To navigate the social world, people must understand and anticipate each other’s thoughts and feelings. How does the brain organize its representations of these hidden mental states? In the first part of my talk, I will describe the 3d Mind Model, which posits that three psychological dimensions describe the way the brain represents mental states: rationality (vs. emotionality), social impact (the extent to which states affect others), and valence (positive vs. negative). FMRI, computational text analysis, and behavior all indicate that the 3d Mind Model is a robust, comprehensive, and generalizable account of mental state representation. In the second part of my talk, I will discuss a key function of this map of the mental world: facilitating the prediction of mental state dynamics. I will present evidence from experience sampling and fMRI studies which indicates that representing mental states along the dimensions of the 3d Mind Model facilitates accurate and efficient social prediction. I will conclude by discussing new data from statistical learning and artificial neural network experiments, which suggest that the goal of prediction shapes how the brain generates mental state concepts in the first place.