Adaptive target regulation for multi-agent cooperation in collective-risk dilemmas

Hua, Shijia and Liu, Linjie and Szolnoki, Attila (2026) Adaptive target regulation for multi-agent cooperation in collective-risk dilemmas. EXPERT SYSTEMS WITH APPLICATIONS, 324. ISSN 0957-4174 (In Press)

	Text eswa26_ms.pdf - Submitted Version Restricted to Registered users only until 1 September 2026. Download (17MB) | Request a copy
	Text 1-s2_0-S0957417426013849-main.pdf - Published Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Multi-agent systems frequently face critical challenges such as collaborative consistency and collective goal alignment that require large-scale cooperation for effective resolution. The collective-risk social dilemma game offers a theoretical framework for studying situations in which the fulfillment of the expected target determines the collective success. In contrast to traditional models, we assume a dynamical approach in which the target level may be raised, kept, or lowered in a repeated game model in the light of latest collective achievement. Our results reveal a close connection between the adjustment of collective targets and the environmental risks encountered. At low risk, the best protocol is to lower the target level when the first-stage target is not achieved. At medium risk, the optimal policy is to raise the target level when the first-stage target is met, but avoid target reduction otherwise. In high-risk scenarios, raising the target level when the first-stage target is achieved proves to be the most effective way to enforce cooperation. Importantly, adjusting the target at an early stage is more favorable for the emergence of cooperation.

Actions (login required)

Edit Item