I study the computational principles and neural systems that allow the human mind to be so flexible in some cases  — for example, able to generate and understand an infinite number of possible sentences, able to quickly reason about new situations, and able to plan for uncertain futures — while being so profoundly limited in others — for example, easily forgetting names, dates, and phone numbers, and unable to perform multiple simple tasks at once. Why are we so smart sometimes and not others? I approach these questions using information-theory, classic neural network formalisms, and cognitive neuroscientific methods (e.g., fMRI), in pursuit of some general principles.