We incorporate learning in a standard dynamic stochastic general equilibrium model along two empirically-supported dimensions. First, we assume that agents cannot directly observe the individual components of the productivity shock and instead must conduct signal extraction exercises and update...