Updating by averaging: a network perspective
摘要
Epistemologists seek a rule that agents can use to update their credence upon receiving evidence. While Conditionalization has been widely accepted as the standard norm for updating credence, it can be difficult to apply in situations involving opinion reports. This paper explores an alternative approach to updating credence through pooling functions. These functions, originally used for generating collective credence for a group, offer a promising method for agents to update credence with opinion reports. Although the formal properties of these functions have been extensively studied, few attempts have been made to explore the consequences of adopting them, especially how they shape epistemic communities. Using the Zollman effect as the primary test case, we can employ network models to examine the formal properties of each pooling function and their effects on an epistemic community, thereby gaining insights into the appropriate rule to adopt.