Motor adaptation displays a structure-learning effect: adapta-tion to a new perturbation occurs more quickly when the sub-ject has prior exposure to perturbations with related structure.Although this ‘learning-to-learn’ effect is well documented, itsunderlying computational mechanisms are poorly understood.We present a new model of motor structure learning, approach-ing it from the point of view of deep reinforcement learning.Previous work outside of motor control has shown how recur-rent neural networks can account for learning-to-learn effects.We leverage this insight to address motor learning, by import-ing it into the setting of model-based reinforcement learning.We apply the resulting processing architecture to empiricalfindings from a landmark study of structure learning in target-directed reaching (Braun et al., 2009), and discuss its implica-tions for a wider range of learning-to-learn phenomena.