Imperfect Imitation Can Enhance Cooperation

The promotion of cooperation on spatial lattices is an important issue in evolutionary game theory. This effect clearly depends on the update rule: it diminishes with stochastic imitative rules whereas it increases with unconditional imitation. To study the transition between both regimes, we propose a new evolutionary rule, which stochastically combines unconditional imitation with another imitative rule. We find that, surprinsingly, in many social dilemmas this rule yields higher cooperative levels than any of the two original ones. This nontrivial effect occurs because the basic rules induce a separation of timescales in the microscopic processes at cluster interfaces. The result is robust in the space of 2x2 symmetric games, on regular lattices and on scale-free networks.

💡 Research Summary

The paper investigates how the choice of update rule influences the emergence of cooperation on spatial structures in evolutionary game theory. Classical studies have shown that unconditional imitation (UI) strongly promotes cooperation by rapidly expanding cooperative clusters, whereas stochastic imitation rules such as the Fermi update tend to diminish cooperation because random fluctuations allow defectors to invade. To bridge these two extremes, the authors introduce a mixed imitation rule: at each evolutionary step a player adopts the UI strategy with probability p and a stochastic imitation rule with probability 1‑p. By varying p from 0 (pure stochastic) to 1 (pure UI) they explore a continuum of dynamics.

Extensive simulations are performed on the full parameter space of 2×2 symmetric games—including Prisoner’s Dilemma, Snowdrift, and Stag‑Hunt—using both regular two‑dimensional lattices and Barabási‑Albert scale‑free networks. The most striking result is that intermediate values of p (roughly 0.2–0.5) yield cooperation levels that exceed those obtained with either pure rule. This counter‑intuitive enhancement is explained through a “time‑scale separation” mechanism. UI acts quickly at cluster interfaces, converting neighboring defectors into cooperators and thereby flattening the boundary. Once the interface becomes smooth, UI alone provides no further growth. Stochastic imitation, on the other hand, continuously injects small perturbations (tiny “holes” or mixed sites) at the boundary. These perturbations create fresh opportunities for UI to act again, leading to a cyclic process of boundary roughening and smoothing that repeatedly expands cooperative domains. Consequently, the interior of clusters remains stable under UI, while the periphery is dynamically reshaped by stochastic fluctuations, producing a net increase in overall cooperation.

On regular lattices the phenomenon appears as cooperative islands that periodically develop and fill small gaps, whereas on scale‑free networks the highly connected hubs become the stable cores of cooperation (maintained by UI) and the low‑degree nodes around them are constantly re‑organized by stochastic updates. This hub‑core structure amplifies the spread of cooperative behavior across the network.

Importantly, the advantage of the mixed rule is robust across all examined games. Whether the payoff ordering follows the classic Prisoner’s Dilemma hierarchy (T > R > P > S), the Snowdrift (T > R > S > P) or the Stag‑Hunt (R > T > P > S), intermediate p values consistently produce the highest cooperation fractions. This universality indicates that the effect is not tied to a specific payoff matrix but stems from the intrinsic interplay between fast deterministic imitation and slower stochastic perturbations.

The authors conclude that “imperfect imitation” should not be viewed merely as noise that erodes cooperation. Instead, when combined judiciously with a deterministic rule, it creates a dynamic environment where cooperative clusters can continually renew their growth fronts. This insight reshapes our understanding of evolutionary dynamics on networks and suggests practical implications: in social, economic, or biological systems, allowing a modest amount of exploratory or error‑prone behavior alongside strong conformity may be a more effective strategy for fostering collective cooperation than enforcing either extreme alone.