Research on promoting cooperation among autonomous, self-regarding agents has often focused on the bi-objective optimization problem: minimizing the total incentive cost while maximising the frequency of cooperation. However, the optimal value of social welfare under such constraints remains largely unexplored. In this work, we hypothesise that achieving maximal social welfare is not guaranteed by the minimal incentive cost required to drive agents to a desired cooperative state. To address this gap, we adopt to a single-objective approach focused on maximising social welfare, building upon foundational evolutionary game theory models that examined cost efficiency in finite populations, in both well-mixed and structured population settings. Our analytical model and agent-based simulations show how different interference strategies, including rewarding local versus global behavioural patterns, affect social welfare and dynamics of cooperation. Our results reveal a significant gap in the per-individual incentive cost between optimising for pure cost efficiency or cooperation frequency and optimising for maximal social welfare. Overall, our findings indicate that incentive design, policy, and benchmarking in multi-agent systems and human societies should prioritise welfare-centric objectives over proxy targets of cost or cooperation frequency.
Deep Dive into 사회복지를 최우선으로 하는 인센티브 설계 비용 효율성 및 협력 빈도.
Research on promoting cooperation among autonomous, self-regarding agents has often focused on the bi-objective optimization problem: minimizing the total incentive cost while maximising the frequency of cooperation. However, the optimal value of social welfare under such constraints remains largely unexplored. In this work, we hypothesise that achieving maximal social welfare is not guaranteed by the minimal incentive cost required to drive agents to a desired cooperative state. To address this gap, we adopt to a single-objective approach focused on maximising social welfare, building upon foundational evolutionary game theory models that examined cost efficiency in finite populations, in both well-mixed and structured population settings. Our analytical model and agent-based simulations show how different interference strategies, including rewarding local versus global behavioural patterns, affect social welfare and dynamics of cooperation. Our results reveal a significant gap in the
The evolution of cooperation has long been a central puzzle in evolutionary biology, social sciences and multi-agent systems [1,2,3,4,5]. While classical evolutionary theory emphasizes the survival of the fittest-which in many strategic settings (e.g., the Prisoner's Dilemma) often corresponds to more selfish behaviour-cooperative behaviour is nevertheless pervasive among both humans and animals. This apparent contradiction has motivated extensive research into the mechanisms that enable and explain how cooperation emerges and persists in social dilemmas [6].
To address this puzzle of cooperation [2,7], numerous mechanisms have been proposed to account for the emergence of cooperation, including kin and group selection [8,9], direct and indirect reciprocity [10,11], reward and punishment (incentives) [12,13] and structured populations [14]. A particularly important mechanism examined in recent years involves an external institution that seeks to steer the population towards greater cooperation by selectively investing in individuals [13,15,16,17]. For example, such an institution may provide rewards to cooperators, either at the global population level or in a more localized, neighbourhood-based manner. However, existing research often overlooks social welfare and fails to address the delicate balance between fostering cooperation and optimizing social welfare.
In well-mixed populations, Han and Tran-Thanh (2018) [18] showed that a decision-maker can conditionally reward cooperators based on population composition to guarantee a desired cooperation level, Figure 1: Overall approach. By exploring the use of Social Welfare as a metric of optimisation for the evolutionary model of cooperation under the effect of institutional incentives, this work aims to answer the differentiation between cost optimisation and social welfare optimisation both theoretically and empirically. while minimizing interference cost. Their framework formulates a bi-objective optimisation problem: maximize cooperation frequency while minimizing institutional investment. Later, Duong and Han (2021) [19] provided a rigorous stochastic analysis of institutional incentives, characterizing optimal reward/punishment schemes under different selection intensities and identifying sharp phase transition phenomena in cost efficiency.
However, real populations are seldom well-mixed. Interaction patterns are shaped by spatial or network structures, which can fundamentally alter evolutionary outcomes [14,2,20]. Han et al. (2018) [21] examined external interference in structured populations using agent-based simulations on square lattices (which was then extended to other networks and game-theoretic interactions [22,23]). Their results indicate that local interference strategies, which monitor neighbourhood-level information, can be significantly more cost-efficient than global ones, underscoring the importance of spatial heterogeneity when designing incentive mechanisms. Despite these advances, prior works mainly focus on either (i) maximising cooperation prevalence or (ii) minimizing the institutional cost [19,24]. What is missing is a direct optimisation of social welfare, defined as: Social Welfare = (Total payoff of the population) -(External cost EC).
From a societal or institutional perspective, social welfare is ultimately the most meaningful objective: cooperation is valuable insofar as it generates net benefits relative to the cost of enforcing it [25,26,27,28]. This paper introduces social welfare maximization into the study of cost-efficient external interference. We address the following research questions:
• RQ1: How does optimizing social welfare change the optimal interference strategies in well-mixed populations? Does welfare maximization require less, more, or differently patterned investment compared to minimizing cost alone?
• RQ2: In structured populations, do previously identified cost-efficient local strategies remain superior under the welfare objective? Does spatial structure amplify or diminish the welfare benefits of conditional interference?
Our key contributions are summarized as follows:
• We incorporate social welfare into the analytical framework of institutional incentives in well-mixed populations, extending the interference scheme of Han & Tran-Thanh (2018) [18] and the costefficiency analysis of Duong & Han (2021) [19]. This allows us to examine how welfare maximization reshapes optimal intervention strategies.
• We evaluate interference in spatially structured populations on grids using agent-based simulations, extending Han et al. (2018). We compare global and local interference strategies not only in terms of cooperation frequency or cost efficiency, but through the unified lens of social welfare.
Our overall approach can be summarized in Figure 1. By introducing social welfare into evolutionary models of institutional incentives, this work offers a more realistic and policy-relevant framework for unde
…(Full text truncated)…
This content is AI-processed based on ArXiv data.
