Our recent work has been published in Feb 5 issue of Neuron, and chosen for a cover image. The cover illustration depicts the approach of establishing specific computational accounts for the functions of different brain regions involved in behavioral control by using a combination of computational modeling with behavioral and fMRI data. The graph-based layout with circles denotes state-space representations that are part of a computational framework for understanding how the brain encodes a model of a particular decision-problem or task, while the rectangles symbolize the degree of uncertainty about those state-space representations. In this issue, Lee et al. (pages 687–699) show that a region of inferior lateral prefrontal cortex (the bottom middle brain image) alongside an area of frontopolar cortex (the top middle brain image) contains neural computations that could be used to allocate the degree of control exerted over behavior by two learning systems: a deliberative model-based system that uses state-prediction errors (left middle) to learn a world model, which in turn is used to flexibly compute state-action values (left diagonal), and a model-free system that uses reward-prediction errors (right middle) to directly update the state-action value (right diagonal).