Since the identification of concept cells nearly two decades ago [1], the semantic invariance of single neurons in the human medial temporal lobe has been extensively investigated [2, 3]. Such investigations typically consist of an initial screening to identify semantically-tuned neurons and a subsequent follow-up experiment which uses the identified stimulus-response pairings to address a particular question. While often examined in the service of other scientific investigations, the initial screenings have rarely been the sole focus of investigation [4]. Given the challenges endemic to human single unit studies, these screening sessions present an unparalleled opportunity to examine the relationships between neuronal responses and stimulus information, patient demographics, and clinical outcomes. Here, we investigated the response properties of stimulus-selective neurons from over 10 years of single-unit recordings from the human medial temporal lobe. Our dataset consists of several thousand neurons collected from over 50 patients across 115 experimental sessions during a screening experiment in which static images depicting persons, places, and objects were shown to intracranially-implanted epilepsy patients. Using a mixture of manual and automated methods, we classified over 3700 unique stimuli included in these screenings and utilized an existing data management framework [5] to query the dataset. The size and breadth of this dataset enabled us to examine patterns of neuronal responses which are usually obscured by limited sampling in the neuronal and stimulus spaces. By leveraging the inclusion of both patient-provided personal images and images of clinical staff in the stimulus set, we estimated the likelihood of observing a novel person-specific neuronal response given time since exposure. We additionally investigated the presence of region-specific trends in neuronal responses on the basis of stimulus category, and compared these to previous findings from targeted experiments [6]. Finally, we examined the computational properties underlying the entire population of neuronal responses and compared these properties across putative cell types, regions, stimulus categories, and patient demographic information such as age and clinical outcome.