Individual memories are thought to be stored by ensembles of neurons that are reactivated during recall. While these memory ensembles can now be labelled and manipulated in the rodent brain, they are difficult to measure in humans using non-invasive methods. Here, by adopting a multifaceted approach, I will show how indirect measures of memory ensembles can be obtained from the human brain. I will start by showing how representational fMRI can be used to track how memories contribute to novel decisions. I will show that novel stimuli are initially represented by combining multiple memories in the hippocampus and medial prefrontal cortex, but are eventually supported by independent representations. Secondly, I will show how MRS, fMRI and transcranial direct current stimulation (tDCS) can be combined to investigate storage of associative memories. I will show evidence to suggest that associative memories are stored in balanced excitatory-inhibitory ensembles, lying dormant unless neocortical excitability is modulated. The importance of neocortical inhibition in preventing interference between overlapping memories will also be explored. Finally, I will discuss how balanced memory ensembles may be released during recall and show preliminary multi-unit electrophysiology data to suggest that the hippocampus may play a role.