It is a common tenet of simulation development that the entities that are being simulated must behave in a manner that is consistent with the manner of behavior exhibited by the actual entity. For non-sentient entities, which are subject to fundamental physical laws and constraints, their behavior in simulations is reasonably well understood. These behaving entities can frequently be used in multiple simulations, which offsets the cost of developing the physics based components. The behavior of sentient entities is not as clear-cut. While there are behavioral components that are reasoned, there are also aspects that are not reasoned, at least at a conscious level. Despite seemingly clear doctrine or training, human behavior continues to be a complex mix of emotional responses, reflexes and reasoning, all of which are influenced by numerous dependencies on the other entities and the surrounding conditions in the simulation. Consequently, good, realistic behavior models are typically expensive to build due to the emotional and knowledge acquisition efforts involved, which is then followed by the requisite knowledge modeling effort. Such models also tend to lack reusability in multiple simulations, due to their tailoring for specific environments. Simulation developers then face a dilemma: either expend resources or accept lower fidelity. This paper describes techniques being used to overcome this limitation by establishing knowledge re-use and collaboration methods. The approach consists of four parts: 1) Simplified knowledge acquisition, 2) Modeling emotions and knowledge in a standard formats; 3) Developing knowledge repositories that operate with the standard formats; 4) Modeling dependencies in standard formats. An example of the entity operation from knowledge based in a repository is presented, followed by a more general demonstration of simplified knowledge acquisition and knowledge repositories.