What changes in the brain when we learn? Those changes must include a reorganization of the patterns of activity that a population of neurons can exhibit. We use a brain-computer interface (BCI) paradigm to examine the rules whereby populations of neurons change with learning. With a BCI, we can request that our monkeys exhibit specific patterns of neural activity, and then observe whether they are able to do so, and how they do it. We have found that it is relatively easy for animals to learn to re-associate existing neural activity patterns with new behaviors, but it takes many days to forge entirely new patterns of neural activity. If the population learning rules that are evident using a BCI are also at play during cognitive and motor learning, then in the future we might be able to use these principles to guide and accelerate learning.