Motor, sensory, and cognitive learning require networks of neurons to generate new activity patterns. Because some behaviors are easier to learn than others 1, 2 , we wondered if some neural activity patterns are easier to generate than others. We asked whether the existing network constrains the patterns that a subset of its neurons is capable of exhibiting, and if so, what principles define the constraint. We employed a closed-loop intracortical brain-computer interface (BCI) learning paradigm in which Rhesus monkeys controlled a computer cursor by modulating neural activity patterns in primary motor cortex. Using the BCI paradigm, we could specify and alter how neural activity mapped to cursor velocity. At the start of each session, we observed the characteristic activity patterns of the recorded neural population. These patterns comprise a low-dimensional space (termed the intrinsic manifold, or IM) within the high-dimensional neural firing rate space. They presumably reflect constraints imposed by the underlying neural circuitry. We found that the animals could readily learn to proficiently control the cursor using neural activity patterns that were within the IM. However, animals were less able to learn to proficiently control the cursor using activity patterns that were outside of the IM. This result suggests that the existing structure of a network can shape learning. On the timescale of hours, it appears to be difficult to learn to generate neural activity patterns that are not consistent with the existing network structure. These findings offer a network-level explanation for the observation that we are more readily able to learn new skills when they are related to the skills that we already possess 3, 4 .