In this proposal we aim to study how humans interacting in social and economic networks learn from their own experience and that of their social contacts. We are planning to conduct a number of studies all centring on the question of how (possibly non Bayesian) decision makers learn in networks and how they update their beliefs about the probability of different events if they receive new information from a neighbour in the network. This question has widespread applications for example to explain phenomena like herding, persuasion bias, the importance of airtime or the emergence of political polarization. In the first part of the project we use theoretical models, laboratory experiments, and field data to understand how people update their beliefs if they receive new information from a neighbour in the network. Standard theory would say that agents should rely on Bayesian updating in these situations. This may prove to be arbitrarily complex in a social network. We plan to run several laboratory experiments to study how information updating takes place in networks. Insights gained from our theory can help in the design of some of the experiments. Using a large data-set of 7 million Austrian households will allow us to test some of the predictions derived with our theory in the field. Finally we want to apply the insights gained to develop a flexible model of information updating in networks that works also if agents have only partial information about the network. In the second part of the project we will study how agents learn in social networks more generally. Since information processing is a key ingredient of any learning process, this second part will build on the first part of the study. We will rely mostly on laboratory experiments and possibly some theoretical modelling here.