Brain-machine interfaces allow us to utilize specific neural activity patterns to control a particular behaviour. By establishing a causal link between neural activity and the corresponding behaviour, brain-machine interfaces also serve as a tool for investigating mechanisms of learning and plasticity in the brain. For decades, it has been known that animals, and humans, can be trained to control neural activity in parts of the brain that are involved in producing bodily movements. However, whether this type of self-regulation can be exerted over parts of the brain that are involved in higher cognition, such as memory function, remains an open question. In this thesis, we attempt to answer this question by utilizing brain-machine interfaces to train humans and animals to self-regulate the activity of individual neurons in memory-related brain structures, particularly the hippocampus. Specifically, our objectives are to (1) determine the extent to which animals and humans can learn to self-regulate activity of neurons in memory-related brain structures, and (2) shed light on the neural and behavioural mechanisms that facilitate this type of learning. In humans, we recorded activity of single neurons in individuals with epilepsy to determine whether an arbitrarily selected neuron can be brought under volitional control. Using a visual brain-machine interface, we demonstrated that humans could learn to upregulate the activity of a selected neuron quickly and reliably. Successful learning was correlated with changes in the neural population dynamics, consisting of robust phase-locking of the trained neuron to the local alpha/beta oscillations. We also utilized a similar brain-machine interface to train rats to control the activity of small hippocampal neuronal ensembles. We demonstrated that, like humans, rats could also learn to modulate the activity of individual neurons in the hippocampus. Using video monitoring we identified behavioural correlates of this type of learning which shed light on the behavioural strategies that animals use to learn such an abstract neuroprosthetic skill. Furthermore, by recording simultaneously from multiple electrodes we also demonstrated that learning is associated with increased spike-timing precision between the hippocampus and the closely associated ventral striatum. We further discuss these findings and propose a putative basal ganglia circuit that may play a role in facilitating this type of learning. Our results demonstrate that volitional self-regulation of neural activity is possible in previously uninvestigated memory-related structures, and this type of learning may engage cortico-basal ganglia circuits in a novel manner. These findings open the possibility of utilizing such neuroprosthetic skill learning to further probe learning and plasticity in memory-related structures of the brain and develop novel neuroprosthetics that can probe memory function without any exogenous stimulation.