政府反恐措施与恐怖分子袭击手段存在着相互观察、适应性演化的过程。本文首先构建了政府作为先行动者、恐怖分子作为后行动者的斯塔克伯格博弈模型，进一步建立了恐怖袭击问题的扩展式演化博弈模型。由于扩展式演化博弈的均衡分析需要考虑代际内和代际间不同层次的信念学习，造成了演化分析的复杂性。本文将"心智模型"概念引入到演化博弈，采取了心智模型演化解的分析方法，得出了简化后的演化均衡解。最后，结合新疆墨玉县6.28暴恐事件进行算例分析，比较怀特流形演化解和心智模型演化解的差异。心智模型演化解的结果表明，根据恐怖分子群体中选择袭击比例是否高于临界值，政府策略收敛到防御或者不防御。该求解方法通过将普遍接受的社会规范引入到演化过程分析，不仅简化了怀特流形的分析过程，而且均衡解展示了更为丰富的、更为切合实际管理问题的演化特征。

There are interactive and adaptive evolution courses between the anti-terrorist measures of governments and the attack behaviors of terrorists. Moreover, for a terrorist event, governments and terrorists usually do not take the actions at the same time.
In the paper,Stackelberg game model is first built, where governments decide to defense strategies at the first and terrorists decide the attack strategies in the latter. And then, according to the above extensive-form factor game, the extensive-form evolutionary game model about the terrorism attack problem is built.
Because the equilibrium analysis of extensive-form evolutionary game model needs to consider the belief learning course at the inter-and intra-generation levels, it causes the complexity of evolutionary analysis. Although the conception of "Wright Manifold" that can be used to analyze the equilibrium solution of extensive-form evolutionary game model has been put forward, the method is difficult to understand and apply. However, if some conventional social norm are considered, the above analytical process can be simplified. Through introduceing the concept of "mental model" into evolutionary game theory, the new analytical method of evolutionary solution based mental model is proposed. For example, given terrorists take the attack strategy, the commonsensible behavior of any governmental forces should be the defense strategy. Thus, the utility functions of terrorists and governments are simplified. The replicator dynamic function of terrorists can be described as dy_i/dt=y_i·(1-y_j)·(-f-c₂)<0, where y_i is the proportion of terrorists taking attack strategy, t is the evolutionary period, and f and c₂ are the terrorists' payoffs. Similarly, the replicator dynamic function of governemnts can be described as dx_i/dt=x_i·(1-x_i)·(y_i·b-c₁), where x_i is the proportion of governments taking defense strategy, and b and c₁ are the governments' payoffs. The evolutionary solution based mental model is that terrorists take the no-attack strategy, and whether governments take the defense strategy depends on the terrorists' proportion of taking the attack strategy. Lastly, based on the 6.28 terrorism incident in Moyu County, Xinjiang, the paper compares two evolutionary solutions of Wright Manifold and Mental Model.
By introducing the universal accepted social norm into evolutionary analysis, the analytical method based on mental model does not only simplify the course of existed "Wright Manifold" evolutionary analysis method proposed by Cressman (2000), but also demonstrate the more complex and realistic evolutionary characteristics. The new evolutionary solution of mental model has the high application worth. Under the overall trend of the effective control terrorist event, the anti-terrorist measurements of government force should flexibly adjust with the proportional change of terrorists taking the attack strategy. Thus, there is no best or consistent anti-terrorist strategy (evolutionary equilibrium).