Humans and other primates are capable of recognizing whether they have encountered a specific image before, even after seeing it only once briefly, and despite having seen thousands of other images. Incredibly, this memory is true even when the image is highly similar to a previously seen image. This remarkable mnemonic ability is thought to depend on the transformation of overlapping sensory representations, allowing for distinct memories— a process often attributed to pattern separation in the hippocampus (HC). While prior work has established that HC supports memory-specific representations that can distinguish between similar experiences, it remains unresolved whether these transformations occur exclusively in HC or are supported by processing in other brain regions, such as the inferotemporal cortex (ITC), a high-level visual area traditionally associated with object recognition. This thesis investigates the neural mechanisms that contribute to memory fidelity by examining how visual and memory-related brain regions represent repeated and visually similar images. Specifically, the presented experiment recorded neural responses from both HC and ITC in rhesus monkeys performing a visual recognition memory task involving novel, repeated, and highly similar lure images. These results were analyzed to determine how neural population activity in each region aligned with behavioral performance, testing whether sharpening of memory signals, defined as nonlinear deviations from visual similarity, could be observed in either area. The results revealed clear evidence for behavioral sharpening: monkeys’ performance had fewer mnemonic errors than expected based on the degree of visual similarity. This sharpening effect was reflected in neural activity within HC, consistent with its proposed role in pattern separation. However, a key and unexpected finding was that ITC responses also reflected sharpening behavior, even though ITC is not classically associated with memory-specific computations. These findings challenge a strict focus on contributions of HC to highly similar image memory, and instead support a more distributed account of visual memory, implicating ITC as contributing to the conversion from image to memory representations.