Objective Excessive activation of defence modules leads to some dysfunctional outcomes, which can be broadly classified to defence activation disorders. Defence activation disorders have high mortality, low fertility, high prevalence and high heritability. In this study, agent-based simulation model is formulated for solving this evolutionary paradox.Methods The emotional system is considered as a superordinate cognitive module for grasping the average resource amount and the average diminishing returns of resources, based on the Marginal Value Theorem. Under the assumption, the evolutionary ecological model was proposed and analysed.Results Individuals utilising suboptimal strategies can be stably maintained in agent-based evolutionary simulation environments. Individuals were adapted to have different d-values according to the local niche. The simulation runs stably within the calibrated range of the variables for a long time. Agents establish locally optimal strategies based on their given d-values, and the relative proportion of subpopulation maintained stably in the heterogeneous habitat with the resource gradient.Conclusion This study verifies the evolutionary mechanism of defence activation disorders in computer-simulated environments by using agent-based modelling with the Marginal Value Theorem. Balancing selection appears to be a plausible evolutionary mechanism that makes the suboptimal levels of defence activation the evolutionarily stable strategies.