Promoting cooperation in the voluntary prisoner's dilemma game via reinforcement learning

Chaos An Interdisciplinary Journal of Nonlinear Science, Год журнала: 2025, Номер 35(4)

Опубликована: Апрель 1, 2025

Reinforcement learning technology has been empirically demonstrated to facilitate cooperation in game models. However, traditional research primarily focused on two-strategy frameworks (cooperation and defection), which inadequately captures the complexity of real-world scenarios. To address this limitation, we integrated Q-learning into prisoner's dilemma game, incorporating three strategies: cooperation, defection, going it alone. We defined each agent's state based number neighboring agents opting for included social payoff Q-table update process. Numerical simulations indicate that framework significantly enhances average as degree social-attention increases. This phenomenon occurs because enables individuals move beyond narrow self-interest consider broader benefits. Additionally, conducted a thorough analysis mechanisms underlying enhancement cooperation.

Язык: Английский

---

Network switching can improve system order

Chaos Solitons & Fractals, Год журнала: 2025, Номер 196, С. 116399 - 116399

Опубликована: Апрель 19, 2025

Язык: Английский

---

Social Touch as an Allostatic Tool in Cooperative Multi-Agent Reinforcement Learning

Опубликована: Апрель 25, 2025

Язык: Английский

---

The evolution of cooperation under falsified payoff information

Chaos Solitons & Fractals, Год журнала: 2025, Номер 197, С. 116463 - 116463

Опубликована: Май 2, 2025

Язык: Английский

---

Effects of value-driven social learning on cooperation in the prisoner’s dilemma games

Chaos An Interdisciplinary Journal of Nonlinear Science, Год журнала: 2024, Номер 34(12)

Опубликована: Дек. 1, 2024

Despite the growing attention and research on impact of Q-learning-based strategy updating evolution cooperation, joint role individual learners social in evolutionary games has seldom been considered. Here, we propose a value-driven learning model that incorporates shape parameter, β, to characterize degree radicalism or conservatism learning. Using prisoner’s dilemma game square lattice as paradigm, our simulation results show cooperation level non-trivial dependence density ρ, strength b. We find both β ρ have nonmonotonic effects cooperation; specifically, moderate levels can facilitate remarkably, when slightly conservative, form favorable region with appropriate ρ. Moreover, demonstrated play key formation network reciprocity, whereas dual support exploitation. Our reveal critical balance between maximize provide insights into understanding collective behavior multi-agent systems.

Язык: Английский

---

Constructive agents nullify the ability of destructive agents to foster cooperation in public goods games

Chaos An Interdisciplinary Journal of Nonlinear Science, Год журнала: 2024, Номер 34(10)

Опубликована: Окт. 1, 2024

Existing studies have revealed a paradoxical phenomenon in public goods games, wherein destructive agents, harming both cooperators and defectors, can unexpectedly bolster cooperation. Building upon this intriguing premise, our paper introduces novel concept: constructive which confer additional benefits to defectors. We investigate the impact of these agents on cooperation dynamics within framework games. Employing replicator dynamics, we find that unlike mere presence does not significantly alter defective equilibrium. However, when from are outweighed by damage inflicted addition affect ability sustain In scenario, be maintained through cyclic dominance between cooperators, with adding complexity but fundamentally changing Conversely, if surpass harm caused nullifies foster Our results highlight nuanced role emphasizing necessity carefully assessing incentive balances encouraging

Язык: Английский

---

An adaptive exploration mechanism for Q-learning in spatial public goods games

Chaos Solitons & Fractals, Год журнала: 2024, Номер 189, С. 115705 - 115705

Опубликована: Ноя. 7, 2024

Язык: Английский

---