This document summarizes research on opinion formation and consensus building in social networks using Bayesian learning models. It presents three key models: 1) a voter dynamics model where individuals update their states based on neighbors, 2) an opinion formation model where individuals treat neighbors' states as evidence in a Bayesian updating process, and 3) a model allowing for innovation where a Dirichlet process prior enables the introduction of new states. Simulations show the impact of different prior probabilities on consensus formation and the role of priors in allowing for innovation and the emergence of new ideas.