An agent has to choose among a finite set of actions with payoffs depending on an unknown state of the world with two (possibly correlated) binary components, (θ1, θ2), θi∈{0,1}. Before making the choice, the agent can learn about the state by dynamically allocating his limited attention...