Evolution of Cooperation on Spatially Embedded Networks

In this work we study the behavior of classical two-person, two-strategies evolutionary games on networks embedded in a Euclidean two-dimensional space with different kinds of degree distributions and topologies going from regular to random, and to scale-free ones. Using several imitative microscopic dynamics, we study the evolution of global cooperation on the above network classes and find that specific topologies having a hierarchical structure and an inhomogeneous degree distribution, such as Apollonian and grid-based networks, are very conducive to cooperation. Spatial scale-free networks are still good for cooperation but to a lesser degree. Both classes of networks enhance average cooperation in all games with respect to standard random geometric graphs and regular grids by shifting the boundaries between cooperative and defective regions. These findings might be useful in the design of interaction structures that maintain cooperation when the agents are constrained to live in physical two-dimensional space.

💡 Research Summary

This paper investigates how the spatial embedding of interaction networks influences the emergence of cooperation in classic two‑person, two‑strategy evolutionary games. The authors consider a variety of network topologies placed on a two‑dimensional Euclidean plane: regular square lattices, random geometric graphs (RGGs), spatially embedded scale‑free networks generated by a distance‑biased preferential attachment rule, and two hierarchical constructions – the Apollonian network and a grid‑based scale‑free variant. Each node represents an agent that repeatedly plays either the Prisoner’s Dilemma, Snowdrift, or Stag‑Hunt game with its immediate neighbours. The payoff matrix is parametrised by the standard T‑S plane, allowing a systematic sweep of game regimes.

Four imitation‑based update rules are employed to model strategy evolution: (i) replicator dynamics (random neighbour imitation proportional to payoff), (ii) unconditional imitation of the best‑performing neighbour, (iii) the Fermi rule (probabilistic adoption based on payoff differences), and (iv) a deterministic “best‑neighbor” rule. Simulations start from a 50 % random mix of cooperators and defectors and run for 10 000 generations; results are averaged over 30 independent runs for each (T,S) pair.

The main findings are threefold. First, networks that combine a heterogeneous degree distribution with high clustering and a clear hierarchical modularity – notably the Apollonian network – produce the highest overall cooperation levels across all three games. In these structures, high‑degree hub nodes act as cooperation “anchors” while dense local clustering protects cooperative clusters from invasion by defectors. Second, a grid‑based spatial scale‑free network, which retains the hierarchical hub‑spoke pattern but with slightly lower clustering, also outperforms plain lattices and RGGs, though its cooperation rates are modestly below those of the Apollonian case. Third, purely spatial networks with uniform degree (regular lattices) or low clustering (RGGs) display the poorest performance; the cooperative region in the T‑S plane shrinks dramatically, especially for the Prisoner’s Dilemma. Spatially embedded scale‑free networks without hierarchical organization sit in an intermediate position: degree heterogeneity helps, but the lack of strong local clustering limits their ability to sustain cooperation.

Phase‑diagram analysis shows that the boundary separating cooperative from defective regimes shifts markedly depending on the underlying topology. Hierarchical networks expand the cooperative region, allowing cooperation to persist even at higher temptation values (large T). This shift is attributed to two structural mechanisms: (a) assortative degree correlations that place high‑degree cooperators near each other, and (b) high clustering that creates tightly knit neighbourhoods where mutual cooperation yields a locally higher payoff than defection.

The authors discuss practical implications: in real‑world systems where agents are constrained to physical space—such as wireless sensor networks, robotic swarms, or urban social settings—designing interaction graphs with hierarchical hubs and strong local clustering can markedly improve collective outcomes. They acknowledge limitations, including the focus on static 2‑D networks, a limited set of update rules, and binary strategies, and suggest extensions toward dynamic topologies, multi‑strategy games, and asynchronous updating.

In conclusion, the study demonstrates that spatially embedded networks possessing both degree heterogeneity and hierarchical modularity are especially conducive to the evolution of cooperation. The Apollonian network emerges as a particularly effective template, offering a concrete design principle for fostering cooperative behaviour in physically constrained complex systems.