Retinal visual prostheses provide hope of restoring sight to patients suffering from retinal degeneration such as retinitis pigmentosa and age-related macular degeneration. Retinal prostheses are used to electrically stimulate residual neurons that are spared in these diseases, namely the retinal ganglion cells (RGCs), eliciting percepts of light termed 'phosphenes'. The elicitation of multiple phosphenes via an electrode array allows patterns to be produced, resulting in a rudimentary form of vision. For such patterns to be produced effectively, the prosthesis must generate well-defined phosphenes. To this end, the hexpolar configuration has been proposed as an alternative to the traditional monopolar or bipolar configurations. It utilizes six electrodes surrounding the stimulating electrode to serve as a combined return, or 'hex guard', purportedly localizing the activation to cells located within them. In this study, the efficacy of the hexpolar configuration in localizing activity was investigated by using patch-clamp electrophysiology to measure the activation thresholds of RGCs to electrical stimulation in isolated rabbit retina. Cells located outside the hex guard were found to have significantly higher relative hexpolar thresholds (>2 fold) as compared to cells located within the hex guard. This confirms the efficacy of the hexpolar configuration in localizing activity to within the hex guard. Furthermore, the effect of using cathodic-first versus anodic-first stimulation on hexpolar threshold and localization was investigated. No significant difference was observed between the two groups, in terms of lowering thresholds or improving localization.