Face perception, a fundamental component of primate social behavior, is supported by a network of specialized visual regions within the ventral visual stream of humans and macaques. Discrete regions, or “patches” within this network respond preferentially to face images over non-face object images, with the majority of visually responsive neurons within these regions responding selectively to faces. In recent years, the functional specialization of neurons within particular fMRI-defined face patches has been studied intensively. In this series of studies, we have investigated one such patch (AF) located in the anterior fundus of the superior temporal sulcus using a new method longitudinal of electrophysiological single unit recording, where individual neurons can be isolated and monitored for several months using chronically implanted electrodes. We made use of this unique opportunity to study the responses of face-selective neurons in two different ways. Our first approach was the longitudinal observation of neurons over time scales relevant to behavioral learning. By observing single-units longitudinally, over several weeks of natural visual experience, as well as throughout the course of an intensive face-learning paradigm, we asked whether individual neurons would change their response selectivity. In the case of the learning paradigm, we asked whether there might be changes in face-selective neural responses that occur while the animal gradually acquired greater sensitivity for the distinctive features of novel face stimuli. For this, we taught the animal to recognize individual monkeys and humans using morphed faces of diminished identity levels. We found AF neurons were remarkably stationary in their responses over time scales of weeks and months, and even for periods as long as one year. Even during periods of intensive training using the stimuli for which the neurons appeared specialized, AF neurons demonstrate very limited change in their response patterns and selectivity. In our second approach- taking advantage of the fact that the neural response selectivity is unchanging across sessions- we accumulated neural responses to a large number of visual stimuli presented over multiple sessions, which allowed for both a broader sampling of the visual stimulus space and an enhanced sensitivity to the preferences of low-firing rate neurons. In this approach, we focused on visual responses to faces, asking how a conjunction of identity, species, viewing angle, and spatial scale shaped single-unit responses in face patch AF. Neurons in AF showed selective responses to each of these manipulations to different degrees. Within the domain of faces, most striking determinants of neural firing were face scale and the systematic, norm-based tuning for face identity. Many neurons were also strongly selective for macaque faces over human faces. Other factors, such as 3-dimensional face rotation, were less important. In addition to insight into the functional compartmentalization of the face-processing network, these results from face patch AF also provide a unique view into the relationship between high-level neural selectivity in the brain and the inherent trade-off between stability and plasticity that accompanies learning-induced changes in behavior.