Experience with a sensorimotor task, such as practicing a piano piece, leads to strong coupling of sensory (visual or auditory) and motor cortices. Here we review behavioral and neurophysiological (M/EEG, TMS and fMRI) research exploring this topic using the brain of musicians as a model system. Our review focuses on a recent body of evidence suggesting that this form of coupling might have (at least) two cognitive functions. First, it leads to the generation of equivalent predictions (concerning both when and what event is more likely to occur) during both perception and production of music. Second, it underpins the common coding of perception and action that supports the integration of the motor output of multiple musicians’ in the context of joint musical tasks. Essentially, training-based coupling of perception and action might scaffold the human ability to represent complex (structured) actions and to entrain multiple agents—via reciprocal prediction and adaptation—in the pursuit of shared goals.