許多企業的經理人一直在尋求有效的政策以促進員工彼此分享知識。在組織中實施激勵知識分享的政策，因為存在著許多的個人與組織的因素，兩者相互作用會影響組織的成員知識分享的行為，政策的效果很難掌握。在實務上，不同的組織激勵的政策或手段的效果，欲通過田野調查或實驗設計來評估政策所帶來效果及對組織成員行為的影響不容易實現。因此，本研究嘗試運用一項新穎的研究策略-代理人基塑模(agent-based modeling)，經由建構一個人造的世界來模擬組織成員知識分享的行為。在這個人造的社會中，研究者考慮了知識分享的報酬、組織成員的行為策略、行為策略的學習與適應機制、不同組織的群體能力水準、互動的網路、知識的選擇模式與不同的組織激勵政策設計等變數，藉以設計相關的實驗。模擬的結果產生了的幾項有趣的發現： (1) 當分享知識的報酬愈高時，代理人知識分享行為會愈多，無論是在那一種的互動網路、群體能力或分享知識的選擇模式的情況之下。(2) 代理人的互動網路是一項重要影響因素，互動網路扮演著知識流通與行為策略學習的管道，它同時會影響個體知識分享的報酬與行為策略的學習。它可能促成不分享的策略的擴散，可能會提升組織激勵的效果，也可能會增強激勵所帶來的副作用。(3) 個體的知識分享與吸收的能力如果存有差異，則能力較好的代理人將會局部地吸引能力較差的互動對象採用他的策略，即使他所用的不是可以獲致最佳報酬策略。(4) 為促進組織成員分享知識，定期審視固定獎勵的作法可以導致比較好的效果，不但可以提升分享知識的行為，也使代理人比較願意採用傾向分享知識的策略，促成組織信任的氣氛。但是，在某些情況下，可能造成反效果。(5) 最後，根據實驗的結果與研究的發現，研究者建構了一個知識分享之組織行為模型，以做為後續實證研究之參考架構。 / Mangers always look for effective policies to prompt knowledge sharing between members in an organization. It is difficult to evaluate the effect of incentive policies for stimulating knowledge sharing because there are many individual and organizational factors. In practice, it is not easy to assess the effects of different incentive policies or methods by the methods of experiment or field investigation. Therefore, a novel research strategy is applied in this study, which is called agent-based modeling. An artificial world was constructed to simulate the knowledge sharing interactions between members in an organization. This study considers some parameters including the payoff of knowledge sharing, the strategies of members, the learning and adaption mechanism of strategies, collective capabilities, interactive network, the selection methods of sharing knowledge and incentive policies to design experiments in the agent-based model. The results of simulations produced some interesting findings: (1) the higher the payoff of sharing knowledge, the more the actions of sharing knowledge is in spite of any kind of interactive networks, collective capabilities, and the selection methods of sharing knowledge. (2) Interactive Network of agents is an important factor, which plays a role of channel of knowledge transition and strategy learning. It simultaneously affects the payoff of knowledge sharing and learning of strategy. It maybe results in the diffusion of strategy of not sharing knowledge, or enhances the effect and side effect brought by incentive policies. (3) Because of difference between agents’ capabilities, agents with better capabilities will locally attract the ones with worse capabilities to learn their strategies, which even are not the best. (4) To enable sharing knowledge between members in an organization, periodic reward will get better results. It does not only increase the action of sharing knowledge, but also make agents to adop the strategies trending toward sharing knowledge. Periodic reward is helpful to form a trustful organization climate. However, in some circumstances, it may get minus effects. (5) Finally, according to experimental results and research findings, an organization behavior model of knowledge sharing has been constructed for the empirical studies in the future.