Lu’s talk will provide a theory for identifying beliefs, information and state-dependent utilities over an objective state space. Lu assumes an agent is Bayesian and signals affect beliefs but not tastes. Under these two assumptions, an analyst who only observes the agent's pre-signal preferences and post-signal random choice can uniquely identify all of the following: (1) the agent's ex-ante prior, (2) the agent's signal structure and (3) the agent's state-dependent utilities.