Behind every wireless system lies a theoretical goal and an immense amount of work to achieve this goal. As the technology behind wireless communication becomes increasingly flexible and capable, researchers and engineers are left with a large number of possible directions in which to set their goals. With all these advances and opportunities, the question of "What is theoretically possible?" becomes more and more relevant. In this talk we will attack this question in relation to networks of intense current interest: cognitive networks. Cognitive networks are networks in which a number of wireless devices, some of which may be more capable and/or adaptive cognitive radios wish to communicate. This talk will first highlight key information theoretic metrics for and models of multi-terminal cognitive networks - how to model the extra capabilities of cognitive radios is key. We then proceed on an intuitive tour of the current state of the art in cognitive networks from an information theoretic perspective, obtaining bounds on the capacity regions of more "intelligent" and "cognitive" networks.