Human being can be called “social animal”. There are no other animals that can communicate with other individuals in common with human. Many researchers are interested in cognitive and neural functions that exist behind advanced human sociality. Most of these researchers break down human sociality into individual cognitive components, such as mirror neuron system, and try to reveal the stand-alone function of these individual components. There are no doubts that these reductionist approaches are valuable for the realization of human sociality. However, we also feel the limitation of these reductionist approaches for the realization of human sociality. In this paper, we propose the new framework “social belief effect” for the realization of human sociality. We define social belief as a subjective belief on an interacting agent (e.g. the agent is a human individual or not). And “social belief effect” means the preparation of executive functions driven by a social belief. In our framework, various cognitive components are dynamically configured by a social belief and this top-down configuration enabled a dynamical social adaptation. To evaluate our framework, we review related previous studies and our behavioral and fMRI experiments for the direct measurement of “social belief effect”, and discuss its feasibility. And we try to model the neural mechanism of “social belief effect" from these previous experimental studies.