We consider a Bayesian decision maker (DM) who, before making a decision, needs to allocate her limited attention across news sources with different biases. When choosing what news to read, the DM expects to receive some additional information in the future beyond her control. We show that the expectation of future information may affect the DM’s optimal learning decision in several ways. In particular, it can rationalize both the choice of news that reinforce or weaken ones prior (own and opposite-biased learning). The DM chooses own-biased learning when she is very certain of her action and, as long as the additional information is sufficiently powerful, opposite-biased learning when she is moderately certain. On the other hand, a very uncertain DM might want to make her choice of news dependent on the bias of the additional information. Applying our rational framework to study how expected future social interactions can impact peoples news consumption decisions, we show that people may (mis)coordinate the type of news they read with respect to their social group.