We start by investigating the conditions that can lead to ‘local standoffs’ (i.e., in which isolated pairs of neighbouring cooperators and defectors cannot overtake one another), and then use exhaustive searches of small square lattices ( 4 × 4 and 6 × 6 ) of degree k = 3 , k = 4 , and k = 6 , to show that two main types of global standoff patterns–‘periodic’ and ‘aperiodic’–are possible by tiling local standoffs across entire spatially structured populations. Of these two types, we argue that only aperiodic global standoffs are likely to be potentially attracting, i.e., capable of emerging spontaneously from non-standoff conditions. Finally, we use stochastic simulation models with comparatively large lattices ( 100 × 100 ) to show that global standoffs in the Prisoner’s Dilemma and Snowdrift games do indeed only (but not always) emerge under the conditions predicted by the small-lattice analysis.