Institutional Incentives and Social Welfare in Social Dilemmas

A new study from arXiv (2605.31330) investigates whether institutional reward and punishment mechanisms can maximize social welfare—total population payoff minus institutional costs—in finite, well-mixed populations playing social dilemmas like the Donation Game and Public Goods Game. Existing research treats incentive design as a bi-objective problem: minimizing cost while achieving high cooperation frequency. This work develops a welfare-centric framework, deriving explicit expressions for expected social welfare under both rewards for cooperators and punishments for defectors. It characterizes how welfare depends on incentive efficiency and selection intensity, identifying parameter regimes where a single optimal incentive exists. The findings have implications for multi-agent and AI systems, as well as human societies, where autonomous agents interact under institutional rules.