Researchers have discovered a mechanism by which cooperation can emerge and be sustained in evolutionary games, even in scenarios where classical theory predicts selfishness. The study, published in Nature Communications, demonstrates that introducing a "nudge" into agents' decision-making processes, combined with adaptive learning, allows cooperative strategies to spread and dominate in dynamic populations. This finding challenges the traditional view that cooperation is inherently fragile and requires very specific conditions for its persistence.

The "nudge" consists of a small external influence that slightly biases an agent's choice towards cooperation, without forcing it. Agents, in turn, adjust their strategies based on the outcomes of previous interactions (adaptive learning). Through computational simulations of games like the prisoner's dilemma, scientists observed that this combination of "nudge" and learning generates a positive feedback loop. Initially, the "nudge" increases the probability of cooperation, leading to better outcomes for cooperators. Adaptively learning agents imitate these successful strategies, reinforcing cooperation in the population.

The results show that cooperation can reach significant and stable levels across a wide range of game parameters, surpassing predictions from models without "nudge" or without learning. This mechanism offers a new perspective on the evolution of cooperation in complex systems, from biology to economics and social interactions. It suggests that small interventions or cognitive biases can have a profound impact on collective behavior, facilitating the emergence of beneficial outcomes for the group. The research opens avenues for exploring how to design environments that foster cooperation in real-world contexts.