The process of learning requires the brain to generate new activity patterns, yet it has been known for some time that the learning speed of certain behaviors can vary significantly. Brain computer interfaces (BCIs) offer a promising setup to study learning since they allow one to concretely link changes in activity to task performance. Using a BCI-driven center-out task, Sadtler et. al. (2014) showed that the speed of learning new activity patterns can differ significantly depending on their relation to Rhesus macaque’s intrinsic neural activity. A mapping between how individual synapses adjust their strength and population-dependent differences in learning speed can be achieved using in silico models, especially when they incorporate known biological constraints on learning. Using recurrent neural networks (RNNs) that learn in a biologically-plausible way via reinforcement learning (RL), we offer a strikingly simple explanation for the distinct learning speeds observed in Sadtler et. al.: a difference in the levels of exploratory activity. We observe that learning speed is strongly modulated by the variance of activity, i.e. exploration. By constructing a model of the center-out task, we show that even if target activity is constrained to lie a fixed distance from an animal’s intrinsic activity, the amount of activity exploration is generally significantly lower outside of the space of intrinsic activity. Our results suggest that the variance of the BCI activity, even when measured prior to learning, should partially explain Sadtler et. al.’s observed differences in learning speed. This simple explanation leads to intuitive insights as to why certain neural activities may be harder to achieve than others and carries implications for obtaining more robust BCI learning.