Coevolutionary dynamics of cooperation and zero-trust institutional enforcement in Prisoner's Dilemma games

Institutional punishment is widely used to promote cooperation, yet most existing studies assume that institutions can perfectly identify individual strategies. Under this idealized assumption, cooperators are typically trusted and exempt from monitoring, while sanctions are imposed exclusively on known defectors, thereby neglecting the informational imperfections inherent in real-world governance. To address these limitations, we adopt the zero-trust principle of "never trust, always verify" and develop a zero-trust institutional framework, in which all individuals are subject to uniform monitoring, and sanctions are applied only upon the actual detection of defection. Given that institutional enforcement intensity in practice is rarely static, we further introduce enforcement intensity as an endogenous variable into the Prisoner's Dilemma game and construct a coevolutionary model in which the enforcement intensity and population state are mutually coupled through a closed feedback loop; that is, higher cooperation levels drive the growth of enforcement, while prevalence of defection leads to institutional relaxation. Under fixed and adaptive enforcement intensities, our theoretical and numerical results reveal distinct evolutionary outcomes characterized by different stable equilibria. Finally, by devising the institutional cost functions, we demonstrate that under fixed enforcement, zero-trust punishment outperforms non-zero-trust one in terms of long-run cost-efficiency at intermediate levels of enforcement. Under adaptive zero-trust enforcement, increasing enforcement responsiveness to population state reduces long-run costs when enforcement itself is relatively inexpensive, albeit with diminishing marginal returns, whereas excessive responsiveness increases overall governance costs when enforcement is expensive.